diff --git a/.gitignore b/.gitignore
index 54c488e68..d500df9bc 100644
--- a/.gitignore
+++ b/.gitignore
@@ -183,6 +183,10 @@ dvclive
 # MacOS
 *.DS_Store
 
+.prettierrc
+
+*.ipynb
+*.pkl
 job_queue.db-shm
 job_queue.db-wal
 
@@ -202,3 +206,8 @@ docs/node_modules/
 !docs/docs/build/**
 !docs/i18n/**/build/
 !docs/i18n/**/build/**
+
+**/csv-train.arrow
+**/dataset_info.json
+
+/RAG_benchmark
diff --git a/AGENTS.md b/AGENTS.md
new file mode 100644
index 000000000..4100dd641
--- /dev/null
+++ b/AGENTS.md
@@ -0,0 +1,126 @@
+# AGENTS.md
+
+This file provides guidance to AI coding agents when working with code in this repository.
+
+## What is DashAI
+
+DashAI is a desktop/web graphical toolbox for training, evaluating, and deploying ML models. FastAPI backend + React frontend, optionally wrapped in PyWebView for a native desktop window. Python >= 3.10.
+
+## Commands
+
+### Backend
+
+```bash
+pip install -e . -r requirements-dev.txt
+pre-commit install
+
+# Run dev server
+python -m DashAI --no-browser --logging-level DEBUG
+
+# Lint / format (must pass both before committing)
+ruff check --fix
+ruff format
+
+# Run all tests (in-memory SQLite — no setup needed)
+pytest tests/
+
+# Run a single test
+pytest tests/back/api/test_components_api.py -v
+pytest tests/back/api/test_components_api.py::test_function_name -v
+
+# Database migrations (auto-runs on startup)
+alembic upgrade head
+```
+
+### Frontend
+
+```bash
+cd DashAI/front
+
+yarn install        # requires Node LTS + Yarn 3.5.0 (enforced by packageManager)
+yarn start          # dev server on http://localhost:3000 (eslint disabled)
+yarn build          # eslint disabled
+yarn test           # eslint disabled
+yarn test FileName.test.tsx
+yarn lint           # runs eslint explicitly (uses eslint.config.js)
+```
+
+**Important:** `start`, `build`, and `test` scripts all set `DISABLE_ESLINT_PLUGIN=true`. ESLint only runs via the dedicated `yarn lint` command. Prettier is configured in `DashAI/front/.prettierrc` and runs via pre-commit.
+
+## Architecture
+
+```
+Browser / PyWebView
+  → React (port 3000 dev / port 8000 prod)
+  → FastAPI (/api/v1/...)
+  → Service layer → SQLite (SQLAlchemy + Alembic)
+  → Huey job queue (async: training, conversion, exploration)
+```
+
+The frontend polls `/api/v1/jobs/{job_id}` for long-running task status.
+
+## Key files
+
+| Path | Purpose |
+|------|---------|
+| `DashAI/__main__.py` | CLI entry point (typer). Starts uvicorn + Huey consumer subprocess |
+| `DashAI/back/app.py` | FastAPI `create_app` factory |
+| `DashAI/back/container.py` | DI container (`kink`) — config, DB engine, ComponentRegistry, job queue |
+| `DashAI/back/initial_components.py` | **Registers all components on startup.** Add new ML components here in `get_initial_components()` |
+| `DashAI/back/dependencies/config_builder.py` | Builds config dict (paths, logging, calls `get_initial_components()`) |
+| `DashAI/back/dependencies/registry/component_registry.py` | `ComponentRegistry` — resolves components by name string |
+| `DashAI/back/api/api_v1/endpoints/` | REST endpoints — each file is a FastAPI router |
+| `DashAI/back/api/api_v1/api.py` | Mounts all endpoint routers on `api_router_v1` |
+| `DashAI/back/core/schema_fields/` | Type system driving dynamic frontend forms |
+| `DashAI/back/pipeline/` | DAG pipeline nodes |
+| `DashAI/back/plugins/` | Plugin system (PyPI packages with `dashai.plugins` entry point) |
+| `DashAI/front/src/components/configurableObject/` | Auto-generates forms from backend component schemas |
+| `DashAI/front/src/pages/generative/SessionRouter.jsx` | Routes `/app/generative/sessions/:id` to RAG or non-RAG view based on session `task_name` |
+| `DashAI/front/src/pages/generative/RAGSession/` | RAG session setup wizard (RAGSessionSetup) + collapsible section components per pipeline stage |
+| `DashAI/front/src/pages/generative/RAGSession/sections/` | Per-stage components: ChunkingSection, RetrieverSection, GeneratorSection, PromptSection |
+| `DashAI/front/src/pages/generative/RAGSession/advanced/` | Advanced configuration modals: CompositeRetrieverBuilder, ChunkingConfigurationStep, RetrieverConfigurationStep, etc. |
+| `DashAI/front/src/pages/generative/RAGSession/components/` | Reusable bodies: GeneratorBody, PromptBody, PresetCard, AdvancedConfigCard |
+
+## Key patterns
+
+**ComponentRegistry** — all ML components (models, metrics, converters, dataloaders, explorers, explainers, tasks, optimizers, jobs, pipeline nodes) are registered at startup and resolved by name string. To add a new component: subclass the relevant base, define its schema, and add it to `get_initial_components()` in `DashAI/back/initial_components.py`.
+
+**Schema / type system** — every component declares parameters using `BaseSchema` + field classes (`IntField`, `FloatField`, `ComponentField`, `UnionType`, etc.) with `MultilingualString` labels. The frontend uses these schemas to auto-generate configuration forms.
+
+**Dependency injection (`kink`)** — singletons (config, DB engine, ComponentRegistry, job queue) live in the `di` container. Use `@inject` to receive them. Config is accessed as `di["config"]` and includes `INITIAL_COMPONENTS`.
+
+**Huey job queue** — in dev mode the Huey consumer runs as a subprocess spawned from `__main__.py`; in PyInstaller bundles it runs as a daemon thread due to limitations with frozen executables.
+
+## RAG Module
+
+DashAI includes a **Retrieval-Augmented Generation (RAG)** module — a 4-stage pipeline (Document Loading → Chunking → Retrieval → Generation) for chatting with your documents.
+
+- **Backend:** `DashAI/back/models/RAG/` — pipeline orchestrator (RAGPipeline), abstract factory (RAGModelsFactory), sub-factories for prompts/chunking/retrievers/LLMs, retriever repository (all SQL), document loader, chunk-set caching.
+- **Frontend:** `DashAI/front/src/pages/generative/RAGSession/` — session setup wizard (RAGSessionSetup, RAGSessionPage) with stage sections (ChunkingSection, RetrieverSection, GeneratorSection, PromptSection) and advanced config modals.
+- **Jobs:** `RAGJob` extends `GenerativeJob` to run the RAG pipeline as a background task.
+- **Full docs (backend + frontend):** See [`docs/rag/`](./docs/rag/) for architecture, data models, retrievers, pipeline orchestration, frontend architecture, testing guide, and known constraints.
+
+## Testing
+
+- Backend tests use in-memory SQLite — no setup needed.
+- Key fixtures in `tests/back/conftest.py`: `test_path`, `test_datasets_path`, `test_job_queue`.
+- Job queue tests: `test_job_queue.set_test_mode(immediate=True)` runs tasks synchronously.
+- API tests use FastAPI `TestClient` defined in `tests/back/api/conftest.py` (module-scoped).
+- Frontend test command passes `--passWithNoTests` in CI.
+
+## Adding new things
+
+**New ML model or converter:** subclass `BaseModel` / `BaseConverter`, define schema, register in `DashAI/back/initial_components.py::get_initial_components()`.
+
+**New API endpoint:** add file in `DashAI/back/api/api_v1/endpoints/`, include its router in `DashAI/back/api/api_v1/api.py`.
+
+**New async job:** subclass `BaseJob`, dispatch via `job_queue.enqueue(...)`, track status through jobs API.
+
+## Data at runtime
+
+Default `~/.DashAI/` (overridable via `DASHAI_LOCAL_PATH` env var):
+`datasets/`, `runs/`, `explanations/`, `notebooks/`, `images/`, `documents/`, `rag/`, `sqlite.db`
+
+## CI
+
+Only a `publish.yml` workflow exists (on push to `production` branch). It builds the frontend, publishes to PyPI, and builds Windows/macOS installers. No CI test workflow is defined.
diff --git a/DashAI/__main__.py b/DashAI/__main__.py
index c2bc3717d..5aa48d4c5 100644
--- a/DashAI/__main__.py
+++ b/DashAI/__main__.py
@@ -65,7 +65,7 @@
 
 
 def open_browser() -> None:
-    _wait_for_backend_server(timeout=120)
+    _wait_for_backend_server(timeout=1200)
     url = "http://localhost:8000/app/"
     webbrowser.open(url=url, new=0, autoraise=True)
 
@@ -119,10 +119,11 @@ def _wait_for_backend_server(host="127.0.0.1", port=8000, timeout=15):
     import socket
     import time
 
+    timeout = 1000
     start_time = time.time()
     while time.time() - start_time < timeout:
         try:
-            with socket.create_connection((host, port), timeout=1):
+            with socket.create_connection((host, port), timeout=timeout):
                 return True
         except (OSError, ConnectionRefusedError):
             time.sleep(0.5)
diff --git a/DashAI/alembic/versions/8d5416cac8f1_merge_rag_and_production_heads.py b/DashAI/alembic/versions/8d5416cac8f1_merge_rag_and_production_heads.py
new file mode 100644
index 000000000..f37cd9196
--- /dev/null
+++ b/DashAI/alembic/versions/8d5416cac8f1_merge_rag_and_production_heads.py
@@ -0,0 +1,26 @@
+"""Merge RAG and production heads
+
+Revision ID: 8d5416cac8f1
+Revises: d4e5f6a7b8c9, f1a2b3c4d5e6
+Create Date: 2026-07-06 12:30:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = '8d5416cac8f1'
+down_revision: Union[str, None] = ('d4e5f6a7b8c9', 'f1a2b3c4d5e6')
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    pass
+
+
+def downgrade() -> None:
+    pass
diff --git a/DashAI/alembic/versions/a1b2c3d4e5f6_add_composite_retriever_support.py b/DashAI/alembic/versions/a1b2c3d4e5f6_add_composite_retriever_support.py
new file mode 100644
index 000000000..c114329b0
--- /dev/null
+++ b/DashAI/alembic/versions/a1b2c3d4e5f6_add_composite_retriever_support.py
@@ -0,0 +1,47 @@
+"""add composite retriever support
+
+Revision ID: a1b2c3d4e5f6
+Revises: f928e0b5203d
+Create Date: 2026-05-22 12:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = 'a1b2c3d4e5f6'
+down_revision: Union[str, None] = 'f928e0b5203d'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    with op.batch_alter_table('rag_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('chk_retriever_type', type_='check')
+        batch_op.add_column(
+            sa.Column('composite_child_ids', sa.JSON(), nullable=True)
+        )
+        batch_op.create_check_constraint(
+            'chk_retriever_type',
+            'NOT (dense_retriever_id IS NOT NULL AND sparse_retriever_id IS NOT NULL)'
+        )
+
+
+def downgrade() -> None:
+    with op.batch_alter_table('rag_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('chk_retriever_type', type_='check')
+        batch_op.drop_column('composite_child_ids')
+        batch_op.create_check_constraint(
+            'chk_retriever_type',
+            (
+                "(class_name = 'SparseRetriever' "
+                "AND sparse_retriever_id IS NOT NULL "
+                "AND dense_retriever_id IS NULL) OR "
+                "(class_name = 'DenseRetriever' "
+                "AND dense_retriever_id IS NOT NULL "
+                "AND sparse_retriever_id IS NULL)"
+            )
+        )
diff --git a/DashAI/alembic/versions/b2c3d4e5f6a7_add_uniqueness_constraints_rag.py b/DashAI/alembic/versions/b2c3d4e5f6a7_add_uniqueness_constraints_rag.py
new file mode 100644
index 000000000..e105106cd
--- /dev/null
+++ b/DashAI/alembic/versions/b2c3d4e5f6a7_add_uniqueness_constraints_rag.py
@@ -0,0 +1,97 @@
+"""add uniqueness constraints to rag tables and fix storage tracing
+
+Revision ID: b2c3d4e5f6a7
+Revises: a1b2c3d4e5f6
+Create Date: 2026-05-25 12:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+revision: str = 'b2c3d4e5f6a7'
+down_revision: Union[str, None] = 'a1b2c3d4e5f6'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ── Chunk: replace useless (id, document_id) with (document_id, chunk_index, chunking_model_id) ──
+    with op.batch_alter_table('chunk', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_chunk_document', type_='unique')
+        batch_op.create_unique_constraint(
+            'uix_document_chunk_index_chunking',
+            ['document_id', 'chunk_index', 'chunking_model_id'],
+        )
+
+    # ── RAGPrompt ──
+    with op.batch_alter_table('rag_prompt', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_prompt_class_params',
+            ['class_name', 'parameters'],
+        )
+
+    # ── RAGGenerationModel ──
+    with op.batch_alter_table('rag_generation_model', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_gen_model_class_params',
+            ['class_name', 'parameters'],
+        )
+
+    # ── RAGChunkingModel ──
+    with op.batch_alter_table('rag_chunking_model', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_chunking_model_class_params',
+            ['class_name', 'parameters'],
+        )
+
+    # ── RAGEmbeddingModel ──
+    with op.batch_alter_table('rag_embedding_model', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_embedding_model_class_params',
+            ['class_name', 'parameters'],
+        )
+
+    # ── RAGSparseRetriever ──
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_sparse_retriever',
+            ['class_name', 'parameters', 'documents_ids', 'chunking_model_id'],
+        )
+
+    # ── RAGDenseRetriever ──
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.create_unique_constraint(
+            'uix_rag_dense_retriever',
+            ['class_name', 'parameters', 'document_ids',
+             'chunking_model_id', 'embedding_model_id'],
+        )
+
+
+def downgrade() -> None:
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_dense_retriever', type_='unique')
+
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_sparse_retriever', type_='unique')
+
+    with op.batch_alter_table('rag_embedding_model', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_embedding_model_class_params', type_='unique')
+
+    with op.batch_alter_table('rag_chunking_model', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_chunking_model_class_params', type_='unique')
+
+    with op.batch_alter_table('rag_generation_model', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_gen_model_class_params', type_='unique')
+
+    with op.batch_alter_table('rag_prompt', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_prompt_class_params', type_='unique')
+
+    with op.batch_alter_table('chunk', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_document_chunk_index_chunking', type_='unique')
+        batch_op.create_unique_constraint(
+            'uix_chunk_document',
+            ['id', 'document_id'],
+        )
diff --git a/DashAI/alembic/versions/c3d4e5f6a7b8_invert_retriever_fk_add_child_table.py b/DashAI/alembic/versions/c3d4e5f6a7b8_invert_retriever_fk_add_child_table.py
new file mode 100644
index 000000000..93de3de17
--- /dev/null
+++ b/DashAI/alembic/versions/c3d4e5f6a7b8_invert_retriever_fk_add_child_table.py
@@ -0,0 +1,134 @@
+"""invert FK direction for retrievers and add rag_retriever_child
+
+Revision ID: c3d4e5f6a7b8
+Revises: b2c3d4e5f6a7
+Create Date: 2026-05-26 12:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+revision: str = 'c3d4e5f6a7b8'
+down_revision: Union[str, None] = 'b2c3d4e5f6a7'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ── Drop old FK from RAGPipeline to RAGRetriever ──
+    with op.batch_alter_table('rag_pipeline', schema=None) as batch_op:
+        batch_op.drop_constraint(
+            'fk_rag_pipeline_retriever_model_id_rag_retriever', type_='foreignkey'
+        )
+        batch_op.drop_column('retriever_model_id')
+
+    # ── Rebuild RAGRetriever: drop old columns, add pipeline_id ──
+    with op.batch_alter_table('rag_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('chk_retriever_type', type_='check')
+        batch_op.drop_constraint(
+            'fk_rag_retriever_dense_retriever_id_rag_dense_retriever', type_='foreignkey'
+        )
+        batch_op.drop_constraint(
+            'fk_rag_retriever_sparse_retriever_id_rag_sparse_retriever', type_='foreignkey'
+        )
+        batch_op.drop_column('dense_retriever_id')
+        batch_op.drop_column('sparse_retriever_id')
+        batch_op.drop_column('composite_child_ids')
+        batch_op.add_column(
+            sa.Column('pipeline_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_retriever_pipeline_id',
+            'rag_pipeline', ['pipeline_id'], ['id'], ondelete='CASCADE',
+        )
+
+    # ── Add bridge_id to RAGSparseRetriever ──
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.add_column(
+            sa.Column('bridge_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_sparse_retriever_bridge_id',
+            'rag_retriever', ['bridge_id'], ['id'], ondelete='CASCADE',
+        )
+
+    # ── Add bridge_id to RAGDenseRetriever ──
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.add_column(
+            sa.Column('bridge_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_dense_retriever_bridge_id',
+            'rag_retriever', ['bridge_id'], ['id'], ondelete='CASCADE',
+        )
+
+    # ── Create RAGRetrieverChild ──
+    op.create_table(
+        'rag_retriever_child',
+        sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+        sa.Column('parent_id', sa.Integer(), nullable=False),
+        sa.Column('child_id', sa.Integer(), nullable=False),
+        sa.Column('child_order', sa.Integer(), nullable=False),
+        sa.ForeignKeyConstraint(
+            ['parent_id'], ['rag_retriever.id'],
+            name='fk_rag_retriever_child_parent', ondelete='CASCADE',
+        ),
+        sa.ForeignKeyConstraint(
+            ['child_id'], ['rag_retriever.id'],
+            name='fk_rag_retriever_child_child', ondelete='CASCADE',
+        ),
+        sa.PrimaryKeyConstraint('id', name='pk_rag_retriever_child'),
+        sa.UniqueConstraint(
+            'parent_id', 'child_order',
+            name='uix_retriever_child_order',
+        ),
+    )
+
+
+def downgrade() -> None:
+    op.drop_table('rag_retriever_child')
+
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('fk_rag_dense_retriever_bridge_id', type_='foreignkey')
+        batch_op.drop_column('bridge_id')
+
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('fk_rag_sparse_retriever_bridge_id', type_='foreignkey')
+        batch_op.drop_column('bridge_id')
+
+    with op.batch_alter_table('rag_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('fk_rag_retriever_pipeline_id', type_='foreignkey')
+        batch_op.drop_column('pipeline_id')
+        batch_op.add_column(
+            sa.Column('composite_child_ids', sa.JSON(), nullable=True)
+        )
+        batch_op.add_column(
+            sa.Column('sparse_retriever_id', sa.Integer(), nullable=True)
+        )
+        batch_op.add_column(
+            sa.Column('dense_retriever_id', sa.Integer(), nullable=True)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_retriever_sparse_retriever_id_rag_sparse_retriever',
+            'rag_sparse_retriever', ['sparse_retriever_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_retriever_dense_retriever_id_rag_dense_retriever',
+            'rag_dense_retriever', ['dense_retriever_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_check_constraint(
+            'chk_retriever_type',
+            'NOT (dense_retriever_id IS NOT NULL AND sparse_retriever_id IS NOT NULL)',
+        )
+
+    with op.batch_alter_table('rag_pipeline', schema=None) as batch_op:
+        batch_op.add_column(
+            sa.Column('retriever_model_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_pipeline_retriever_model_id_rag_retriever',
+            'rag_retriever', ['retriever_model_id'], ['id'], ondelete='CASCADE',
+        )
diff --git a/DashAI/alembic/versions/d4e5f6a7b8c9_add_chunk_set_architecture.py b/DashAI/alembic/versions/d4e5f6a7b8c9_add_chunk_set_architecture.py
new file mode 100644
index 000000000..71b937098
--- /dev/null
+++ b/DashAI/alembic/versions/d4e5f6a7b8c9_add_chunk_set_architecture.py
@@ -0,0 +1,222 @@
+"""add chunk_set architecture: chunk_set, chunk_set_document, retriever_chunk_set
+
+Revision ID: d4e5f6a7b8c9
+Revises: c3d4e5f6a7b8
+Create Date: 2026-05-26 18:00:00.000000
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+revision: str = 'd4e5f6a7b8c9'
+down_revision: Union[str, None] = 'c3d4e5f6a7b8'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ── RAGChunkSet ──
+    op.create_table(
+        'rag_chunk_set',
+        sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+        sa.Column('signature', sa.String(), nullable=False),
+        sa.Column('parameters', sa.JSON(), nullable=True),
+        sa.PrimaryKeyConstraint('id', name='pk_rag_chunk_set'),
+        sa.UniqueConstraint('signature', name='uq_rag_chunk_set_signature'),
+    )
+
+    # ── RAGChunkSetDocument ──
+    op.create_table(
+        'rag_chunk_set_document',
+        sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+        sa.Column('chunk_set_id', sa.Integer(), nullable=False),
+        sa.Column('document_id', sa.Integer(), nullable=False),
+        sa.ForeignKeyConstraint(
+            ['chunk_set_id'], ['rag_chunk_set.id'],
+            name='fk_chunk_set_doc_set', ondelete='CASCADE',
+        ),
+        sa.ForeignKeyConstraint(
+            ['document_id'], ['document.id'],
+            name='fk_chunk_set_doc_document', ondelete='CASCADE',
+        ),
+        sa.PrimaryKeyConstraint('id', name='pk_rag_chunk_set_document'),
+        sa.UniqueConstraint(
+            'chunk_set_id', 'document_id', name='uix_chunk_set_document',
+        ),
+    )
+
+    # ── Chunk: replace chunking_model_id with chunk_set_id, add metadata ──
+    with op.batch_alter_table('chunk', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_document_chunk_index_chunking', type_='unique')
+        batch_op.drop_constraint(
+            'fk_chunk_chunking_model_id_rag_chunking_model', type_='foreignkey'
+        )
+        batch_op.drop_column('chunking_model_id')
+        batch_op.add_column(
+            sa.Column('chunk_set_id', sa.Integer(), nullable=False)
+        )
+        batch_op.add_column(
+            sa.Column('metadata', sa.JSON(), nullable=True)
+        )
+        batch_op.create_foreign_key(
+            'fk_chunk_chunk_set_id',
+            'rag_chunk_set', ['chunk_set_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_chunk_set_doc_index',
+            ['chunk_set_id', 'document_id', 'chunk_index'],
+        )
+
+    # ── RAGSparseRetriever: drop documents_ids + chunking_model_id, add chunk_set_id ──
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_sparse_retriever', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_sparse_retriever_chunking_model_id_rag_chunking_model',
+            type_='foreignkey',
+        )
+        batch_op.drop_column('documents_ids')
+        batch_op.drop_column('chunking_model_id')
+        batch_op.add_column(
+            sa.Column('chunk_set_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_sparse_retriever_chunk_set_id',
+            'rag_chunk_set', ['chunk_set_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_rag_sparse_retriever',
+            ['class_name', 'parameters', 'chunk_set_id'],
+        )
+
+    # ── RAGDenseRetriever: drop document_ids + chunking_model_id, add chunk_set_id ──
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_dense_retriever', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_dense_retriever_chunking_model_id_rag_chunking_model',
+            type_='foreignkey',
+        )
+        batch_op.drop_column('document_ids')
+        batch_op.drop_column('chunking_model_id')
+        batch_op.add_column(
+            sa.Column('chunk_set_id', sa.Integer(), nullable=False)
+        )
+        batch_op.create_foreign_key(
+            'fk_rag_dense_retriever_chunk_set_id',
+            'rag_chunk_set', ['chunk_set_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_rag_dense_retriever',
+            ['class_name', 'parameters', 'chunk_set_id', 'embedding_model_id'],
+        )
+
+    # ── RAGEmbeddingMatrix: chunking_model_id → chunk_set_id ──
+    with op.batch_alter_table('rag_embedding_matrix', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_document_chunking_embedding', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_embedding_matrix_chunking_model_id_rag_chunking_model',
+            type_='foreignkey',
+        )
+        batch_op.alter_column('chunking_model_id', new_column_name='chunk_set_id')
+        batch_op.create_foreign_key(
+            'fk_rag_embedding_matrix_chunk_set_id',
+            'rag_chunk_set', ['chunk_set_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_document_chunk_set_embedding',
+            ['document_id', 'chunk_set_id', 'embedding_model_id'],
+        )
+
+    # ── RAGRetrieverChunkSet ──
+    op.create_table(
+        'rag_retriever_chunk_set',
+        sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+        sa.Column('retriever_id', sa.Integer(), nullable=False),
+        sa.Column('chunk_set_id', sa.Integer(), nullable=False),
+        sa.ForeignKeyConstraint(
+            ['retriever_id'], ['rag_retriever.id'],
+            name='fk_retriever_chunk_set_retriever', ondelete='CASCADE',
+        ),
+        sa.ForeignKeyConstraint(
+            ['chunk_set_id'], ['rag_chunk_set.id'],
+            name='fk_retriever_chunk_set_chunk_set', ondelete='CASCADE',
+        ),
+        sa.PrimaryKeyConstraint('id', name='pk_rag_retriever_chunk_set'),
+        sa.UniqueConstraint(
+            'retriever_id', 'chunk_set_id', name='uix_retriever_chunk_set',
+        ),
+    )
+
+
+def downgrade() -> None:
+    op.drop_table('rag_retriever_chunk_set')
+
+    with op.batch_alter_table('rag_embedding_matrix', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_document_chunk_set_embedding', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_embedding_matrix_chunk_set_id', type_='foreignkey',
+        )
+        batch_op.alter_column('chunk_set_id', new_column_name='chunking_model_id')
+        batch_op.create_foreign_key(
+            'fk_rag_embedding_matrix_chunking_model_id_rag_chunking_model',
+            'rag_chunking_model', ['chunking_model_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_document_chunking_embedding',
+            ['document_id', 'chunking_model_id', 'embedding_model_id'],
+        )
+
+    with op.batch_alter_table('rag_dense_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_dense_retriever', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_dense_retriever_chunk_set_id', type_='foreignkey',
+        )
+        batch_op.drop_column('chunk_set_id')
+        batch_op.add_column(sa.Column('chunking_model_id', sa.Integer(), nullable=False))
+        batch_op.add_column(sa.Column('document_ids', sa.JSON(), nullable=False))
+        batch_op.create_foreign_key(
+            'fk_rag_dense_retriever_chunking_model_id_rag_chunking_model',
+            'rag_chunking_model', ['chunking_model_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_rag_dense_retriever',
+            ['class_name', 'parameters', 'document_ids',
+             'chunking_model_id', 'embedding_model_id'],
+        )
+
+    with op.batch_alter_table('rag_sparse_retriever', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_rag_sparse_retriever', type_='unique')
+        batch_op.drop_constraint(
+            'fk_rag_sparse_retriever_chunk_set_id', type_='foreignkey',
+        )
+        batch_op.drop_column('chunk_set_id')
+        batch_op.add_column(sa.Column('chunking_model_id', sa.Integer(), nullable=False))
+        batch_op.add_column(sa.Column('documents_ids', sa.JSON(), nullable=False))
+        batch_op.create_foreign_key(
+            'fk_rag_sparse_retriever_chunking_model_id_rag_chunking_model',
+            'rag_chunking_model', ['chunking_model_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_rag_sparse_retriever',
+            ['class_name', 'parameters', 'documents_ids', 'chunking_model_id'],
+        )
+
+    with op.batch_alter_table('chunk', schema=None) as batch_op:
+        batch_op.drop_constraint('uix_chunk_set_doc_index', type_='unique')
+        batch_op.drop_constraint('fk_chunk_chunk_set_id', type_='foreignkey')
+        batch_op.drop_column('chunk_set_id')
+        batch_op.drop_column('metadata')
+        batch_op.add_column(sa.Column('chunking_model_id', sa.Integer(), nullable=False))
+        batch_op.create_foreign_key(
+            'fk_chunk_chunking_model_id_rag_chunking_model',
+            'rag_chunking_model', ['chunking_model_id'], ['id'], ondelete='CASCADE',
+        )
+        batch_op.create_unique_constraint(
+            'uix_document_chunk_index_chunking',
+            ['document_id', 'chunk_index', 'chunking_model_id'],
+        )
+
+    op.drop_table('rag_chunk_set_document')
+    op.drop_table('rag_chunk_set')
diff --git a/DashAI/alembic/versions/f06652057903_add_rag_tables.py b/DashAI/alembic/versions/f06652057903_add_rag_tables.py
new file mode 100644
index 000000000..e489b4de5
--- /dev/null
+++ b/DashAI/alembic/versions/f06652057903_add_rag_tables.py
@@ -0,0 +1,166 @@
+"""add rag tables
+
+Revision ID: f06652057903
+Revises: b4f9e70098e7
+Create Date: 2026-04-17 02:11:23.522691
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = 'f06652057903'
+down_revision: Union[str, None] = 'b4f9e70098e7'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.create_table('document',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('file_name', sa.String(), nullable=False),
+    sa.Column('file_type', sa.String(), nullable=False),
+    sa.Column('file_path', sa.String(), nullable=False),
+    sa.Column('file_hash', sa.String(), nullable=False),
+    sa.Column('optional_metadata', sa.JSON(), nullable=True),
+    sa.Column('created', sa.DateTime(), nullable=False),
+    sa.Column('last_modified', sa.DateTime(), nullable=False),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_document')),
+    sa.UniqueConstraint('file_hash', name=op.f('uq_document_file_hash'))
+    )
+    op.create_table('rag_chunking_model',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_chunking_model'))
+    )
+    op.create_table('rag_embedding_model',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_embedding_model'))
+    )
+    op.create_table('rag_generation_model',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_generation_model'))
+    )
+    op.create_table('rag_prompt',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('name', sa.String(), nullable=True),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.Column('created', sa.DateTime(), nullable=False),
+    sa.Column('last_modified', sa.DateTime(), nullable=False),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_prompt'))
+    )
+    op.create_table('chunk',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('document_id', sa.Integer(), nullable=False),
+    sa.Column('chunk_index', sa.Integer(), nullable=False),
+    sa.Column('chunking_model_id', sa.Integer(), nullable=False),
+    sa.Column('text', sa.Text(), nullable=False),
+    sa.ForeignKeyConstraint(['chunking_model_id'], ['rag_chunking_model.id'], name=op.f('fk_chunk_chunking_model_id_rag_chunking_model'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['document_id'], ['document.id'], name=op.f('fk_chunk_document_id_document'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_chunk')),
+    sa.UniqueConstraint('id', 'document_id', name='uix_chunk_document')
+    )
+    op.create_table('rag_dense_retriever',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.Column('embedding_model_id', sa.Integer(), nullable=False),
+    sa.Column('document_ids', sa.JSON(), nullable=False),
+    sa.Column('chunking_model_id', sa.Integer(), nullable=False),
+    sa.ForeignKeyConstraint(['chunking_model_id'], ['rag_chunking_model.id'], name=op.f('fk_rag_dense_retriever_chunking_model_id_rag_chunking_model'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['embedding_model_id'], ['rag_embedding_model.id'], name=op.f('fk_rag_dense_retriever_embedding_model_id_rag_embedding_model'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_dense_retriever'))
+    )
+    op.create_table('rag_embedding_matrix',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('document_id', sa.Integer(), nullable=False),
+    sa.Column('chunking_model_id', sa.Integer(), nullable=False),
+    sa.Column('embedding_model_id', sa.Integer(), nullable=False),
+    sa.Column('storage_folder', sa.String(), nullable=False),
+    sa.Column('matrix_shape', sa.JSON(), nullable=False),
+    sa.Column('created', sa.DateTime(), nullable=False),
+    sa.Column('last_modified', sa.DateTime(), nullable=False),
+    sa.ForeignKeyConstraint(['chunking_model_id'], ['rag_chunking_model.id'], name=op.f('fk_rag_embedding_matrix_chunking_model_id_rag_chunking_model'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['document_id'], ['document.id'], name=op.f('fk_rag_embedding_matrix_document_id_document'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['embedding_model_id'], ['rag_embedding_model.id'], name=op.f('fk_rag_embedding_matrix_embedding_model_id_rag_embedding_model'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_embedding_matrix')),
+    sa.UniqueConstraint('document_id', 'chunking_model_id', 'embedding_model_id', name='uix_document_chunking_embedding')
+    )
+    op.create_table('rag_sparse_retriever',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.Column('storage_folder', sa.String(), nullable=False),
+    sa.Column('documents_ids', sa.JSON(), nullable=False),
+    sa.Column('chunking_model_id', sa.Integer(), nullable=False),
+    sa.ForeignKeyConstraint(['chunking_model_id'], ['rag_chunking_model.id'], name=op.f('fk_rag_sparse_retriever_chunking_model_id_rag_chunking_model'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_sparse_retriever'))
+    )
+    op.create_table('rag_retriever',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('class_name', sa.String(), nullable=False),
+    sa.Column('dense_retriever_id', sa.Integer(), nullable=True),
+    sa.Column('sparse_retriever_id', sa.Integer(), nullable=True),
+    sa.CheckConstraint("(class_name = 'SparseRetriever' AND sparse_retriever_id IS NOT NULL AND dense_retriever_id IS NULL) OR (class_name = 'DenseRetriever' AND dense_retriever_id IS NOT NULL AND sparse_retriever_id IS NULL)", name=op.f('ck_chk_retriever_type')),
+    sa.ForeignKeyConstraint(['dense_retriever_id'], ['rag_dense_retriever.id'], name=op.f('fk_rag_retriever_dense_retriever_id_rag_dense_retriever'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['sparse_retriever_id'], ['rag_sparse_retriever.id'], name=op.f('fk_rag_retriever_sparse_retriever_id_rag_sparse_retriever'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_retriever'))
+    )
+    op.create_table('rag_pipeline',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('session_id', sa.Integer(), nullable=False),
+    sa.Column('name', sa.String(), nullable=False),
+    sa.Column('description', sa.String(), nullable=True),
+    sa.Column('parameters', sa.JSON(), nullable=True),
+    sa.Column('chunking_model_id', sa.Integer(), nullable=False),
+    sa.Column('retriever_model_id', sa.Integer(), nullable=False),
+    sa.Column('prompt_id', sa.Integer(), nullable=False),
+    sa.Column('generation_model_id', sa.Integer(), nullable=False),
+    sa.ForeignKeyConstraint(['chunking_model_id'], ['rag_chunking_model.id'], name=op.f('fk_rag_pipeline_chunking_model_id_rag_chunking_model'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['generation_model_id'], ['rag_generation_model.id'], name=op.f('fk_rag_pipeline_generation_model_id_rag_generation_model'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['prompt_id'], ['rag_prompt.id'], name=op.f('fk_rag_pipeline_prompt_id_rag_prompt'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['retriever_model_id'], ['rag_retriever.id'], name=op.f('fk_rag_pipeline_retriever_model_id_rag_retriever'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['session_id'], ['generative_session.id'], name=op.f('fk_rag_pipeline_session_id_generative_session'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_pipeline'))
+    )
+    op.create_table('rag_document_pipeline_session_link',
+    sa.Column('id', sa.Integer(), autoincrement=True, nullable=False),
+    sa.Column('document_id', sa.Integer(), nullable=False),
+    sa.Column('session_id', sa.Integer(), nullable=False),
+    sa.Column('pipeline_id', sa.Integer(), nullable=False),
+    sa.ForeignKeyConstraint(['document_id'], ['document.id'], name=op.f('fk_rag_document_pipeline_session_link_document_id_document'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['pipeline_id'], ['rag_pipeline.id'], name=op.f('fk_rag_document_pipeline_session_link_pipeline_id_rag_pipeline'), ondelete='CASCADE'),
+    sa.ForeignKeyConstraint(['session_id'], ['generative_session.id'], name=op.f('fk_rag_document_pipeline_session_link_session_id_generative_session'), ondelete='CASCADE'),
+    sa.PrimaryKeyConstraint('id', name=op.f('pk_rag_document_pipeline_session_link')),
+    sa.UniqueConstraint('document_id', 'pipeline_id', name='uix_document_pipeline'),
+    sa.UniqueConstraint('document_id', 'session_id', name='uix_document_session'),
+    sa.UniqueConstraint('session_id', 'pipeline_id', name='uix_session_pipeline')
+    )
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_table('rag_document_pipeline_session_link')
+    op.drop_table('rag_pipeline')
+    op.drop_table('rag_retriever')
+    op.drop_table('rag_sparse_retriever')
+    op.drop_table('rag_embedding_matrix')
+    op.drop_table('rag_dense_retriever')
+    op.drop_table('chunk')
+    op.drop_table('rag_prompt')
+    op.drop_table('rag_generation_model')
+    op.drop_table('rag_embedding_model')
+    op.drop_table('rag_chunking_model')
+    op.drop_table('document')
+    # ### end Alembic commands ###
diff --git a/DashAI/alembic/versions/f928e0b5203d_rag_tables_added.py b/DashAI/alembic/versions/f928e0b5203d_rag_tables_added.py
new file mode 100644
index 000000000..4e7eb013d
--- /dev/null
+++ b/DashAI/alembic/versions/f928e0b5203d_rag_tables_added.py
@@ -0,0 +1,36 @@
+"""RAG tables added
+
+Revision ID: f928e0b5203d
+Revises: f06652057903
+Create Date: 2026-04-24 18:55:45.236676
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = 'f928e0b5203d'
+down_revision: Union[str, None] = 'f06652057903'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('plugin', schema=None) as batch_op:
+        batch_op.add_column(sa.Column('latest_version', sa.String(), nullable=False))
+        batch_op.drop_column('lastest_version')
+
+    # ### end Alembic commands ###
+
+
+def downgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    with op.batch_alter_table('plugin', schema=None) as batch_op:
+        batch_op.add_column(sa.Column('lastest_version', sa.VARCHAR(), nullable=False))
+        batch_op.drop_column('latest_version')
+
+    # ### end Alembic commands ###
diff --git a/DashAI/back/api/api_v1/api.py b/DashAI/back/api/api_v1/api.py
index 1beb875a5..87b7ee2dc 100644
--- a/DashAI/back/api/api_v1/api.py
+++ b/DashAI/back/api/api_v1/api.py
@@ -5,6 +5,7 @@
 from DashAI.back.api.api_v1.endpoints.datafile import router as datafile_router
 from DashAI.back.api.api_v1.endpoints.dataset_source import router as dataset_source
 from DashAI.back.api.api_v1.endpoints.datasets import router as datasets
+from DashAI.back.api.api_v1.endpoints.documents import router as documents
 from DashAI.back.api.api_v1.endpoints.explainers import router as explainers
 from DashAI.back.api.api_v1.endpoints.explorers import router as explorers
 from DashAI.back.api.api_v1.endpoints.folders import router as folders
@@ -22,6 +23,7 @@
 from DashAI.back.api.api_v1.endpoints.pipelines import router as pipelines
 from DashAI.back.api.api_v1.endpoints.plugins import router as plugins
 from DashAI.back.api.api_v1.endpoints.predict import router as predict
+from DashAI.back.api.api_v1.endpoints.prompts import router as prompts
 from DashAI.back.api.api_v1.endpoints.runs import router as runs
 from DashAI.back.api.api_v1.endpoints.scoring import router as scoring
 
@@ -29,6 +31,7 @@
 api_router_v1.include_router(converters, prefix="/converter")
 api_router_v1.include_router(components, prefix="/component")
 api_router_v1.include_router(datasets, prefix="/dataset")
+api_router_v1.include_router(documents, prefix="/document")
 api_router_v1.include_router(model_sessions, prefix="/model-session")
 api_router_v1.include_router(explainers, prefix="/explainer")
 api_router_v1.include_router(explorers, prefix="/explorer")
@@ -39,6 +42,7 @@
 api_router_v1.include_router(generative_process, prefix="/generative-process")
 api_router_v1.include_router(pipelines, prefix="/pipelines")
 api_router_v1.include_router(plugins, prefix="/plugin")
+api_router_v1.include_router(prompts, prefix="/prompt")
 api_router_v1.include_router(notebook, prefix="/notebook")
 api_router_v1.include_router(metrics, prefix="/metrics")
 api_router_v1.include_router(hardware, prefix="/hardware")
diff --git a/DashAI/back/api/api_v1/endpoints/components.py b/DashAI/back/api/api_v1/endpoints/components.py
index 3c2d5cc53..f80446c49 100644
--- a/DashAI/back/api/api_v1/endpoints/components.py
+++ b/DashAI/back/api/api_v1/endpoints/components.py
@@ -235,6 +235,8 @@ async def get_components(
         for component_dict in selected_components.values()
     ]
 
+    return out
+
 
 @router.get("/{id}/")
 @inject
@@ -315,6 +317,25 @@ async def update_component() -> None:
         status_code=status.HTTP_501_NOT_IMPLEMENTED, detail="Method not implemented"
     )
 
+@router.get("/{component_name}/children/")
+async def get_child_components(
+    component_name:str,
+    recursive:bool=False,
+    component_registry: "ComponentRegistry" = Depends(lambda: di["component_registry"])):
+    """Get child components of a specific component.
+
+    Args:
+        component_name (str): The name of the component to get children for.
+        recursive (bool): Whether to get child components recursively.
+
+    Returns:
+        List[Dict[str, Any]]: A list of child component dictionaries.
+    """
+    children_list = component_registry.get_child_components(component_name, recursive=recursive)
+    # Remove "class" key from each child component
+    cleaned_children = [_delete_class(child) for child in children_list]
+
+    return cleaned_children
 
 @router.get("/image/{component_name}/", status_code=status.HTTP_200_OK)
 async def get_component_image(
diff --git a/DashAI/back/api/api_v1/endpoints/documents.py b/DashAI/back/api/api_v1/endpoints/documents.py
new file mode 100644
index 000000000..41d29b632
--- /dev/null
+++ b/DashAI/back/api/api_v1/endpoints/documents.py
@@ -0,0 +1,469 @@
+import json
+import logging
+import os
+from datetime import datetime
+from typing import Any, Dict, List
+from urllib.parse import quote
+
+from fastapi import (
+    APIRouter,
+    Depends,
+    File,
+    Form,
+    HTTPException,
+    Request,
+    Response,
+    UploadFile,
+    status,
+)
+from fastapi.responses import FileResponse
+from kink import di
+from sqlalchemy import exc
+from sqlalchemy.orm import sessionmaker
+
+from DashAI.back.api.api_v1.schemas import DocumentResponse
+from DashAI.back.dependencies.database.models import (
+    Document as DocumentDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    GenerativeSession,
+)
+from DashAI.back.models.RAG.utils import hash_function
+
+router = APIRouter()
+log = logging.getLogger(__name__)
+
+
+base_url = "/api/v1/document"
+
+
+@router.get("/", response_model=List[DocumentResponse])
+async def get_all_documents(
+    request: Request,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get all documents with file_url included."""
+    with session_factory() as db:
+        try:
+            documents: List[DocumentDBModel] = db.query(DocumentDBModel).all()
+
+            documents_responses = []
+            base = str(request.base_url).rstrip("/")
+            for doc in documents:
+                documents_responses.append(
+                    DocumentResponse(
+                        id=doc.id,
+                        file_name=doc.file_name,
+                        file_type=doc.file_type,
+                        file_hash=doc.file_hash,
+                        created=doc.created,
+                        last_modified=doc.last_modified,
+                        optional_metadata=doc.optional_metadata,
+                        related_sessions=[
+                            session.id for session in doc.get_related_sessions
+                        ]
+                        if getattr(doc, "get_related_sessions", None)
+                        else None,
+                        file_url=f"{base}{base_url}/{doc.id}/download",
+                    )
+                )
+
+            return documents_responses
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.get("/{document_id}", response_model=DocumentResponse)
+async def get_document(
+    document_id: int,
+    request: Request,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get metadata of a document by its ID."""
+    with session_factory() as db:
+        try:
+            document = db.get(DocumentDBModel, document_id)
+            if not document:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Document {document_id} does not exist in DB.",
+                )
+            base = str(request.base_url).rstrip("/")
+            return DocumentResponse(
+                id=document.id,
+                file_name=document.file_name,
+                file_type=document.file_type,
+                file_hash=document.file_hash,
+                created=document.created,
+                last_modified=document.last_modified,
+                optional_metadata=document.optional_metadata,
+                related_sessions=[
+                    session.id for session in document.get_related_sessions
+                ]
+                if getattr(document, "get_related_sessions", None)
+                else None,
+                file_url=f"{base}{base_url}/{document.id}/download",
+            )
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.get("/{document_id}/download", response_class=FileResponse)
+async def download_document(
+    document_id: int,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Download the actual file content of a document."""
+    with session_factory() as db:
+        document = db.query(DocumentDBModel).filter_by(id=document_id).first()
+        if not document:
+            raise HTTPException(
+                status_code=404,
+                detail=f"Document {document_id} not found",
+            )
+        if not os.path.exists(document.file_path):
+            raise HTTPException(
+                status_code=500,
+                detail=f"File not found at {document.file_path}",
+            )
+        # Detect media type by file extension
+        ext = os.path.splitext(document.file_name)[1].lower()
+        if ext == ".txt":
+            media_type = "text/plain"
+        elif ext == ".pdf":
+            media_type = "application/pdf"
+        else:
+            media_type = "application/octet-stream"
+
+        encoded_name = quote(document.file_name, safe="")
+        if ext == ".pdf":
+            return FileResponse(
+                path=document.file_path,
+                media_type=media_type,
+                headers={
+                    "Content-Disposition": (f"inline; filename*=UTF-8''{encoded_name}"),
+                },
+            )
+        else:
+            return FileResponse(
+                path=document.file_path,
+                media_type=media_type,
+                headers={
+                    "Content-Disposition": (
+                        f"attachment; filename*=UTF-8''{encoded_name}"
+                    ),
+                },
+            )
+
+
+@router.post("/", response_model=DocumentResponse, status_code=status.HTTP_201_CREATED)
+async def upload_document(
+    file: UploadFile = File(...),
+    metadata: str = Form(...),
+    config: Dict[str, Any] = Depends(lambda: di["config"]),
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+    request: Request = None,  # type: ignore[assignment]
+):
+    """
+    Upload a new document to the RAG system with file content and metadata.
+    The metadata should be a JSON string with required fields and an 'optional_metadata'
+    dict.
+    """
+    docs_folder_path = config["DOCUMENTS_PATH"]
+    if not docs_folder_path.exists():
+        raise HTTPException(
+            status_code=500,
+            detail=f"Documents folder {docs_folder_path} does not exist.",
+        )
+    try:
+        metadata_dict = json.loads(metadata)
+    except json.JSONDecodeError:
+        raise HTTPException(
+            status_code=400, detail="Invalid JSON in metadata"
+        ) from None
+
+    file_name = metadata_dict.get("file_name")
+    last_modified = metadata_dict.get("last_modified")
+    if not file_name or not last_modified:
+        raise HTTPException(
+            status_code=400,
+            detail="Missing required metadata fields: 'filename' and 'last_modified'",
+        )
+    optional_metadata = metadata_dict.get("optional_metadata", {})
+    if not isinstance(optional_metadata, dict):
+        raise HTTPException(
+            status_code=400,
+            detail="'optional_metadata' must be a dictionary",
+        )
+    try:
+        content_bytes = await file.read()
+    except Exception:
+        raise HTTPException(
+            status_code=400, detail="Failed to read file content"
+        ) from None
+
+    file_content_hash = hash_function(content_bytes)
+    base = str(request.base_url).rstrip("/") if request else ""
+    with session_factory() as db:
+        existing_doc: DocumentDBModel = (
+            db.query(DocumentDBModel).filter_by(file_hash=file_content_hash).first()
+        )
+        if existing_doc:
+            # Update existing document's information
+            existing_doc.file_name = file_name
+            existing_doc.optional_metadata = optional_metadata
+            existing_doc.created = datetime.now()
+            db.commit()
+            return DocumentResponse(
+                id=existing_doc.id,
+                file_name=existing_doc.file_name,
+                file_type=existing_doc.file_type,
+                file_hash=existing_doc.file_hash,
+                created=existing_doc.created,
+                last_modified=existing_doc.last_modified,
+                related_sessions=existing_doc.get_related_sessions,
+                optional_metadata=existing_doc.optional_metadata,
+                file_url=f"{base}{base_url}/{existing_doc.id}/download",
+            )
+        else:
+            # Create a new document entry
+            try:
+                file_path = docs_folder_path / file_name
+
+                with open(file_path, "wb") as f:
+                    f.write(content_bytes)
+
+                # Get file type from file extension
+                file_type = os.path.splitext(file_name)[1].lstrip(".") or "unknown"
+
+                doc = DocumentDBModel(
+                    file_name=file_name,
+                    file_type=file_type,
+                    file_path=str(file_path),
+                    file_hash=file_content_hash,
+                    optional_metadata=optional_metadata or None,
+                )
+                db.add(doc)
+                db.commit()
+                db.refresh(doc)
+
+                return DocumentResponse(
+                    id=doc.id,
+                    file_name=doc.file_name,
+                    file_type=doc.file_type,
+                    file_hash=doc.file_hash,
+                    created=doc.created,
+                    last_modified=doc.last_modified,
+                    related_sessions=doc.get_related_sessions,
+                    optional_metadata=doc.optional_metadata,
+                    file_url=f"{base}{base_url}/{doc.id}/download",
+                )
+            except exc.SQLAlchemyError as e:
+                log.exception(e)
+                raise HTTPException(
+                    status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                    detail="Internal database error",
+                ) from e
+
+
+@router.get("/session/{session_id}", response_model=List[DocumentResponse])
+async def get_documents_by_session(
+    session_id: int,
+    request: Request,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get all documents associated with a specific RAG session."""
+    with session_factory() as db:
+        try:
+            # Get the session
+            session = db.query(GenerativeSession).filter_by(id=session_id).first()
+            if not session:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Session {session_id} does not exist.",
+                )
+
+            # Extract document IDs from session parameters
+            document_ids = session.parameters.get("documents", [])
+            if not document_ids:
+                return []
+
+            # Query documents by IDs
+            documents = (
+                db.query(DocumentDBModel)
+                .filter(DocumentDBModel.id.in_(document_ids))
+                .all()
+            )
+
+            documents_responses = []
+            base = str(request.base_url).rstrip("/")
+            for doc in documents:
+                documents_responses.append(
+                    DocumentResponse(
+                        id=doc.id,
+                        file_name=doc.file_name,
+                        file_type=doc.file_type,
+                        file_hash=doc.file_hash,
+                        created=doc.created,
+                        last_modified=doc.last_modified,
+                        optional_metadata=doc.optional_metadata,
+                        related_sessions=[
+                            session.id for session in doc.get_related_sessions
+                        ]
+                        if doc.pipeline_links
+                        else None,
+                        file_url=f"{base}{base_url}/{doc.id}/download",
+                    )
+                )
+
+            return documents_responses
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.get("/related-sessions/{document_id}", response_model=List[int])
+async def get_related_sessions(
+    document_id: int,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get all generative session IDs related to a specific document."""
+    with session_factory() as db:
+        try:
+            document = db.query(DocumentDBModel).filter_by(id=document_id).first()
+            if not document:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Document {document_id} does not exist in DB.",
+                )
+            if not document.get_related_sessions:
+                return []
+            related_session_ids = [
+                session.id for session in document.get_related_sessions
+            ]
+            return related_session_ids
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.delete("/{document_id}", status_code=status.HTTP_204_NO_CONTENT)
+async def delete_document(
+    document_id: int,
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Delete a document from the RAG system by its ID."""
+    with session_factory() as db:
+        try:
+            document = db.query(DocumentDBModel).filter_by(id=document_id).first()
+            if not document:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Document {document_id} does not exist in DB.",
+                )
+
+            # Delete the physical file
+            if os.path.exists(document.file_path):
+                os.remove(document.file_path)
+
+            # Delete the database record
+            db.delete(document)
+            db.commit()
+
+            return Response(status_code=status.HTTP_204_NO_CONTENT)
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+        except OSError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Error deleting physical file",
+            ) from e
+
+
+@router.put("/{document_id}", response_model=DocumentResponse)
+async def update_document_metadata(
+    document_id: int,
+    metadata: str = Form(...),
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+    request: Request = None,  # type: ignore[assignment]
+):
+    """Update a document's metadata."""
+    base = str(request.base_url).rstrip("/") if request else ""
+    with session_factory() as db:
+        try:
+            document = db.query(DocumentDBModel).filter_by(id=document_id).first()
+            if not document:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail=f"Document {document_id} does not exist in DB.",
+                )
+
+            try:
+                metadata_dict = json.loads(metadata)
+            except json.JSONDecodeError:
+                raise HTTPException(
+                    status_code=400, detail="Invalid JSON in metadata"
+                ) from None
+
+            # Update fields that are allowed to be modified
+            if "file_name" in metadata_dict:
+                document.file_name = metadata_dict["file_name"]
+                # Update file_type when file_name changes
+                document.file_type = (
+                    os.path.splitext(metadata_dict["file_name"])[1].lstrip(".")
+                    or "unknown"
+                )
+
+            if "optional_metadata" in metadata_dict:
+                optional_metadata = metadata_dict["optional_metadata"]
+                if not isinstance(optional_metadata, dict):
+                    raise HTTPException(
+                        status_code=400,
+                        detail="'optional_metadata' must be a dictionary",
+                    )
+                document.optional_metadata = optional_metadata
+
+            document.last_modified = datetime.now()
+
+            db.commit()
+            db.refresh(document)
+
+            return DocumentResponse(
+                id=document.id,
+                file_name=document.file_name,
+                file_hash=document.file_hash,
+                created=document.created,
+                optional_metadata=document.optional_metadata,
+                related_sessions=[
+                    session.id for session in document.get_related_sessions
+                ]
+                if document.get_related_sessions
+                else None,
+                file_url=f"{base}{base_url}/{document.id}/download",
+            )
+
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
diff --git a/DashAI/back/api/api_v1/endpoints/generative_process.py b/DashAI/back/api/api_v1/endpoints/generative_process.py
index 1a1791906..58cc3cf1a 100644
--- a/DashAI/back/api/api_v1/endpoints/generative_process.py
+++ b/DashAI/back/api/api_v1/endpoints/generative_process.py
@@ -121,7 +121,7 @@ async def upload_generative_process(
 
 @router.get("/{process_id}", status_code=status.HTTP_200_OK, response_model=None)
 async def get_generative_process(
-    process_id: str,
+    process_id: int,
     session_factory: "sessionmaker" = Depends(lambda: di["session_factory"]),
     component_registry: "ComponentRegistry" = Depends(lambda: di["component_registry"]),
 ):
@@ -185,7 +185,7 @@ async def get_generative_process(
     "/{process_id}", status_code=status.HTTP_204_NO_CONTENT, response_model=None
 )
 async def delete_generative_process(
-    process_id: str,
+    process_id: int,
     session_factory: "sessionmaker" = Depends(lambda: di["session_factory"]),
 ):
     """Delete a generative process by its ID.
@@ -233,7 +233,7 @@ async def delete_generative_process(
     "/session/{session_id}", status_code=status.HTTP_200_OK, response_model=None
 )
 async def get_generative_process_by_session_id(
-    session_id: str,
+    session_id: int,
     session_factory: "sessionmaker" = Depends(lambda: di["session_factory"]),
     component_registry: "ComponentRegistry" = Depends(lambda: di["component_registry"]),
 ):
diff --git a/DashAI/back/api/api_v1/endpoints/generative_session.py b/DashAI/back/api/api_v1/endpoints/generative_session.py
index 8211dbcf9..b5a1feb41 100644
--- a/DashAI/back/api/api_v1/endpoints/generative_session.py
+++ b/DashAI/back/api/api_v1/endpoints/generative_session.py
@@ -1,6 +1,12 @@
 import logging
+import shutil
 from datetime import datetime
-from typing import TYPE_CHECKING, Union
+from pathlib import Path
+from typing import TYPE_CHECKING, Final, Union
+
+COMPOSITE_RETRIEVER_NAMES: Final = frozenset(
+    {"SequentialRetriever", "ParallelRetriever"}
+)
 
 from fastapi import APIRouter, Depends, HTTPException, status
 from kink import di
@@ -10,11 +16,22 @@
     GenerativeSessionParams,
 )
 from DashAI.back.dependencies.database.models import (
+    Document,
     GenerativeProcess,
     GenerativeSession,
     GenerativeSessionParameterHistory,
     ProcessData,
+    RAGChunkingModel,
+    RAGDenseRetriever,
+    RAGEmbeddingMatrix,
+    RAGEmbeddingModel,
+    RAGPipeline,
+    RAGPrompt,
+    RAGRetriever,
+    RAGRetrieverChild,
+    RAGSparseRetriever,
 )
+from DashAI.back.tasks.RAG_task import RAGTask
 
 if TYPE_CHECKING:
     from sqlalchemy.orm import sessionmaker
@@ -26,6 +43,254 @@
 log = logging.getLogger(__name__)
 
 
+def _delete_path(path_value: str | None) -> None:
+    if not path_value:
+        return
+    path = Path(path_value)
+    if path.exists():
+        shutil.rmtree(path, ignore_errors=True)
+
+
+def _other_sessions_with_same_config(
+    db,
+    session_id: int,
+    expected_parameters: dict,
+    *,
+    keys: tuple[str, ...],
+) -> bool:
+    """Return True when any other session matches all keys in `keys`.
+
+    This avoids generator/list comprehensions for clarity.
+    """
+    other_sessions = (
+        db.query(GenerativeSession).filter(GenerativeSession.id != session_id).all()
+    )
+
+    for other_session in other_sessions:
+        other_parameters = other_session.parameters or {}
+        all_keys_match = True
+        for key in keys:
+            if other_parameters.get(key) != expected_parameters.get(key):
+                all_keys_match = False
+                break
+        if all_keys_match:
+            return True
+
+    return False
+
+
+def _cleanup_orphaned_rag_resources(
+    db,
+    session_id: int,
+    old_parameters: dict,
+    new_parameters: dict | None = None,
+) -> None:
+    if not old_parameters:
+        return
+
+    def _component_changed(key: str) -> bool:
+        if new_parameters is None:
+            return True
+        return old_parameters.get(key) != new_parameters.get(key)
+
+    other_sessions = (
+        db.query(GenerativeSession).filter(GenerativeSession.id != session_id).all()
+    )
+
+    documents_ids = old_parameters.get("documents") or []
+    documents_ids = sorted(documents_ids)
+
+    # ── Retriever cleanup MUST run BEFORE chunking cleanup ──────────
+    # _cleanup_unit_retriever queries RAGChunkingModel to obtain
+    # chunking_model_id. If chunking models are deleted first, the
+    # query returns None and retriever rows are silently skipped.
+    # ─────────────────────────────────────────────────────────────────
+    retriever_model_params = old_parameters.get("retriever_model")
+    if not retriever_model_params:
+        retriever_model_params = {}
+
+    retriever_component_name = retriever_model_params.get("component", "")
+
+    should_cleanup_retriever = (
+        bool(retriever_model_params)
+        and _component_changed("retriever_model")
+        and not _other_sessions_with_same_config(
+            db,
+            session_id,
+            old_parameters,
+            keys=("documents", "chunking_model", "retriever_model"),
+        )
+    )
+
+    if should_cleanup_retriever:
+        if retriever_component_name in COMPOSITE_RETRIEVER_NAMES:
+            _cleanup_composite_retriever(db, old_parameters, documents_ids, session_id)
+        else:
+            _cleanup_unit_retriever(
+                db, old_parameters, documents_ids, retriever_model_params, session_id
+            )
+
+    # ── Chunking model cleanup (AFTER retriever) ──────────────────────
+    chunking_model_params = old_parameters.get("chunking_model")
+    if not chunking_model_params:
+        chunking_model_params = {}
+
+    should_cleanup_chunking = (
+        bool(chunking_model_params)
+        and _component_changed("chunking_model")
+        and not _other_sessions_with_same_config(
+            db, session_id, old_parameters, keys=("documents", "chunking_model")
+        )
+    )
+
+    if should_cleanup_chunking:
+        chunking_models = (
+            db.query(RAGChunkingModel)
+            .filter(
+                RAGChunkingModel.class_name == chunking_model_params.get("component"),
+                RAGChunkingModel.parameters == chunking_model_params.get("params"),
+            )
+            .all()
+        )
+        for chunking_model in chunking_models:
+            db.delete(chunking_model)
+
+
+def _find_pipeline_id(db, session_id: int) -> int | None:
+    pipeline = db.query(RAGPipeline).filter_by(session_id=session_id).first()
+    return pipeline.id if pipeline else None
+
+
+def _cleanup_composite_retriever(
+    db,
+    old_parameters,
+    documents_ids,
+    session_id,
+) -> None:
+    pipeline_id = _find_pipeline_id(db, session_id)
+    if pipeline_id is None:
+        return
+
+    retriever_component_name = old_parameters.get("retriever_model", {}).get(
+        "component"
+    )
+
+    composite_bridges = (
+        db.query(RAGRetriever)
+        .filter(
+            RAGRetriever.pipeline_id == pipeline_id,
+            RAGRetriever.class_name == retriever_component_name,
+        )
+        .all()
+    )
+
+    for bridge in composite_bridges:
+        child_links = (
+            db.query(RAGRetrieverChild)
+            .filter_by(parent_id=bridge.id)
+            .order_by(RAGRetrieverChild.child_order)
+            .all()
+        )
+        for link in child_links:
+            child_bridge = db.query(RAGRetriever).get(link.child_id)
+            if child_bridge is None:
+                continue
+            if child_bridge.sparse_detail:
+                _delete_path(child_bridge.sparse_detail.storage_folder)
+                db.delete(child_bridge.sparse_detail)
+            elif child_bridge.dense_detail:
+                db.delete(child_bridge.dense_detail)
+        db.delete(bridge)
+
+
+def _cleanup_unit_retriever(
+    db,
+    old_parameters,
+    documents_ids,
+    retriever_model_params,
+    session_id,
+) -> None:
+    # Previously queried non-existent columns (chunking_model_id, document_ids)
+    # on RAGDenseRetriever/RAGSparseRetriever. Fixed after schema refactor:
+    # trace through the pipeline chain to obtain the valid chunk_set_id column.
+    retriever_params = retriever_model_params.get("params", {})
+
+    pipeline_id = _find_pipeline_id(db, session_id)
+    if pipeline_id is None:
+        return
+
+    bridge = (
+        db.query(RAGRetriever)
+        .filter(
+            RAGRetriever.pipeline_id == pipeline_id,
+            RAGRetriever.class_name == retriever_model_params.get("component"),
+        )
+        .first()
+    )
+    if bridge is None:
+        return
+
+    if "encoding_model" in retriever_params:
+        dense_retriever = (
+            db.query(RAGDenseRetriever)
+            .filter(
+                RAGDenseRetriever.bridge_id == bridge.id,
+                RAGDenseRetriever.class_name == retriever_model_params.get("component"),
+                RAGDenseRetriever.parameters == retriever_params,
+            )
+            .first()
+        )
+        if dense_retriever is None:
+            return
+
+        chunk_set_id = dense_retriever.chunk_set_id
+        embedding_model_id = dense_retriever.embedding_model_id
+
+        embedding_matrices = (
+            db.query(RAGEmbeddingMatrix)
+            .filter(
+                RAGEmbeddingMatrix.chunk_set_id == chunk_set_id,
+                RAGEmbeddingMatrix.embedding_model_id == embedding_model_id,
+                RAGEmbeddingMatrix.document_id.in_(documents_ids),
+            )
+            .all()
+        )
+
+        matrix_ids = []
+        for matrix in embedding_matrices:
+            _delete_path(matrix.storage_folder)
+            matrix_ids.append(matrix.id)
+
+        if matrix_ids:
+            db.query(RAGEmbeddingMatrix).filter(
+                RAGEmbeddingMatrix.id.in_(matrix_ids)
+            ).delete(synchronize_session=False)
+
+        embedding_model = db.query(RAGEmbeddingModel).get(embedding_model_id)
+        if embedding_model is not None:
+            db.delete(embedding_model)
+
+        db.delete(dense_retriever)
+        db.delete(bridge)
+    else:
+        sparse_retriever = (
+            db.query(RAGSparseRetriever)
+            .filter(
+                RAGSparseRetriever.bridge_id == bridge.id,
+                RAGSparseRetriever.class_name
+                == retriever_model_params.get("component"),
+                RAGSparseRetriever.parameters == retriever_params,
+            )
+            .first()
+        )
+        if sparse_retriever is None:
+            return
+
+        _delete_path(sparse_retriever.storage_folder)
+        db.delete(sparse_retriever)
+        db.delete(bridge)
+
+
 @router.post("/", status_code=status.HTTP_201_CREATED)
 async def upload_generative_session(
     params: GenerativeSessionParams,
@@ -55,22 +320,53 @@ async def upload_generative_session(
                     f"generative model.",
                 )
 
-            # Validate the model parameters
+            # Check if the task is registered
             try:
-                model_class.SCHEMA.model_validate(params.parameters)
-            except ValueError as e:
+                task_class = component_registry[params.task_name]["class"]
+            except KeyError as e:
                 raise HTTPException(
                     status_code=status.HTTP_400_BAD_REQUEST,
-                    detail=f"Invalid parameters for model {params.model_name}: {e}",
+                    detail=f"Task {params.task_name} is not registered.",
                 ) from e
 
-            # Check if the task is registered
+            # RAG Task specific handling
+            # Frontend will send the ids of the documents to be used in the
+            # RAG session but RAG pipeline expects the documents paths of the
+            # backend-stored documents
+            if task_class == RAGTask:
+                try:
+                    assert params.parameters["documents"]
+                    assert isinstance(params.parameters["documents"], list)
+                    assert len(params.parameters["documents"]) > 0
+
+                except (AssertionError, KeyError) as e:
+                    raise HTTPException(
+                        status_code=status.HTTP_400_BAD_REQUEST,
+                        detail="RAG Task requires a non-empty list of document IDs.",
+                    ) from e
+
+                documents_ids = []
+                for doc_id in params.parameters["documents"]:
+                    document = db.get(Document, doc_id)
+                    if not document:
+                        raise HTTPException(
+                            status_code=status.HTTP_400_BAD_REQUEST,
+                            detail=f"Document with ID {doc_id} does not exist.",
+                        )
+                    documents_ids.append(document.id)
+
+                params.parameters["documents"] = documents_ids
+            # Continue with the session creation
+
+            # Normalise frontend properties wrapper and validate
+            from DashAI.back.core.schema_fields.utils import normalize_payload
+            params.parameters = normalize_payload(params.parameters)
             try:
-                task_class = component_registry[params.task_name]["class"]
-            except KeyError as e:
+                model_class.SCHEMA.model_validate(params.parameters)
+            except ValueError as e:
                 raise HTTPException(
                     status_code=status.HTTP_400_BAD_REQUEST,
-                    detail=f"Task {params.task_name} is not registered.",
+                    detail=f"Invalid parameters for model {params.model_name}: {e}",
                 ) from e
 
             # Check if the task is a subclass of BaseGenerativeTask
@@ -257,6 +553,8 @@ async def delete_generative_session(
                     detail=f"Generative session {session_id} does not exist in DB.",
                 )
 
+            old_parameters = dict(session.parameters or {})
+
             # Delete all the processes associated with the session
             processes = (
                 db.query(GenerativeProcess)
@@ -286,6 +584,7 @@ async def delete_generative_session(
                 db.delete(entry)
             # Finally, delete the session itself
             db.delete(session)
+            _cleanup_orphaned_rag_resources(db, session_id, old_parameters)
             db.commit()
         except exc.SQLAlchemyError as e:
             log.exception(e)
@@ -405,6 +704,7 @@ async def update_generative_session_params(
     session_id: int,
     new_params: dict,
     session_factory: "sessionmaker" = Depends(lambda: di["session_factory"]),
+    component_registry: "ComponentRegistry" = Depends(lambda: di["component_registry"]),
 ):
     """Update the parameters of a generative session and log the change.
 
@@ -438,7 +738,43 @@ async def update_generative_session_params(
                     detail=f"Generative session {session_id} does not exist in DB.",
                 )
 
-            updated_parameters = {**session.parameters, **new_params}
+            old_parameters = dict(session.parameters or {})
+            updated_parameters = {**old_parameters, **new_params}
+
+            if "prompt_id" in updated_parameters:
+                if "prompt" not in updated_parameters:
+                    prompt_db = db.get(RAGPrompt, updated_parameters["prompt_id"])
+                    if prompt_db:
+                        raw_params = dict(prompt_db.parameters or {})
+                        prompt_params = {
+                            "template": raw_params.get(
+                                "template",
+                                raw_params.get("templates", {}).get("en", ""),
+                            ),
+                        }
+                        if "language" in raw_params:
+                            prompt_params["language"] = raw_params["language"]
+                        updated_parameters["prompt"] = {
+                            "component": prompt_db.class_name,
+                            "params": prompt_params,
+                        }
+                del updated_parameters["prompt_id"]
+
+            try:
+                task_class = component_registry[session.task_name]["class"]
+            except KeyError:
+                task_class = None
+
+            if task_class is not None and task_class == RAGTask:
+                from DashAI.back.models.RAG.RAG_pipeline import RAGPipeline
+
+                try:
+                    RAGPipeline.SCHEMA.model_validate(updated_parameters)
+                except ValueError as e:
+                    raise HTTPException(
+                        status_code=status.HTTP_400_BAD_REQUEST,
+                        detail=f"Invalid parameters for RAG session {session_id}: {e}",
+                    ) from e
 
             session_params_entry = GenerativeSessionParameterHistory(
                 session_id=session.id,
@@ -450,11 +786,17 @@ async def update_generative_session_params(
             session.parameters = updated_parameters
             session.last_modified = datetime.now()
 
+            _cleanup_orphaned_rag_resources(
+                db, session_id, old_parameters, updated_parameters
+            )
             db.commit()
             db.refresh(session)
 
             return {"id": session.id, "parameters": session.parameters}
+        except HTTPException:
+            raise
         except exc.SQLAlchemyError as e:
+            db.rollback()
             log.exception(e)
             raise HTTPException(
                 status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
@@ -506,16 +848,18 @@ async def get_generative_session_parameters_history(
                 .all()
             )
 
-            # Convert the objects to dictionaries
-            return [
-                {
-                    "id": entry.id,
-                    "session_id": entry.session_id,
-                    "parameters": entry.parameters,
-                    "modified_at": entry.modified_at,
-                }
-                for entry in parameters_history
-            ]
+            # Convert the objects to dictionaries (explicit loop for clarity)
+            history_list = []
+            for entry in parameters_history:
+                history_list.append(
+                    {
+                        "id": entry.id,
+                        "session_id": entry.session_id,
+                        "parameters": entry.parameters,
+                        "modified_at": entry.modified_at,
+                    }
+                )
+            return history_list
         except exc.SQLAlchemyError as e:
             log.exception(e)
             raise HTTPException(
@@ -570,13 +914,18 @@ async def get_parameter_history_entry(
                 .all()
             )
 
-            parameters_history = [p.__dict__ for p in parameters_history]
+            params_history_dicts = []
+            for p in parameters_history:
+                params_history_dicts.append(dict(p.__dict__))
 
             events = []
-            prev_params = parameters_history[0]["parameters"]
+            if not params_history_dicts:
+                return []
+
+            prev_params = params_history_dicts[0]["parameters"]
 
-            for i in range(1, len(parameters_history)):
-                curr = parameters_history[i]
+            for i in range(1, len(params_history_dicts)):
+                curr = params_history_dicts[i]
                 curr_params = curr["parameters"]
                 changes = []
 
diff --git a/DashAI/back/api/api_v1/endpoints/prompts.py b/DashAI/back/api/api_v1/endpoints/prompts.py
new file mode 100644
index 000000000..e3170390a
--- /dev/null
+++ b/DashAI/back/api/api_v1/endpoints/prompts.py
@@ -0,0 +1,365 @@
+import logging
+from datetime import datetime
+from typing import Any, Dict
+
+from fastapi import APIRouter, Depends, HTTPException, status
+from kink import di
+from sqlalchemy import exc, select
+from sqlalchemy.orm import sessionmaker
+
+from DashAI.back.api.api_v1.schemas.rag_prompt import (
+    RAGPromptSchema,
+    RAGPromptUpdateSchema,
+)
+from DashAI.back.dependencies.database.models import (
+    GenerativeSession,
+    GenerativeSessionParameterHistory,
+    RAGPrompt,
+)
+from DashAI.back.dependencies.registry import ComponentRegistry
+from DashAI.back.models.RAG.prompts.generation.default_QA_rag_generation_prompt import (
+    DefaultQARAGGenerationPrompt,
+)
+from DashAI.back.models.RAG.prompts.generation.default_rag_generation_prompt import (
+    DefaultRAGGenerationPrompt,
+)
+from DashAI.back.models.RAG.prompts.prompt import Prompt
+
+router = APIRouter()
+log = logging.getLogger(__name__)
+
+
+def _validate_prompt_template(
+    prompt_class_name: str,
+    template: str,
+    component_registry: ComponentRegistry,
+) -> None:
+    if prompt_class_name not in component_registry:
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"Component {prompt_class_name} is not registered.",
+        )
+
+    prompt_component = component_registry[prompt_class_name]
+    prompt_class = prompt_component["class"]
+    if not issubclass(prompt_class, Prompt):
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=f"Component {prompt_class_name} is not a valid Prompt subclass.",
+        )
+
+    if not prompt_class.validate_template(template):
+        raise HTTPException(
+            status_code=status.HTTP_400_BAD_REQUEST,
+            detail=(
+                f"Invalid template for prompt {prompt_class_name}. "
+                f"Required tokens are: {prompt_class.get_required_placeholders()}"
+            ),
+        )
+
+
+def _serialize_prompt(prompt: RAGPrompt) -> Dict[str, Any]:
+    return {
+        "id": prompt.id,
+        "class_name": prompt.class_name,
+        "name": prompt.name,
+        "parameters": prompt.parameters,
+        "created": prompt.created,
+        "last_modified": prompt.last_modified,
+    }
+
+
+def _build_session_prompt_name(base_name: str, session_id: int, db) -> str:
+    candidate = f"{base_name} - session {session_id}"
+    suffix = 2
+    while db.execute(select(RAGPrompt.id).where(RAGPrompt.name == candidate)).scalar():
+        candidate = f"{base_name} - session {session_id} ({suffix})"
+        suffix += 1
+    return candidate
+
+
+@router.post("/", status_code=status.HTTP_201_CREATED)
+async def create_rag_prompt(
+    prompt: RAGPromptSchema,
+    component_registry: ComponentRegistry = Depends(lambda: di["component_registry"]),
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Create a new RAGPrompt entry in the database."""
+    with session_factory() as db:
+        if not prompt.parameters:
+            raise HTTPException(
+                status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                detail="Prompt parameters are required and must include a template.",
+            )
+
+        if "templates" in prompt.parameters:
+            for _lang, tmpl in prompt.parameters["templates"].items():
+                _validate_prompt_template(prompt.class_name, tmpl, component_registry)
+        elif "template" in prompt.parameters:
+            _validate_prompt_template(
+                prompt.class_name,
+                prompt.parameters["template"],
+                component_registry,
+            )
+        else:
+            raise HTTPException(
+                status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                detail="Prompt parameters must include 'template' or 'templates'.",
+            )
+        new_prompt = RAGPrompt(
+            class_name=prompt.class_name,
+            name=prompt.name,
+            parameters=prompt.parameters,
+        )
+        db.add(new_prompt)
+        db.commit()
+        db.refresh(new_prompt)
+        return {"id": new_prompt.id}
+
+
+@router.patch("/{prompt_id}", status_code=status.HTTP_200_OK)
+async def update_rag_prompt(
+    prompt_id: int,
+    prompt: RAGPromptUpdateSchema,
+    component_registry: ComponentRegistry = Depends(lambda: di["component_registry"]),
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Update an existing prompt in place."""
+    with session_factory() as db:
+        try:
+            existing_prompt = db.get(RAGPrompt, prompt_id)
+            if existing_prompt is None:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail="Prompt not found",
+                )
+
+            changed = False
+
+            if prompt.name is not None:
+                new_name = prompt.name.strip()
+                if not new_name:
+                    raise HTTPException(
+                        status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                        detail="Prompt name cannot be empty",
+                    )
+                if new_name != existing_prompt.name:
+                    existing_prompt.name = new_name
+                    changed = True
+
+            if prompt.parameters is not None:
+                if "templates" in prompt.parameters:
+                    for _lang, tmpl in prompt.parameters["templates"].items():
+                        _validate_prompt_template(
+                            existing_prompt.class_name,
+                            tmpl,
+                            component_registry,
+                        )
+                elif "template" in prompt.parameters:
+                    _validate_prompt_template(
+                        existing_prompt.class_name,
+                        prompt.parameters["template"],
+                        component_registry,
+                    )
+                else:
+                    raise HTTPException(
+                        status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                        detail=(
+                            "Prompt parameters must include 'template' or 'templates'."
+                        ),
+                    )
+                existing_prompt.parameters = prompt.parameters
+                changed = True
+
+            if not changed:
+                raise HTTPException(
+                    status_code=status.HTTP_304_NOT_MODIFIED,
+                    detail="Record not modified",
+                )
+
+            existing_prompt.last_modified = datetime.now()
+            db.commit()
+            db.refresh(existing_prompt)
+            return _serialize_prompt(existing_prompt)
+        except HTTPException:
+            raise
+        except exc.SQLAlchemyError as e:
+            db.rollback()
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.post("/{prompt_id}/sessions/{session_id}", status_code=status.HTTP_201_CREATED)
+async def update_rag_prompt_for_session(
+    prompt_id: int,
+    session_id: int,
+    prompt: RAGPromptUpdateSchema,
+    component_registry: ComponentRegistry = Depends(lambda: di["component_registry"]),
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Create a session-scoped copy of a prompt and attach it to the session."""
+    with session_factory() as db:
+        try:
+            existing_prompt = db.get(RAGPrompt, prompt_id)
+            if existing_prompt is None:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail="Prompt not found",
+                )
+
+            session = db.get(GenerativeSession, session_id)
+            if session is None:
+                raise HTTPException(
+                    status_code=status.HTTP_404_NOT_FOUND,
+                    detail="Generative session not found",
+                )
+
+            if prompt.parameters is not None:
+                if "templates" in prompt.parameters:
+                    for _lang, tmpl in prompt.parameters["templates"].items():
+                        _validate_prompt_template(
+                            existing_prompt.class_name,
+                            tmpl,
+                            component_registry,
+                        )
+                elif "template" in prompt.parameters:
+                    _validate_prompt_template(
+                        existing_prompt.class_name,
+                        prompt.parameters["template"],
+                        component_registry,
+                    )
+                else:
+                    raise HTTPException(
+                        status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
+                        detail=(
+                            "Prompt parameters must include 'template' or 'templates'."
+                        ),
+                    )
+                new_parameters = prompt.parameters
+            else:
+                new_parameters = existing_prompt.parameters
+
+            # Make parameters unique per session to avoid UNIQUE constraint
+            new_parameters = dict(new_parameters or {})
+            new_parameters["cloned_for_session"] = session_id
+
+            if prompt.name:
+                candidate_name = prompt.name
+            elif existing_prompt.name:
+                candidate_name = existing_prompt.name
+            else:
+                candidate_name = existing_prompt.class_name
+
+            base_name = (candidate_name or "").strip()
+            new_name = _build_session_prompt_name(base_name, session_id, db)
+
+            new_prompt = RAGPrompt(
+                class_name=existing_prompt.class_name,
+                name=new_name,
+                parameters=new_parameters,
+            )
+            db.add(new_prompt)
+            db.commit()
+            db.refresh(new_prompt)
+
+            session_parameters = dict(session.parameters or {})
+            session_parameters["prompt_id"] = new_prompt.id
+            session.parameters = session_parameters
+            session.last_modified = datetime.now()
+            db.add(
+                GenerativeSessionParameterHistory(
+                    session_id=session.id,
+                    parameters=session_parameters,
+                    modified_at=datetime.now(),
+                )
+            )
+            db.commit()
+            db.refresh(session)
+
+            return {
+                "prompt": _serialize_prompt(new_prompt),
+                "session_id": session.id,
+                "parameters": session.parameters,
+            }
+        except HTTPException:
+            raise
+        except exc.SQLAlchemyError as e:
+            db.rollback()
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
+
+
+@router.get("/", status_code=status.HTTP_200_OK)
+async def get_all_prompts(
+    session_factory: sessionmaker = Depends(lambda: di["session_factory"]),
+):
+    """Get all available RAG prompts.
+
+    Parameters
+    ----------
+    session_factory : Callable[..., ContextManager[Session]]
+        A factory that creates a context manager that handles a SQLAlchemy session.
+        The generated session can be used to access and query the database.
+
+    Returns
+    -------
+    list
+        A list of dictionaries with all prompts on the database.
+
+    Raises
+    ------
+    HTTPException
+        If there's an internal database error.
+    """
+    with session_factory() as db:
+        try:
+            prompts = db.query(RAGPrompt).all()
+
+            if len(prompts) == 0:
+                default_generation_prompt = RAGPrompt(
+                    class_name=DefaultRAGGenerationPrompt.__name__,
+                    name="Default RAG Generation Prompt",
+                    parameters={
+                        "templates": DefaultRAGGenerationPrompt.metadata["templates"],
+                        "language": "en",
+                    },
+                )
+                default_qa_prompt = RAGPrompt(
+                    class_name=DefaultQARAGGenerationPrompt.__name__,
+                    name="Default QA RAG Generation Prompt",
+                    parameters={
+                        "templates": DefaultQARAGGenerationPrompt.metadata["templates"],
+                        "language": "en",
+                    },
+                )
+                db.add(default_generation_prompt)
+                db.add(default_qa_prompt)
+                db.commit()
+                prompts = db.query(RAGPrompt).all()
+
+            prompt_responses = []
+            for prompt in prompts:
+                prompt_responses.append(
+                    {
+                        "id": prompt.id,
+                        "class_name": prompt.class_name,
+                        "name": prompt.name,
+                        "parameters": prompt.parameters,
+                        "created": prompt.created,
+                        "last_modified": prompt.last_modified,
+                    }
+                )
+
+            return prompt_responses
+        except exc.SQLAlchemyError as e:
+            log.exception(e)
+            raise HTTPException(
+                status_code=status.HTTP_500_INTERNAL_SERVER_ERROR,
+                detail="Internal database error",
+            ) from e
diff --git a/DashAI/back/api/api_v1/schemas/__init__.py b/DashAI/back/api/api_v1/schemas/__init__.py
index e69de29bb..9d35150b8 100644
--- a/DashAI/back/api/api_v1/schemas/__init__.py
+++ b/DashAI/back/api/api_v1/schemas/__init__.py
@@ -0,0 +1 @@
+from DashAI.back.api.api_v1.schemas.document import DocumentResponse
\ No newline at end of file
diff --git a/DashAI/back/api/api_v1/schemas/document.py b/DashAI/back/api/api_v1/schemas/document.py
new file mode 100644
index 000000000..1bde38a24
--- /dev/null
+++ b/DashAI/back/api/api_v1/schemas/document.py
@@ -0,0 +1,15 @@
+from datetime import datetime
+from typing import Any, Dict, List, Optional
+
+from pydantic import BaseModel
+
+
+class DocumentResponse(BaseModel):
+    id: int
+    file_name: str
+    file_type: str
+    file_hash: str
+    created: datetime
+    optional_metadata: Optional[Dict[str, Any]]
+    related_sessions: List[int] | None
+    file_url: str
diff --git a/DashAI/back/api/api_v1/schemas/job_params.py b/DashAI/back/api/api_v1/schemas/job_params.py
index 167e54dab..816e0f22e 100644
--- a/DashAI/back/api/api_v1/schemas/job_params.py
+++ b/DashAI/back/api/api_v1/schemas/job_params.py
@@ -15,5 +15,6 @@ class JobParams(BaseModel):
         "ConverterJob",
         "GenerativeJob",
         "PipelineJob",
+        "RAGJob",
     ]
     kwargs: dict
diff --git a/DashAI/back/api/api_v1/schemas/rag_prompt.py b/DashAI/back/api/api_v1/schemas/rag_prompt.py
new file mode 100644
index 000000000..317c8d303
--- /dev/null
+++ b/DashAI/back/api/api_v1/schemas/rag_prompt.py
@@ -0,0 +1,14 @@
+from typing import Any, Dict, Optional
+
+from pydantic import BaseModel
+
+
+class RAGPromptSchema(BaseModel):
+    class_name: str
+    name: str
+    parameters: Optional[Dict[str, Any]] = None
+
+
+class RAGPromptUpdateSchema(BaseModel):
+    name: Optional[str] = None
+    parameters: Optional[Dict[str, Any]] = None
diff --git a/DashAI/back/app.py b/DashAI/back/app.py
index f4bc5088a..a6160f992 100644
--- a/DashAI/back/app.py
+++ b/DashAI/back/app.py
@@ -80,6 +80,7 @@ def create_app(
     _create_path_if_not_exists(config["EXPLANATIONS_PATH"])
     _create_path_if_not_exists(config["NOTEBOOK_PATH"])
     _create_path_if_not_exists(config["RUNS_PATH"])
+    _create_path_if_not_exists(config["DOCUMENTS_PATH"])
     _create_path_if_not_exists(config["DATAFILE_PATH"])
 
     logger.debug("3. Creating app container and setting up dependency injection.")
diff --git a/DashAI/back/config.py b/DashAI/back/config.py
index 728c8a760..b56a4821e 100644
--- a/DashAI/back/config.py
+++ b/DashAI/back/config.py
@@ -20,5 +20,8 @@ class DefaultSettings(BaseSettings):
     IMAGES_PATH: str = "images"
     RUNS_PATH: str = "runs"
     EXPLANATIONS_PATH: str = "explanations"
+    EXPLORATIONS_PATH: str = "explorations"
+    DOCUMENTS_PATH: str = "documents"
+    RAG_PATH: str = "rag"
     NOTEBOOK_PATH: str = "notebook"
     DATAFILE_PATH: str = "datafiles"
diff --git a/DashAI/back/config_object.py b/DashAI/back/config_object.py
index b994e8e93..3eb485d61 100644
--- a/DashAI/back/config_object.py
+++ b/DashAI/back/config_object.py
@@ -47,5 +47,11 @@ def validate_and_transform(self, raw_data: dict) -> dict:
         ValidationError
             If the given data does not follow the schema associated with the model.
         """
-        schema_instance = self.SCHEMA.model_validate(raw_data)
+        try:
+            from pydantic import ValidationError
+            schema_instance = self.SCHEMA.model_validate(raw_data)
+        except ValidationError as e:
+            # print full error message
+            print(e.json())
+            raise e
         return fill_objects(schema_instance)
diff --git a/DashAI/back/converters/hugging_face/embedding.py b/DashAI/back/converters/hugging_face/embedding.py
index cbc726d35..fe5f98d14 100644
--- a/DashAI/back/converters/hugging_face/embedding.py
+++ b/DashAI/back/converters/hugging_face/embedding.py
@@ -1,3 +1,9 @@
+from typing import List
+
+import numpy as np
+import torch
+from transformers import AutoModel, AutoTokenizer
+
 """HuggingFace embedding converter with lazy-loaded dependencies."""
 
 from typing import TYPE_CHECKING
@@ -178,12 +184,41 @@ def get_output_type(self, column_name: str = None) -> DashAIDataType:
 
     def _load_model(self):
         """Load the embedding model and tokenizer."""
-        from transformers import AutoModel, AutoTokenizer
 
         self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
         self.model = AutoModel.from_pretrained(self.model_name).to(self.device)
         self.model.eval()
 
+    def _encode_texts(self, texts:List[str])-> List[np.ndarray]:
+        """Encode a list of texts into embeddings."""
+        # Tokenize
+        encoded = self.tokenizer(
+            texts,
+            padding=True,
+            truncation=True,
+            max_length=self.max_length,
+            return_tensors="pt",
+        )
+
+        # Move to device
+        encoded = {k: v.to(self.device) for k, v in encoded.items()}
+
+        # Get embeddings
+        with torch.no_grad():
+            outputs = self.model(**encoded)
+            hidden_states = outputs.last_hidden_state
+
+            # Apply pooling strategy
+            if self.pooling_strategy == "mean":
+                embeddings = torch.mean(hidden_states, dim=1)
+            elif self.pooling_strategy == "cls":
+                embeddings = hidden_states[:, 0]
+            else:  # max pooling
+                embeddings = torch.max(hidden_states, dim=1)[0]
+
+        embeddings_np = embeddings.cpu().numpy()
+        return embeddings_np
+
     def _process_batch(self, batch: "DashAIDataset") -> "DashAIDataset":
         """Encode a batch of text columns into dense embedding vectors.
 
@@ -203,7 +238,6 @@ def _process_batch(self, batch: "DashAIDataset") -> "DashAIDataset":
             dense embedding vector column(s).
         """
         import pyarrow as pa
-        import torch
 
         from DashAI.back.dataloaders.classes.dashai_dataset import DashAIDataset
 
@@ -213,32 +247,7 @@ def _process_batch(self, batch: "DashAIDataset") -> "DashAIDataset":
             # Get text data from dataset
             texts = [row[column] if row[column] is not None else "" for row in batch]
 
-            # Tokenize
-            encoded = self.tokenizer(
-                texts,
-                padding=True,
-                truncation=True,
-                max_length=self.max_length,
-                return_tensors="pt",
-            )
-
-            # Move to device
-            encoded = {k: v.to(self.device) for k, v in encoded.items()}
-
-            # Get embeddings
-            with torch.no_grad():
-                outputs = self.model(**encoded)
-                hidden_states = outputs.last_hidden_state
-
-                # Apply pooling strategy
-                if self.pooling_strategy == "mean":
-                    embeddings = torch.mean(hidden_states, dim=1)
-                elif self.pooling_strategy == "cls":
-                    embeddings = hidden_states[:, 0]
-                else:  # max pooling
-                    embeddings = torch.max(hidden_states, dim=1)[0]
-
-            embeddings_np = embeddings.cpu().numpy()
+            embeddings_np = self._encode_texts(texts)
 
             # Append one column per embedding dimension
             for i in range(embeddings_np.shape[1]):
diff --git a/DashAI/back/core/schema_fields/base_schema.py b/DashAI/back/core/schema_fields/base_schema.py
index 3371ffd4b..3a4571c4f 100644
--- a/DashAI/back/core/schema_fields/base_schema.py
+++ b/DashAI/back/core/schema_fields/base_schema.py
@@ -46,3 +46,15 @@ def replace_defs_in_schema(schema: dict):
 
 class BaseSchema(BaseModel):
     pass
+
+    """ @classmethod
+    def model_validate(cls, raw_data: dict):
+        schema_fields = cls.model_fields
+        for field_name, field in schema_fields.items():
+            if field_name not in raw_data:
+                continue
+            if field.annotation._name == 'Optional':
+                if raw_data[field_name] == "":
+                    raw_data[field_name] = None
+
+        return super().model_validate(raw_data) """
diff --git a/DashAI/back/core/schema_fields/utils.py b/DashAI/back/core/schema_fields/utils.py
index 17ac63b39..f6ec0a601 100644
--- a/DashAI/back/core/schema_fields/utils.py
+++ b/DashAI/back/core/schema_fields/utils.py
@@ -7,6 +7,70 @@
     from DashAI.back.dependencies.registry import ComponentRegistry
 
 
+def normalize_payload(data):
+    """Recursively normalizes the frontend ``properties`` wrapper into the
+    canonical ``{component, params}`` format.
+
+    This is the factored-out version of the unwrapping logic found in
+    ``model_factory._process_param`` (approximately line 162).
+
+    The frontend auto-generated forms (``generateInitialValues`` in
+    ``utils/schema.js``) produce a ``properties`` wrapper structure:
+
+    .. code-block:: python
+
+        {
+            "properties": {
+                "component": "ParentModelName",
+                "params": {
+                    "comp": {
+                        "component": "ActualModelName",
+                        "params": { ... sub‑params ... },
+                    },
+                },
+            },
+        }
+
+    This function converts it into the canonical form that
+    ``fill_objects()`` and ``model_class.SCHEMA.model_validate()`` expect:
+
+    .. code-block:: python
+
+        {"component": "ActualModelName", "params": { ... sub‑params ... }}
+
+    Any scalar, list, or already-canonical dict values are returned
+    unchanged.
+
+    Parameters
+    ----------
+    data : Any
+        Arbitrary JSON-like data (output of ``model_dump()`` or raw
+        frontend payload) that may contain ``properties`` wrappers.
+
+    Returns
+    -------
+    Any
+        The same data with all ``properties`` wrappers replaced by the
+        canonical ``{component, params}`` form.
+    """
+    if isinstance(data, dict):
+        if "properties" in data and len(data) == 1:
+            inner = data["properties"]
+            if isinstance(inner, dict) and "component" in inner:
+                raw_params = inner.get("params", {})
+                comp = raw_params.get("comp", {})
+                if isinstance(comp, dict) and "component" in comp:
+                    return {
+                        "component": comp["component"],
+                        "params": normalize_payload(comp.get("params", {})),
+                    }
+                return normalize_payload(inner)
+        return {k: normalize_payload(v) for k, v in data.items()}
+    if isinstance(data, list):
+        return [normalize_payload(item) for item in data]
+    return data
+
+
 @inject
 def fill_objects(
     schema_instance: "BaseSchema",
diff --git a/DashAI/back/dependencies/config_builder.py b/DashAI/back/dependencies/config_builder.py
index 3323340b9..69e729451 100644
--- a/DashAI/back/dependencies/config_builder.py
+++ b/DashAI/back/dependencies/config_builder.py
@@ -63,5 +63,7 @@ def build_config_dict(
     config["BACK_PATH"] = pathlib.Path(config["BACK_PATH"]).absolute()
     config["LOGGING_LEVEL"] = getattr(logging, logging_level)
     config["INITIAL_COMPONENTS"] = get_initial_components()
+    config["DOCUMENTS_PATH"] = local_path / config["DOCUMENTS_PATH"]
+    config["RAG_PATH"] = local_path / config["RAG_PATH"]
 
     return config
diff --git a/DashAI/back/dependencies/database/models.py b/DashAI/back/dependencies/database/models.py
index cdf57d968..938b0c1fd 100644
--- a/DashAI/back/dependencies/database/models.py
+++ b/DashAI/back/dependencies/database/models.py
@@ -7,6 +7,7 @@
     JSON,
     BigInteger,
     Boolean,
+    Column,
     DateTime,
     Enum,
     Float,
@@ -307,7 +308,7 @@ class Plugin(Base):
         Boolean, nullable=False, default=False, server_default="0"
     )
     installed_version: Mapped[str] = mapped_column(String, nullable=False)
-    lastest_version: Mapped[str] = mapped_column(String, nullable=False)
+    latest_version: Mapped[str] = mapped_column(String, nullable=False)
     tags: Mapped[List["Tag"]] = relationship(
         back_populates="plugin", cascade="all, delete", lazy="selectin"
     )
@@ -547,6 +548,11 @@ class GenerativeSession(Base):
         "GenerativeProcess", cascade="all, delete-orphan", back_populates="session"
     )
 
+    # Relationship with RAGDocumentPipelineSessionLink
+    pipeline_links: Mapped[List["RAGDocumentPipelineSessionLink"]] = relationship(
+        back_populates="session"
+    )
+
 
 class Pipeline(Base):
     __tablename__ = "pipeline"
@@ -746,6 +752,517 @@ def delete_result(self) -> None:
             self.end_time = None
 
 
+"""
+RAG tables
+"""
+
+
+class Document(Base):
+    __tablename__ = "document"
+    """
+    Table to store all the information about a document.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    file_name: Mapped[str] = mapped_column(String, nullable=False)
+    file_type: Mapped[str] = mapped_column(String, nullable=False)
+    file_path: Mapped[str] = mapped_column(String, nullable=False)
+    file_hash: Mapped[str] = mapped_column(String, nullable=False, unique=True)
+    optional_metadata: Mapped[Dict[str, Any]] = mapped_column(JSON, nullable=True)
+    created: Mapped[DateTime] = mapped_column(DateTime, default=datetime.now)
+    last_modified: Mapped[DateTime] = mapped_column(
+        DateTime,
+        default=datetime.now,
+        onupdate=datetime.now,
+    )
+
+    # Create a relationship for the related sessions and related chunks
+    pipeline_links: Mapped[List["RAGDocumentPipelineSessionLink"]] = relationship(
+        "RAGDocumentPipelineSessionLink",
+        cascade="all, delete-orphan",
+        back_populates="document",
+    )
+
+    chunks: Mapped[List["Chunk"]] = relationship(
+        "Chunk",
+        back_populates="document",
+        cascade="all, delete-orphan",
+    )
+
+    embedding_matrices: Mapped[List["RAGEmbeddingMatrix"]] = relationship(
+        "RAGEmbeddingMatrix", back_populates="document", cascade="all, delete-orphan"
+    )
+
+    @property
+    def get_related_sessions(self) -> List["GenerativeSession"]:
+        """Return a list of sessions related to the document."""
+        return [link.session for link in self.pipeline_links]
+
+    @property
+    def get_related_pipelines(self) -> List["RAGPipeline"]:
+        """Return a list of pipelines related to the document."""
+        return [link.pipeline for link in self.pipeline_links]
+
+    def get_embedding_matrix(
+        self, chunk_set_id: int, embedding_model_id: int
+    ) -> Optional["RAGEmbeddingMatrix"]:
+        for matrix in self.embedding_matrices:
+            if (
+                matrix.chunk_set_id == chunk_set_id
+                and matrix.embedding_model_id == embedding_model_id
+            ):
+                return matrix
+        return None
+
+
+class Chunk(Base):
+    __tablename__ = "chunk"
+    """
+    Table to store all the information about a chunk.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False
+    )
+    document_id: Mapped[int] = mapped_column(
+        ForeignKey("document.id", ondelete="CASCADE"), nullable=False
+    )
+    chunk_index: Mapped[int] = mapped_column(Integer, nullable=False)
+    text: Mapped[str] = mapped_column(Text, nullable=False)
+    chunk_metadata: Mapped[Optional[dict]] = mapped_column("metadata", JSON, nullable=True)
+
+    document: Mapped["Document"] = relationship("Document", back_populates="chunks")
+    chunk_set: Mapped["RAGChunkSet"] = relationship("RAGChunkSet", back_populates="chunks")
+
+    __table_args__ = (
+        UniqueConstraint(
+            "chunk_set_id", "document_id", "chunk_index",
+            name="uix_chunk_set_doc_index",
+        ),
+    )
+
+
+class RAGChunkSet(Base):
+    __tablename__ = "rag_chunk_set"
+    """
+    Canonical identity for a set of chunks produced by a processing pipeline.
+    Two sessions with the same documents + same pipeline config share the same
+    chunk set (same signature → same row).
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    signature: Mapped[str] = mapped_column(String, nullable=False, unique=True)
+    parameters: Mapped[Optional[dict]] = mapped_column(JSON, nullable=True)
+
+    chunks: Mapped[List["Chunk"]] = relationship(
+        "Chunk", back_populates="chunk_set", cascade="all, delete-orphan",
+    )
+    documents: Mapped[List["RAGChunkSetDocument"]] = relationship(
+        "RAGChunkSetDocument", back_populates="chunk_set", cascade="all, delete-orphan",
+    )
+    embedding_matrices: Mapped[List["RAGEmbeddingMatrix"]] = relationship(
+        "RAGEmbeddingMatrix", back_populates="chunk_set",
+    )
+    retriever_links: Mapped[List["RAGRetrieverChunkSet"]] = relationship(
+        "RAGRetrieverChunkSet", back_populates="chunk_set", cascade="all, delete-orphan",
+    )
+
+
+class RAGChunkSetDocument(Base):
+    __tablename__ = "rag_chunk_set_document"
+    """Which documents belong to a chunk set."""
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False
+    )
+    document_id: Mapped[int] = mapped_column(
+        ForeignKey("document.id", ondelete="CASCADE"), nullable=False
+    )
+
+    chunk_set: Mapped["RAGChunkSet"] = relationship(
+        "RAGChunkSet", back_populates="documents",
+    )
+    document: Mapped["Document"] = relationship("Document")
+
+    __table_args__ = (
+        UniqueConstraint(
+            "chunk_set_id", "document_id",
+            name="uix_chunk_set_document",
+        ),
+    )
+
+
+class RAGRetrieverChunkSet(Base):
+    __tablename__ = "rag_retriever_chunk_set"
+    """Links a retriever to the chunk set it operates on."""
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    retriever_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_retriever.id", ondelete="CASCADE"), nullable=False
+    )
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False
+    )
+
+    chunk_set: Mapped["RAGChunkSet"] = relationship(
+        "RAGChunkSet", back_populates="retriever_links",
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "retriever_id", "chunk_set_id",
+            name="uix_retriever_chunk_set",
+        ),
+    )
+
+
+class RAGPrompt(Base):
+    __tablename__ = "rag_prompt"
+    """
+    Table to store all the information about a RAG prompt.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    name: Mapped[str] = mapped_column(String, nullable=True)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+    created: Mapped[DateTime] = mapped_column(DateTime, default=datetime.now)
+    last_modified: Mapped[DateTime] = mapped_column(
+        DateTime,
+        default=datetime.now,
+        onupdate=datetime.now,
+    )
+
+    # Relationship with RAGPipeline
+    pipelines: Mapped[List["RAGPipeline"]] = relationship(
+        "RAGPipeline", back_populates="prompt", cascade="all, delete-orphan"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters",
+            name="uix_rag_prompt_class_params",
+        ),
+    )
+
+
+class RAGGenerationModel(Base):
+    __tablename__ = "rag_generation_model"
+    """
+    Table to store all the information about a generation model.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+
+    # Relationships
+    pipelines: Mapped[List["RAGPipeline"]] = relationship(
+        back_populates="generation_model"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters",
+            name="uix_rag_gen_model_class_params",
+        ),
+    )
+
+
+class RAGPipeline(Base):
+    __tablename__ = "rag_pipeline"
+    """
+    Table to store all the information about a RAG pipeline.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    session_id: Mapped[int] = mapped_column(
+        ForeignKey("generative_session.id", ondelete="CASCADE"), nullable=False
+    )
+    name: Mapped[str] = mapped_column(String, nullable=False)
+    description: Mapped[str] = mapped_column(String, nullable=True)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+
+    chunking_model_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunking_model.id", ondelete="CASCADE"), nullable=False
+    )
+    prompt_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_prompt.id", ondelete="CASCADE"), nullable=False
+    )
+    generation_model_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_generation_model.id", ondelete="CASCADE"), nullable=False
+    )
+
+    # Relationships
+    chunking_model: Mapped["RAGChunkingModel"] = relationship(
+        "RAGChunkingModel", back_populates="pipelines"
+    )
+    retriever: Mapped["RAGRetriever"] = relationship(
+        "RAGRetriever", back_populates="pipeline", uselist=False,
+    )
+    prompt: Mapped["RAGPrompt"] = relationship("RAGPrompt", back_populates="pipelines")
+    generation_model: Mapped["RAGGenerationModel"] = relationship(
+        "RAGGenerationModel", back_populates="pipelines"
+    )
+    pipeline_links: Mapped[List["RAGDocumentPipelineSessionLink"]] = relationship(
+        "RAGDocumentPipelineSessionLink", back_populates="pipeline"
+    )
+
+
+class RAGChunkingModel(Base):
+    __tablename__ = "rag_chunking_model"
+    """
+    Table to store all the information about a chunking model.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+
+    pipelines: Mapped[List["RAGPipeline"]] = relationship(
+        "RAGPipeline", back_populates="chunking_model"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters",
+            name="uix_rag_chunking_model_class_params",
+        ),
+    )
+
+
+class RAGRetriever(Base):
+    __tablename__ = "rag_retriever"
+    """
+    Canonical identity row for every retriever — unit (sparse/dense) or composite.
+
+    Sub-tables (RAGSparseRetriever, RAGDenseRetriever) reference this row
+    via bridge_id. Composite children are stored in rag_retriever_child.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    pipeline_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_pipeline.id", ondelete="CASCADE"), nullable=False
+    )
+
+    pipeline: Mapped["RAGPipeline"] = relationship(back_populates="retriever")
+    children: Mapped[List["RAGRetrieverChild"]] = relationship(
+        "RAGRetrieverChild", foreign_keys="RAGRetrieverChild.parent_id",
+        back_populates="parent", cascade="all, delete-orphan",
+    )
+    sparse_detail: Mapped[Optional["RAGSparseRetriever"]] = relationship(
+        "RAGSparseRetriever", back_populates="bridge", uselist=False,
+    )
+    dense_detail: Mapped[Optional["RAGDenseRetriever"]] = relationship(
+        "RAGDenseRetriever", back_populates="bridge", uselist=False,
+    )
+
+
+class RAGRetrieverChild(Base):
+    __tablename__ = "rag_retriever_child"
+    """
+    Link table for composite retrievers: maps a parent composite to its
+    ordered child retrievers. Both parent and child reference rag_retriever.id.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    parent_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_retriever.id", ondelete="CASCADE"), nullable=False,
+    )
+    child_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_retriever.id", ondelete="CASCADE"), nullable=False,
+    )
+    child_order: Mapped[int] = mapped_column(Integer, nullable=False)
+
+    parent: Mapped["RAGRetriever"] = relationship(
+        "RAGRetriever", foreign_keys=[parent_id], back_populates="children",
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "parent_id", "child_order",
+            name="uix_retriever_child_order",
+        ),
+    )
+
+
+class RAGSparseRetriever(Base):
+    __tablename__ = "rag_sparse_retriever"
+    """
+    Table to store all the information about a sparse retriever.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    bridge_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_retriever.id", ondelete="CASCADE"), nullable=False,
+    )
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False,
+    )
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+    storage_folder: Mapped[str] = mapped_column(String, nullable=False)
+
+    bridge: Mapped["RAGRetriever"] = relationship(
+        "RAGRetriever", back_populates="sparse_detail",
+    )
+    chunk_set: Mapped["RAGChunkSet"] = relationship("RAGChunkSet")
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters", "chunk_set_id",
+            name="uix_rag_sparse_retriever",
+        ),
+    )
+
+
+class RAGDenseRetriever(Base):
+    __tablename__ = "rag_dense_retriever"
+    """
+    Table to store all the information about a dense retriever.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    bridge_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_retriever.id", ondelete="CASCADE"), nullable=False,
+    )
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False,
+    )
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+    embedding_model_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_embedding_model.id", ondelete="CASCADE"), nullable=False
+    )
+    bridge: Mapped["RAGRetriever"] = relationship(
+        "RAGRetriever", back_populates="dense_detail",
+    )
+    chunk_set: Mapped["RAGChunkSet"] = relationship("RAGChunkSet")
+    embedding_model: Mapped["RAGEmbeddingModel"] = relationship(
+        "RAGEmbeddingModel", back_populates="dense_retrievers"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters", "chunk_set_id", "embedding_model_id",
+            name="uix_rag_dense_retriever",
+        ),
+    )
+
+
+class RAGEmbeddingModel(Base):
+    __tablename__ = "rag_embedding_model"
+    """
+    Table to store embedding model configurations.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    class_name: Mapped[str] = mapped_column(String, nullable=False)
+    parameters: Mapped[JSON] = mapped_column(JSON, nullable=True)
+
+    # Relationships
+    embedding_matrices: Mapped[List["RAGEmbeddingMatrix"]] = relationship(
+        back_populates="embedding_model", cascade="all, delete-orphan"
+    )
+    dense_retrievers: Mapped[List["RAGDenseRetriever"]] = relationship(
+        back_populates="embedding_model"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "class_name", "parameters",
+            name="uix_rag_embedding_model_class_params",
+        ),
+    )
+
+
+class RAGEmbeddingMatrix(Base):
+    __tablename__ = "rag_embedding_matrix"
+    """
+    Table to store embedding matrices for each unique combination of
+    document, chunk set, and embedding model.
+    """
+    id: Mapped[int] = mapped_column(primary_key=True, autoincrement=True)
+    document_id: Mapped[int] = mapped_column(
+        ForeignKey("document.id", ondelete="CASCADE"), nullable=False
+    )
+    chunk_set_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_chunk_set.id", ondelete="CASCADE"), nullable=False
+    )
+    embedding_model_id: Mapped[int] = mapped_column(
+        ForeignKey("rag_embedding_model.id", ondelete="CASCADE"), nullable=False
+    )
+    storage_folder: Mapped[str] = mapped_column(String, nullable=False)
+    matrix_shape: Mapped[List[int]] = mapped_column(
+        JSON, nullable=False
+    )
+    created: Mapped[DateTime] = mapped_column(DateTime, default=datetime.now)
+    last_modified: Mapped[DateTime] = mapped_column(
+        DateTime, default=datetime.now, onupdate=datetime.now,
+    )
+
+    document: Mapped["Document"] = relationship(
+        "Document", back_populates="embedding_matrices"
+    )
+    chunk_set: Mapped["RAGChunkSet"] = relationship(
+        "RAGChunkSet", back_populates="embedding_matrices",
+    )
+    embedding_model: Mapped["RAGEmbeddingModel"] = relationship(
+        "RAGEmbeddingModel", back_populates="embedding_matrices"
+    )
+
+    __table_args__ = (
+        UniqueConstraint(
+            "document_id", "chunk_set_id", "embedding_model_id",
+            name="uix_document_chunk_set_embedding",
+        ),
+    )
+
+    @property
+    def num_chunks(self) -> int:
+        """Return the number of chunks in this embedding matrix."""
+        return self.matrix_shape[0] if self.matrix_shape else 0
+
+    @property
+    def embedding_dimension(self) -> int:
+        """Return the embedding dimension."""
+        return self.matrix_shape[1] if len(self.matrix_shape) > 1 else 0
+
+    @classmethod
+    def get_by_tuple(
+        cls, session, document_id: int, chunk_set_id: int, embedding_model_id: int
+    ) -> Optional["RAGEmbeddingMatrix"]:
+        return (
+            session.query(cls)
+            .filter(
+                cls.document_id == document_id,
+                cls.chunk_set_id == chunk_set_id,
+                cls.embedding_model_id == embedding_model_id,
+            )
+            .first()
+        )
+
+
+"""
+RAG relationship tables
+"""
+
+
+class RAGDocumentPipelineSessionLink(Base):
+    __tablename__ = "rag_document_pipeline_session_link"
+
+    id = Column(Integer, primary_key=True, autoincrement=True)
+    document_id = Column(
+        Integer, ForeignKey("document.id", ondelete="CASCADE"), nullable=False
+    )
+    session_id = Column(
+        Integer, ForeignKey("generative_session.id", ondelete="CASCADE"), nullable=False
+    )
+    pipeline_id = Column(
+        Integer, ForeignKey("rag_pipeline.id", ondelete="CASCADE"), nullable=False
+    )
+
+    # Relationships
+    document = relationship("Document", back_populates="pipeline_links")
+    pipeline = relationship("RAGPipeline", back_populates="pipeline_links")
+    session = relationship("GenerativeSession", back_populates="pipeline_links")
+
+    __table_args__ = (
+        UniqueConstraint("document_id", "session_id", name="uix_document_session"),
+        UniqueConstraint("session_id", "pipeline_id", name="uix_session_pipeline"),
+        UniqueConstraint("document_id", "pipeline_id", name="uix_document_pipeline"),
+    )
+
+
 class Datafile(Base):
     __tablename__ = "datafile"
 
diff --git a/DashAI/back/dependencies/database/utils.py b/DashAI/back/dependencies/database/utils.py
index 184034def..a5a42ee22 100644
--- a/DashAI/back/dependencies/database/utils.py
+++ b/DashAI/back/dependencies/database/utils.py
@@ -64,7 +64,7 @@ def add_plugin_to_db(
                     author=raw_plugin.author,
                     verified=raw_plugin.verified,
                     installed_version=raw_plugin.installed_version,
-                    lastest_version=raw_plugin.lastest_version,
+                    latest_version=raw_plugin.lastest_version,
                     summary=raw_plugin.summary,
                     description=raw_plugin.description,
                     description_content_type=raw_plugin.description_content_type,
diff --git a/DashAI/back/dependencies/registry/component_registry.py b/DashAI/back/dependencies/registry/component_registry.py
index d99145aaf..e9b1e0209 100644
--- a/DashAI/back/dependencies/registry/component_registry.py
+++ b/DashAI/back/dependencies/registry/component_registry.py
@@ -217,6 +217,7 @@ def register_component(self, new_component: Type) -> None:
             "metadata": _metadata,
             "description": getattr(new_component, "DESCRIPTION", None),
             "display_name": getattr(new_component, "DISPLAY_NAME", None),
+            "flags": getattr(new_component, "FLAGS", []),
             "color": getattr(new_component, "COLOR", None),
         }
 
diff --git a/DashAI/back/initial_components.py b/DashAI/back/initial_components.py
index 49ea258f1..160008c18 100644
--- a/DashAI/back/initial_components.py
+++ b/DashAI/back/initial_components.py
@@ -112,6 +112,7 @@
 from DashAI.back.job.model_job import ModelJob
 from DashAI.back.job.pipeline_job import PipelineJob
 from DashAI.back.job.predict_job import PredictJob
+from DashAI.back.job.rag_job import RAGJob
 
 # Metrics
 from DashAI.back.metrics.classification.accuracy import Accuracy
@@ -172,6 +173,9 @@
 from DashAI.back.models.hugging_face.opus_mt_fr_en_transformer import (
     OpusMtFrEnTransformer,
 )
+from DashAI.back.models.hugging_face.phi_4_mini_instruct_model import (
+    Phi4MiniInstructModel,
+)
 from DashAI.back.models.hugging_face.pixart_sigma_model import PixArtSigmaModel
 from DashAI.back.models.hugging_face.qwen_model import QwenModel
 from DashAI.back.models.hugging_face.roberta_transformer import RobertaTransformer
@@ -209,6 +213,57 @@
 from DashAI.back.models.hugging_face.xlnet_transformer import XlnetTransformer
 from DashAI.back.models.lenet5_image_classifier import LeNet5ImageClassifier
 from DashAI.back.models.mlp_image_classifier import MLPImageClassifier
+from DashAI.back.models.RAG import RAGPipeline
+from DashAI.back.models.RAG.chunking_models import (
+    CharacterChunkModel,
+    RecursiveCharacterChunkModel,
+    TokenChunkModel,
+)
+from DashAI.back.models.RAG.embeddings.dense import (
+    BERTEmbedding,
+    DistilBERTEmbedding,
+    E5Embedding,
+    GemmaEmbedding,
+    HuggingFaceEmbedding,
+    InstructorEmbedding,
+    LaBSEmbedding,
+    OpenAIEmbedding,
+    RoBERTaEmbedding,
+    SentenceTransformerEmbedding,
+)
+from DashAI.back.models.RAG.prompts import (
+    CustomAugmentationPrompt,
+    CustomRAGGenerationPrompt,
+    DefaultAugmentationPrompt,
+    DefaultQARAGGenerationPrompt,
+    DefaultRAGGenerationPrompt,
+)
+from DashAI.back.models.RAG.retrievers.composite.mmr_reranker_retriever import (
+    MMRRerankerRetriever,
+)
+from DashAI.back.models.RAG.retrievers.composite.parallel_retriever import (
+    ParallelRetriever,
+)
+from DashAI.back.models.RAG.retrievers.composite.sequential_retriever import (
+    SequentialRetriever,
+)
+from DashAI.back.models.RAG.retrievers.dense.dense_embedding_retriever import (
+    DenseEmbeddingRetriever,
+)
+from DashAI.back.models.RAG.retrievers.sparse.bm25_retriever import (
+    BM25Retriever,
+    BM25VectorizerModel,
+)
+from DashAI.back.models.RAG.retrievers.sparse.tfidf_retriever import (
+    TFIDFRetriever,
+    TFIDFVectorizerModel,
+)
+from DashAI.back.models.remote_models.deepseek_text_to_text_generation_model import (
+    DeepSeekTextToTextGenerationModel,
+)
+from DashAI.back.models.remote_models.openai_text_to_text_generation_model import (
+    OpenAITextToTextGenerationModel,
+)
 from DashAI.back.models.resnet18_image_classifier import ResNet18ImageClassifier
 from DashAI.back.models.resnet50_image_classifier import ResNet50ImageClassifier
 from DashAI.back.models.scikit_learn.adaboost_classifier import AdaBoostClassifier
@@ -279,10 +334,11 @@
 
 # Plugins
 from DashAI.back.plugins.utils import get_available_plugins
-
-# Tasks
 from DashAI.back.tasks.controlnet_task import ControlNetTask
 from DashAI.back.tasks.image_classification_task import ImageClassificationTask
+
+# Tasks
+from DashAI.back.tasks.RAG_task import RAGTask
 from DashAI.back.tasks.regression_task import RegressionTask
 from DashAI.back.tasks.tabular_classification_task import TabularClassificationTask
 from DashAI.back.tasks.text_classification_task import TextClassificationTask
@@ -314,6 +370,7 @@ def get_initial_components():
         TextToImageGenerationTask,
         TextToTextGenerationTask,
         ControlNetTask,
+        RAGTask,
         ImageClassificationTask,
         # Models
         AdaBoostClassifier,
@@ -340,6 +397,8 @@ def get_initial_components():
         HistGradientBoostingClassifier,
         HistGradientBoostingRegression,
         KNeighborsClassifier,
+        QwenModel,
+        RAGPipeline,
         KNeighborsRegression,
         LassoRegression,
         LinearRegression,
@@ -379,6 +438,9 @@ def get_initial_components():
         StableDiffusionV3Model,
         StableDiffusionXLModel,
         StableDiffusionXLV1ControlNet,
+        OpenAITextToTextGenerationModel,
+        DeepSeekTextToTextGenerationModel,
+        Phi4MiniInstructModel,
         SVC,
         SVR,
         T5SmallTransformer,
@@ -433,6 +495,7 @@ def get_initial_components():
         DatasetJob,
         GenerativeJob,
         PipelineJob,
+        RAGJob,
         # Explainers
         KernelShap,
         PartialDependence,
@@ -495,6 +558,38 @@ def get_initial_components():
         SMOTEConverter,
         SMOTEENNConverter,
         RandomUnderSamplerConverter,
+        # RAG
+        RAGPipeline,
+        # Chunking Models
+        CharacterChunkModel,
+        RecursiveCharacterChunkModel,
+        TokenChunkModel,
+        # Encodings
+        OpenAIEmbedding,
+        HuggingFaceEmbedding,
+        SentenceTransformerEmbedding,
+        BERTEmbedding,
+        DistilBERTEmbedding,
+        RoBERTaEmbedding,
+        E5Embedding,
+        GemmaEmbedding,
+        InstructorEmbedding,
+        LaBSEmbedding,
+        # Prompts
+        DefaultRAGGenerationPrompt,
+        CustomRAGGenerationPrompt,
+        DefaultQARAGGenerationPrompt,
+        DefaultAugmentationPrompt,
+        CustomAugmentationPrompt,
+        # Retrievers
+        BM25Retriever,
+        BM25VectorizerModel,
+        TFIDFRetriever,
+        TFIDFVectorizerModel,
+        DenseEmbeddingRetriever,
+        MMRRerankerRetriever,
+        SequentialRetriever,
+        ParallelRetriever,
     ]
 
     # Obtener plugins instalados
diff --git a/DashAI/back/job/generative_job.py b/DashAI/back/job/generative_job.py
index e52c58a72..17af4d511 100644
--- a/DashAI/back/job/generative_job.py
+++ b/DashAI/back/job/generative_job.py
@@ -111,7 +111,23 @@ def run(
         component_registry = di["component_registry"]
         session_factory = di["session_factory"]
         config = di["config"]
-        # (Lazy imports removed to avoid duplicate and unused imports warnings)
+
+        from DashAI.back.job.rag_job import RAGJob
+        from DashAI.back.tasks.RAG_task import RAGTask
+
+        with session_factory() as db:
+            process = db.get(
+                GenerativeProcess, self.kwargs.get("generative_process_id")
+            )
+            if process is not None:
+                session = db.get(GenerativeSession, process.session_id)
+                if session is not None:
+                    task_class = component_registry[session.task_name]["class"]
+                    if issubclass(task_class, RAGTask):
+                        rag_job = RAGJob(**self.kwargs)
+                        rag_job.run()
+                        return
+
         model = None
         generative_process = None
         with session_factory() as db:
diff --git a/DashAI/back/job/rag_job.py b/DashAI/back/job/rag_job.py
new file mode 100644
index 000000000..01253b582
--- /dev/null
+++ b/DashAI/back/job/rag_job.py
@@ -0,0 +1,313 @@
+import logging
+from typing import TYPE_CHECKING, Any
+
+from kink import inject
+from sqlalchemy import exc
+
+from DashAI.back.core.enums.status import RunStatus
+from DashAI.back.dependencies.database.models import (
+    GenerativeProcess,
+    GenerativeSession,
+    ProcessData,
+)
+from DashAI.back.job.base_job import BaseJob, JobError
+from DashAI.back.models.RAG.document_loader import DocumentLoader
+from DashAI.back.models.RAG.pipeline_repository import PipelineRepository
+from DashAI.back.models.RAG.rag_models_factory import RAGModelsFactory
+from DashAI.back.models.RAG.RAG_pipeline import (
+    RAGPipeline,
+    RAGPipelineConfig,
+)
+from DashAI.back.tasks.RAG_task import RAGTask
+
+if TYPE_CHECKING:
+    from sqlalchemy.orm import sessionmaker
+
+log = logging.getLogger(__name__)
+
+# Known RAG pipeline parameter keys accepted from generative_session.parameters.
+_RAG_PARAM_KEYS: frozenset = frozenset(
+    {"documents", "prompt", "chunking_model", "retriever_model", "generation_model"}
+)
+
+
+class RAGJob(BaseJob):
+    """RAGJob handles the full RAG pipeline lifecycle as a background job."""
+
+    @inject
+    def set_status_as_delivered(
+        self, session_factory: "sessionmaker" = lambda di: di["session_factory"]
+    ) -> None:
+        """Set the status of the job as delivered."""
+        generative_process_id: int = self.kwargs["generative_process_id"]
+
+        with session_factory() as db:
+            process: GenerativeProcess = db.get(
+                GenerativeProcess, generative_process_id
+            )
+            if not process:
+                raise JobError(
+                    f"Generative process {generative_process_id} does not exist in DB."
+                )
+            try:
+                process.set_status_as_delivered()
+                db.commit()
+            except exc.SQLAlchemyError as e:
+                log.exception(e)
+                raise JobError("Internal database error") from e
+
+    @inject
+    def set_status_as_error(
+        self, session_factory: "sessionmaker" = lambda di: di["session_factory"]
+    ) -> None:
+        """Set the status of the job as error."""
+        generative_process_id: int = self.kwargs.get("generative_process_id")
+        if generative_process_id is None:
+            return
+
+        with session_factory() as db:
+            process: GenerativeProcess = db.get(
+                GenerativeProcess, generative_process_id
+            )
+            if not process:
+                return
+            try:
+                process.set_status_as_error()
+                db.commit()
+            except exc.SQLAlchemyError as e:
+                log.exception(e)
+
+    @inject
+    def get_job_name(self) -> str:
+        """Get a descriptive name for the job."""
+        generative_process_id = self.kwargs.get("generative_process_id")
+        if not generative_process_id:
+            return "RAG Process"
+
+        from kink import di
+
+        session_factory = di["session_factory"]
+
+        try:
+            with session_factory() as db:
+                process: GenerativeProcess = db.get(
+                    GenerativeProcess, generative_process_id
+                )
+                if process:
+                    session: GenerativeSession = db.get(
+                        GenerativeSession, process.session_id
+                    )
+                    if session and session.name:
+                        return f"RAG: {session.name}"
+                    return f"RAG Process #{generative_process_id}"
+        except Exception as e:
+            log.exception(f"Error getting job name: {e}")
+
+        return f"RAG Process #{generative_process_id}"
+
+    @inject
+    def run(self) -> None:
+        """Execute the full RAG pipeline lifecycle as a background job.
+
+        Loads the generative process and session, filters parameters via
+        whitelist, instantiates the RAG pipeline, runs inference, and
+        persists the output.
+
+        Raises:
+            JobError: If any stage of execution fails.
+        """
+        import gc
+
+        import torch
+        from kink import di
+
+        component_registry = di["component_registry"]
+        session_factory = di["session_factory"]
+        config = di["config"]
+
+        model = None
+        generative_process = None
+        # NOTE: The DB session spans the entire job lifecycle including LLM
+        # inference, which risks connection timeouts for long-running
+        # generations. No obvious solution yet.
+        with session_factory() as db:
+            try:
+                generative_process_id: int = self.kwargs["generative_process_id"]
+
+                try:
+                    generative_process: GenerativeProcess = db.get(
+                        GenerativeProcess, generative_process_id
+                    )
+                    if not generative_process:
+                        raise JobError(
+                            f"Generative process {generative_process_id} "
+                            "not found in DB."
+                        )
+                    log.debug("Loaded generative process %d", generative_process_id)
+                except Exception as e:
+                    log.exception(e)
+                    if generative_process:
+                        generative_process.set_status_as_error()
+                        db.commit()
+                    raise JobError("Error retrieving generative process.") from e
+
+                try:
+                    generative_session: GenerativeSession = db.get(
+                        GenerativeSession, generative_process.session_id
+                    )
+                    if not generative_session:
+                        raise JobError(
+                            f"Session {generative_process.session_id} not found in DB."
+                        )
+                    log.debug(
+                        "Loaded generative session %d", generative_session.id
+                    )
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError("Error retrieving generative session.") from e
+
+                try:
+                    # Whitelist-only: accept only known RAG pipeline keys.
+                    # generative_session.parameters is a JSON column that may
+                    # contain legacy or metadata keys (e.g. prompt_id).
+                    raw_params = dict(generative_session.parameters)
+                    extra_keys: set[str] = set(raw_params) - _RAG_PARAM_KEYS
+                    if extra_keys:
+                        log.debug(
+                            "Filtered extra session parameter keys: %s",
+                            sorted(extra_keys),
+                        )
+                    clean_params = {}
+                    for key in raw_params:
+                        if key in _RAG_PARAM_KEYS:
+                            clean_params[key] = raw_params[key]
+                    pipeline_config = RAGPipelineConfig.from_kwargs(
+                        db=db,
+                        component_registry=component_registry,
+                        session_id=generative_session.id,
+                        env_rag_path=config["RAG_PATH"],
+                        **clean_params,
+                    )
+                    log.debug("Created RAG pipeline config for session %d", generative_session.id)
+                    models = RAGModelsFactory(
+                        db,
+                        component_registry,
+                        config["RAG_PATH"],
+                    )
+                    repo = PipelineRepository(db)
+                    doc_loader = DocumentLoader(db)
+                    model: RAGPipeline = RAGPipeline(
+                        pipeline_config,
+                        models,
+                        repo,
+                        doc_loader,
+                    )
+                    log.debug("Created RAG pipeline model for session %d", generative_session.id)
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError(
+                        "Error instantiating RAG pipeline with given parameters."
+                    ) from e
+
+                input_data = generative_process.input
+
+                try:
+                    task = RAGTask()
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError("Error instantiating RAG task.") from e
+
+                try:
+                    history = [
+                        (proc.input[0].data, proc.output[0].data)
+                        for proc in db.query(GenerativeProcess)
+                        .filter(GenerativeProcess.session_id == generative_session.id)
+                        .filter(GenerativeProcess.status == RunStatus.FINISHED)
+                        .all()
+                    ]
+                    input_data = task.prepare_for_task(
+                        input_data,
+                        history=history,
+                    )
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError("Error preparing task with history.") from e
+
+                try:
+                    log.debug("Starting generation for process %d", generative_process_id)
+                    generative_process.set_status_as_started()
+                    db.commit()
+                except exc.SQLAlchemyError as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError(
+                        "Failed to update process status in database."
+                    ) from e
+
+                try:
+                    output: Any = model.generate(input_data)
+                    log.debug("Generation completed for process %d", generative_process_id)
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.add(
+                        ProcessData(
+                            data=f"Error details: {str(e)}",
+                            data_type="str",
+                            process_id=generative_process.id,
+                            is_input=False,
+                        )
+                    )
+                    db.commit()
+                    raise JobError("Error during RAG model generation.") from e
+
+                try:
+                    output = task.process_output(
+                        output, images_path=config["IMAGES_PATH"]
+                    )
+                    outputs_for_database = []
+                    for o in output:
+                        if not isinstance(o, tuple) or len(o) != 2:
+                            raise JobError(
+                                "Output from task must be a list of "
+                                "tuples (data, type)."
+                            )
+                        output_data, output_type = o
+                        process_data = ProcessData(
+                            data=output_data,
+                            data_type=output_type,
+                            process_id=generative_process.id,
+                            is_input=False,
+                        )
+                        outputs_for_database.append(process_data)
+
+                    db.add_all(outputs_for_database)
+                    db.commit()
+
+                    db.refresh(generative_process)
+                    generative_process.set_status_as_finished()
+                    db.commit()
+                    log.debug("Output processing completed for process %d", generative_process_id)
+                except Exception as e:
+                    log.exception(e)
+                    generative_process.set_status_as_error()
+                    db.commit()
+                    raise JobError(
+                        "Error processing and saving RAG generation output."
+                    ) from e
+
+            finally:
+                if model:
+                    del model
+                if torch.cuda.is_available():
+                    torch.cuda.empty_cache()
+                gc.collect()
diff --git a/DashAI/back/models/RAG/RAG_pipeline.py b/DashAI/back/models/RAG/RAG_pipeline.py
new file mode 100644
index 000000000..a61cd9979
--- /dev/null
+++ b/DashAI/back/models/RAG/RAG_pipeline.py
@@ -0,0 +1,502 @@
+"""RAG pipeline orchestration.
+
+RAGPipeline receives its dependencies injected (config, factories,
+repository, loader) rather than constructing them from raw kwargs.
+Orchestrates: document loading → chunk-set creation → chunking →
+retrieval → prompt formatting → LLM generation.
+"""
+
+from __future__ import annotations
+
+import logging
+from dataclasses import dataclass
+from typing import TYPE_CHECKING, Any, Dict, List, Tuple
+
+log = logging.getLogger(__name__)
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    component_field,
+    int_field,
+    list_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.dependencies.database.models import (
+    RAGChunkSet,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGPipeline as PipelineDBModel,
+)
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.base_generative_model import BaseGenerativeModel
+from DashAI.back.models.RAG.chunk_set_utils import get_or_create_chunk_set
+from DashAI.back.models.RAG.chunking_models.chunking_model_factory import (
+    ChunkingFactoryResult,
+)
+from DashAI.back.models.RAG.document_loader import DocumentLoader
+from DashAI.back.models.RAG.documents import BaseDocument, Chunk
+from DashAI.back.models.RAG.llm_factory import LLMFactoryResult
+from DashAI.back.models.RAG.pipeline_repository import PipelineRepository
+from DashAI.back.models.RAG.prompts.prompt_factory import PromptFactoryResult
+from DashAI.back.models.RAG.rag_models_factory import RAGModelsFactory
+from DashAI.back.models.RAG.retrievers.retriever_factory import (
+    RetrieverFactoryResult,
+)
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+
+if TYPE_CHECKING:
+    from DashAI.back.models.RAG.chunking_models.base_chunking_model import (
+        BaseChunkingModel,
+    )
+    from DashAI.back.models.RAG.prompts import Prompt
+    from DashAI.back.models.RAG.retrievers.retriever_model import RetrieverModel
+
+
+class RAGPipelineError(Exception):
+    """Base exception for RAG pipeline errors."""
+
+
+class RAGPipelineParametersError(RAGPipelineError):
+    """Backward-compatible alias for RAGPipelineConfigError."""
+
+
+class RAGPipelineConfigError(RAGPipelineError):
+    """Invalid or missing parameters in pipeline configuration."""
+
+
+class RAGPipelineInitializationError(RAGPipelineError):
+    """Error during RAG pipeline initialization."""
+
+
+class RAGPipelineRuntimeError(RAGPipelineError):
+    """Error during RAG pipeline execution."""
+
+
+class RAGDatabaseError(RAGPipelineError):
+    """Database-related error in RAG pipeline."""
+
+
+@dataclass(frozen=True)
+class ModelRef:
+    """Parsed reference to a component model.
+
+    Represents a ``{component: str, params: dict}`` entry from the
+    pipeline parameter payload.
+    """
+
+    component: str
+    params: Dict[str, Any]
+
+
+@dataclass(frozen=True)
+class ChunkReference:
+    """A single chunk with its document metadata."""
+
+    document_id: int
+    document_name: str
+    document_position: int
+    text: str
+
+    def to_dict(self) -> Dict[str, Any]:
+        """Return a JSON-serializable dict representation."""
+        return {
+            "document_id": self.document_id,
+            "document_name": self.document_name,
+            "document_position": self.document_position,
+            "text": self.text,
+        }
+
+
+@dataclass(frozen=True)
+class RAGGenerationOutput:
+    """Typed output from RAGPipeline.generate()."""
+
+    message: Any
+    chunks: Dict[str, "ChunkReference"]
+
+
+@dataclass(frozen=True)
+class RAGPipelineConfig:
+    """Structured, validated representation of pipeline initialisation kwargs.
+
+    Parses the raw kwargs dict once and provides typed field access,
+    eliminating magic strings in the pipeline's __init__.
+    """
+
+    session_id: int
+    db: Session
+    component_registry: ComponentRegistry
+    env_rag_path: str
+    documents: List[int]
+    prompt: ModelRef
+    chunking_model: ModelRef
+    retriever_model: ModelRef
+    generation_model: ModelRef
+
+    _MODEL_KEYS: Tuple[str, str, str, str] = (
+        "prompt",
+        "chunking_model",
+        "retriever_model",
+        "generation_model",
+    )
+    _INFRA_KEYS: Tuple[str, str, str, str] = (
+        "session_id",
+        "db",
+        "component_registry",
+        "env_rag_path",
+    )
+    _PARAM_KEYS: Tuple[str] = ("documents",)
+
+    @classmethod
+    def validate_model_refs(cls, params: Dict[str, Any]) -> None:
+        for key in cls._MODEL_KEYS:
+            if key not in params:
+                raise RAGPipelineConfigError(f"Missing '{key}'")
+            raw = params[key]
+            if not isinstance(raw, dict):
+                raise RAGPipelineConfigError(
+                    f"'{key}' must be a dict, got {type(raw).__name__}"
+                )
+            if "component" not in raw:
+                raise RAGPipelineConfigError(f"Missing 'component' in '{key}'")
+            if "params" not in raw:
+                raise RAGPipelineConfigError(f"Missing 'params' in '{key}'")
+        if "documents" not in params:
+            raise RAGPipelineConfigError("Missing required parameter 'documents'")
+
+    @classmethod
+    def from_kwargs(cls, **kwargs: Any) -> "RAGPipelineConfig":
+        missing: List[str] = []
+        for key in cls._INFRA_KEYS:
+            if key not in kwargs:
+                missing.append(key)
+        for key in cls._PARAM_KEYS:
+            if key not in kwargs:
+                missing.append(key)
+        for key in cls._MODEL_KEYS:
+            if key not in kwargs:
+                missing.append(key)
+        if missing:
+            raise RAGPipelineConfigError(
+                f"Missing required parameters: {sorted(missing)}"
+            )
+
+        model_refs: Dict[str, ModelRef] = {}
+        for key in cls._MODEL_KEYS:
+            raw = kwargs[key]
+            if not isinstance(raw, dict):
+                raise RAGPipelineConfigError(
+                    f"'{key}' must be a dict, got {type(raw).__name__}"
+                )
+            if "component" not in raw:
+                raise RAGPipelineConfigError(f"Missing 'component' in '{key}'")
+            if "params" not in raw:
+                raise RAGPipelineConfigError(f"Missing 'params' in '{key}'")
+            model_refs[key] = ModelRef(
+                component=raw["component"],
+                params=raw["params"],
+            )
+
+        all_known: set[str] = (
+            set(cls._INFRA_KEYS) | set(cls._PARAM_KEYS) | set(cls._MODEL_KEYS)
+        )
+        extra: set[str] = set(kwargs) - all_known
+        if extra:
+            log.warning(
+                "Unknown parameters passed to RAGPipelineConfig.from_kwargs(): %s. "
+                "These keys are not recognized and will cause the pipeline to fail. "
+                "Check if session parameters contain extra metadata keys.",
+                sorted(extra),
+            )
+            raise RAGPipelineConfigError(
+                f"Unknown parameters: {sorted(extra)}. "
+                "These keys are not recognized pipeline configuration parameters."
+            )
+
+        return cls(
+            session_id=kwargs["session_id"],
+            db=kwargs["db"],
+            component_registry=kwargs["component_registry"],
+            env_rag_path=kwargs["env_rag_path"],
+            documents=kwargs["documents"],
+            prompt=model_refs["prompt"],
+            chunking_model=model_refs["chunking_model"],
+            retriever_model=model_refs["retriever_model"],
+            generation_model=model_refs["generation_model"],
+        )
+
+
+class RAGPipelineSchema(BaseSchema):
+    documents: schema_field(
+        list_field(int_field(gt=0)),
+        placeholder=None,
+        description=MultilingualString(
+            en="List of document IDs to use in the RAG pipeline.",
+            es="Lista de IDs de documentos a usar en el pipeline RAG.",
+        ),
+    )  # type: ignore
+
+    prompt: schema_field(
+        component_field(parent="Prompt"),
+        placeholder={"component": "DefaultRAGGenerationPrompt", "params": {}},
+        description=MultilingualString(
+            en="Prompt template used in the RAG pipeline.",
+            es="Plantilla de prompt usada en el pipeline RAG.",
+        ),
+    )  # type: ignore
+
+    chunking_model: schema_field(
+        component_field(parent="BaseChunkingModel"),
+        description=MultilingualString(
+            en="Chunking model used to split documents into smaller pieces.",
+            es="Modelo de fragmentación para dividir documentos en piezas.",
+        ),
+        placeholder={"component": "CharacterChunkModel", "params": {}},
+    )  # type: ignore
+
+    retriever_model: schema_field(
+        component_field(parent="RetrieverModel"),
+        placeholder={"component": "TFIDFRetriever", "params": {}},
+        description=MultilingualString(
+            en="Retriever component used in the RAG pipeline.",
+            es="Componente recuperador usado en el pipeline RAG.",
+        ),
+    )  # type: ignore
+
+    generation_model: schema_field(
+        component_field(parent="TextToTextGenerationTaskModel"),
+        placeholder={"component": "", "params": {}},
+        description=MultilingualString(
+            en="Text generation model used in the RAG pipeline.",
+            es="Modelo de generación de texto usado en el pipeline RAG.",
+        ),
+    )  # type: ignore
+
+
+class RAGPipeline(BaseGenerativeModel):
+    """Retrieval-Augmented Generation pipeline.
+
+    Receives dependencies injected — does not construct factories,
+    repositories, or loaders. The caller (RAGJob) builds them from
+    the current DB session and passes them in.
+
+    Orchestrates: document loading → chunk-set creation → chunking →
+    retrieval → prompt formatting → LLM generation.
+    """
+
+    COMPATIBLE_COMPONENTS: List[str] = ["RAGTask"]
+    SCHEMA: type[BaseSchema] = RAGPipelineSchema
+
+    session_id: int
+    pipeline_id: int
+    documents_ids: List[int]
+    documents: Dict[int, BaseDocument]
+    prompt_model: Prompt
+    chunking_model_id: int
+    chunking_model: BaseChunkingModel
+    chunks: Dict[int, Dict[int, Chunk]]
+    retriever: RetrieverModel
+    llm_model: TextToTextGenerationTaskModel
+
+    DISPLAY_NAME: str = MultilingualString(
+        en="RAG Pipeline",
+        es="Flujo de RAG",
+        pt="Pipeline RAG",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Pipeline for Retrieval-Augmented Generation (RAG) tasks, orchestrating document loading, chunking, retrieval, prompt formatting, and LLM generation.",
+        es="Pipeline para tareas de Generación Aumentada por Recuperación (RAG), orquestando la carga de documentos, chunking, recuperación, formateo de prompts y generación con LLM.",
+        pt="Pipeline para tarefas de Geração Aumentada por Recuperação (RAG), orquestrando carregamento de documentos, chunking, recuperação, formatação de prompts e geração com LLM.",
+    )
+    COLOR: str = "#e12885"
+    ICON: str = "Grading"
+    MODEL_NAME: str = "RAGPipeline"
+
+    def __init__(
+        self,
+        config: RAGPipelineConfig,
+        models: RAGModelsFactory,
+        repo: PipelineRepository,
+        doc_loader: DocumentLoader,
+    ) -> None:
+        pipeline_record: PipelineDBModel = repo.ensure_db_record(
+            config.session_id,
+        )
+
+        documents = doc_loader.load(config.documents)
+
+        chunk_set: RAGChunkSet = get_or_create_chunk_set(
+            db=config.db,
+            document_ids=config.documents,
+            pipeline_config={
+                "chunking_model": {
+                    "component": config.chunking_model.component,
+                    "params": config.chunking_model.params,
+                }
+            },
+        )
+
+        prompt_result: PromptFactoryResult = models.create_prompt(
+            config.prompt.component,
+            config.prompt.params,
+        )
+
+        chunking_result: ChunkingFactoryResult = models.create_chunking_model(
+            documents,
+            chunk_set.id,
+            config.chunking_model.component,
+            config.chunking_model.params,
+        )
+
+        retriever_result: RetrieverFactoryResult = models.create_retriever(
+            pipeline_record.id,
+            chunking_result.chunks,
+            chunk_set.id,
+            config.retriever_model.component,
+            config.retriever_model.params,
+        )
+
+        llm_result: LLMFactoryResult = models.create_llm(
+            config.generation_model.component,
+            config.generation_model.params,
+        )
+
+        repo.update_db_record(
+            session_id=config.session_id,
+            chunking_model_id=chunking_result.db_record_id,
+            prompt_id=prompt_result.db_record_id,
+            generation_model_id=llm_result.db_record_id,
+        )
+
+        self.session_id = config.session_id
+        self.pipeline_id = pipeline_record.id
+        self.documents_ids = config.documents
+        self.documents = documents
+        self.prompt_model = prompt_result.model
+        self.chunking_model_id = chunking_result.db_record_id
+        self.chunking_model = chunking_result.model
+        self.chunks = chunking_result.chunks
+        self.retriever = retriever_result.model
+        self.llm_model = llm_result.model
+
+    def single_interaction(
+        self,
+        query: str,
+    ) -> List[Chunk]:
+        """Retrieve the top-K chunks for a single query.
+
+        Parameters
+        ----------
+        query : str
+            The search query string.
+
+        Returns
+        -------
+        List[Chunk]
+            Ranked list of retrieved chunks.
+
+        Raises
+        ------
+        RAGPipelineRuntimeError
+            If the retriever fails.
+        """
+        try:
+            return self.retriever.retrieve(query)
+        except Exception as e:
+            raise RAGPipelineRuntimeError(f"Document retrieval failed: {str(e)}") from e
+
+    def _build_chunk_references(
+        self,
+        chunks: List[Chunk],
+    ) -> Tuple[str, Dict[str, ChunkReference]]:
+        """Build a text block and reference map from retrieved chunks.
+
+        Parameters
+        ----------
+        chunks : List[Chunk]
+            Chunks returned by the retriever.
+
+        Returns
+        -------
+        Tuple[str, Dict[str, ChunkReference]]
+            Joined chunk texts and a mapping from chunk key to ChunkReference.
+        """
+        chunks_texts: List[str] = []
+        chunk_dict: Dict[str, ChunkReference] = {}
+        for retrieved_chunk in chunks:
+            document_id: int = retrieved_chunk.document_id
+            document: BaseDocument = self.documents[document_id]
+            chunk_position: int = retrieved_chunk.document_position
+            chunk_text: str = retrieved_chunk.text
+            chunk_ref: str = f"{document_id}_{chunk_position}"
+            chunk_dict[chunk_ref] = ChunkReference(
+                document_id=document_id,
+                document_name=document.file_name,
+                document_position=chunk_position,
+                text=chunk_text,
+            )
+            chunks_texts.append(
+                f"Document {document.file_name}, "
+                f"chunk nº {chunk_position}, text:\n {chunk_text}"
+            )
+        return "\n\n".join(chunks_texts), chunk_dict
+
+    def generate(
+        self,
+        input_data: Tuple[Dict[str, str], ...],
+    ) -> RAGGenerationOutput:
+        """Run the full RAG pipeline: retrieve → format → generate.
+
+        Parameters
+        ----------
+        input_data : Tuple[Dict[str, str], ...]
+            Chat-format input with history as earlier entries and the
+            current user message as the last entry.
+
+        Returns
+        -------
+        RAGGenerationOutput
+            The generated message and the chunks used for retrieval.
+
+        Raises
+        ------
+        RAGPipelineRuntimeError
+            If any stage fails (retrieval, formatting, or generation).
+        """
+        if not input_data:
+            raise RAGPipelineRuntimeError("input_data must not be empty.")
+        try:
+            input_dict: Dict[str, str] = input_data[-1]
+            input_message: str = input_dict["content"]
+            history: Tuple[Dict[str, str], ...] = input_data[:-1]
+            chunks: List[Chunk] = self.single_interaction(input_message)
+        except Exception as e:
+            raise RAGPipelineRuntimeError(f"Failed during retrieval: {str(e)}") from e
+        try:
+            chunks_text: str
+            chunk_dict: Dict[str, ChunkReference]
+            chunks_text, chunk_dict = self._build_chunk_references(chunks)
+            prompt: str = self.prompt_model.format(
+                input=input_message,
+                chunks=chunks_text,
+            )
+        except Exception as e:
+            raise RAGPipelineRuntimeError(
+                f"Failed during prompt formatting: {str(e)}"
+            ) from e
+        try:
+            model_input: List[Dict[str, str]] = list(history) + [
+                {"role": "user", "content": prompt}
+            ]
+            # NOTE: Output is not streamed — the user waits for the full response.
+            output: List[Any] = self.llm_model.generate(model_input)
+            return RAGGenerationOutput(message=output[0], chunks=chunk_dict)
+        except Exception as e:
+            raise RAGPipelineRuntimeError(
+                f"Failed during LLM generation: {str(e)}"
+            ) from e
diff --git a/DashAI/back/models/RAG/__init__.py b/DashAI/back/models/RAG/__init__.py
new file mode 100644
index 000000000..bbfd5ce7a
--- /dev/null
+++ b/DashAI/back/models/RAG/__init__.py
@@ -0,0 +1 @@
+from DashAI.back.models.RAG.RAG_pipeline import RAGPipeline  # noqa: F401
diff --git a/DashAI/back/models/RAG/chunk_set_utils.py b/DashAI/back/models/RAG/chunk_set_utils.py
new file mode 100644
index 000000000..7daebf40b
--- /dev/null
+++ b/DashAI/back/models/RAG/chunk_set_utils.py
@@ -0,0 +1,57 @@
+import hashlib
+import json
+from typing import Any, Dict, List
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import (
+    RAGChunkSet,
+    RAGChunkSetDocument,
+)
+
+
+def _build_chunk_set_signature(
+    document_ids: List[int],
+    pipeline_config: Dict[str, Any],
+) -> str:
+    payload = json.dumps(
+        {
+            "doc_ids": sorted(document_ids),
+            "config": dict(sorted(pipeline_config.items())),
+        },
+        sort_keys=True,
+    )
+    return hashlib.sha256(payload.encode()).hexdigest()
+
+
+# NOTE: This does a SELECT then INSERT without a lock, which is safe for
+# a single-user platform but could cause IntegrityError under concurrent
+# session creation with identical document sets.
+def get_or_create_chunk_set(
+    db: Session,
+    document_ids: List[int],
+    pipeline_config: Dict[str, Any],
+) -> RAGChunkSet:
+    signature = _build_chunk_set_signature(document_ids, pipeline_config)
+    existing = db.query(RAGChunkSet).filter_by(signature=signature).first()
+    if existing:
+        return existing
+
+    chunk_set = RAGChunkSet(
+        signature=signature,
+        parameters=pipeline_config,
+    )
+    db.add(chunk_set)
+    db.commit()
+    db.refresh(chunk_set)
+
+    for doc_id in sorted(document_ids):
+        db.add(
+            RAGChunkSetDocument(
+                chunk_set_id=chunk_set.id,
+                document_id=doc_id,
+            )
+        )
+
+    db.commit()
+    return chunk_set
diff --git a/DashAI/back/models/RAG/chunking_models/__init__.py b/DashAI/back/models/RAG/chunking_models/__init__.py
new file mode 100644
index 000000000..49eda0a1f
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/__init__.py
@@ -0,0 +1,6 @@
+from DashAI.back.models.RAG.chunking_models.base_chunking_model import BaseChunkingModel
+from DashAI.back.models.RAG.chunking_models.character_chunk_model import CharacterChunkModel
+from DashAI.back.models.RAG.chunking_models.recursive_character_chunk_model import (
+    RecursiveCharacterChunkModel,
+)
+from DashAI.back.models.RAG.chunking_models.token_chunk_model import TokenChunkModel
\ No newline at end of file
diff --git a/DashAI/back/models/RAG/chunking_models/base_chunking_model.py b/DashAI/back/models/RAG/chunking_models/base_chunking_model.py
new file mode 100644
index 000000000..45d496f96
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/base_chunking_model.py
@@ -0,0 +1,113 @@
+from abc import ABCMeta, abstractmethod
+from typing import Dict, Final, List
+
+from DashAI.back.config_object import ConfigObject
+from DashAI.back.models.RAG.documents.base_document import BaseDocument
+from DashAI.back.models.RAG.documents.chunk import Chunk
+from DashAI.back.models.RAG.exceptions import RAGWorkflowError
+
+
+class BaseChunkingModel(ConfigObject, metaclass=ABCMeta):
+    """
+    Base class for chunking models.
+    This class should be inherited by any specific chunking model implementation.
+    """
+
+    TYPE: Final[str] = "ChunkingModel"
+    REQUIRED_EXTRA_KWARGS: Final[List[str]] = ["documents"]
+
+    def __init__(self, **kwargs):
+        """
+        Initialize the chunking model with any necessary parameters.
+        """
+        self.documents: Dict[int, BaseDocument] = kwargs.pop("documents")
+        self.parameters = self.validate_and_transform(kwargs)
+        self.chunks = self.chunk_documents()
+        self.id = None
+
+    @abstractmethod
+    def chunk_text(self, text: str, **kwargs) -> List[str]:
+        """
+        Chunk the input text into smaller pieces.
+
+        Args:
+            text (str): The input text to be chunked.
+            **kwargs: Additional parameters for chunking.
+
+        Returns:
+            List[str]: A list of text chunks.
+        """
+        raise NotImplementedError("Subclasses must implement the chunk_text method.")
+
+    def chunk_document(self, document: BaseDocument) -> Dict[int, Chunk]:
+        """
+        Chunk a single document using the chunk_text method.
+
+        Args:
+            document (BaseDocument): The input document to be chunked.
+            **kwargs: Additional parameters for chunking, passed to `chunk_text`.
+
+        Returns:
+            Dict[int, Chunk]: A dictionary mapping chunk indices to Chunk objects.
+        """
+        if not isinstance(document, BaseDocument):
+            raise TypeError("Input must be an instance of BaseDocument.")
+        text = document.get_text()
+        chunks_text = self.chunk_text(text)
+        chunks = {}
+        for idx, chunk_text in enumerate(chunks_text):
+            chunk = Chunk(
+                id=None,
+                document_position=idx,
+                document_id=document.id,
+                text=chunk_text,
+            )
+            chunks[idx] = chunk
+        return chunks
+
+    def chunk_documents(self) -> Dict[int, Dict[int, Chunk]]:
+        """
+        Chunk documents using the chunk_document method.
+
+        Returns:
+            Dict[int, Dict[int, Chunk]]: A dictionary mapping document IDs to their respective
+            dictionaries of chunk indices and Chunk objects.
+        """
+        chunked_documents = {}
+        for document_id, document in self.documents.items():
+            chunks = self.chunk_document(document)
+            chunked_documents[document_id] = chunks
+        return chunked_documents
+
+    def get_chunks(self) -> Dict[int, Dict[int, Chunk]]:
+        """
+        Get the chunks generated by the chunking model.
+
+        Returns:
+            Dict[int, Dict[int, Chunk]]: A dictionary mapping document IDs to their respective
+            dictionaries of chunk indices and Chunk objects.
+        """
+        return self.chunks
+
+    def get_id(self) -> int:
+        """
+        Get the ID of the chunking model.
+
+        Returns:
+            int: The ID of the chunking model.
+        """
+        return self.id
+
+    def set_id(self, id: int) -> None:
+        """
+        Set the ID of the chunking model.
+
+        Args:
+            id (int): The ID to set for the chunking model.
+        """
+        if self.id is None:
+            self.id = id
+        else:
+            raise RAGWorkflowError(
+                "Chunking model ID is already set and cannot be modified."
+            )
diff --git a/DashAI/back/models/RAG/chunking_models/character_chunk_model.py b/DashAI/back/models/RAG/chunking_models/character_chunk_model.py
new file mode 100644
index 000000000..dd04002a3
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/character_chunk_model.py
@@ -0,0 +1,89 @@
+from typing import List
+
+from pydantic import field_validator
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    int_field,
+    schema_field,
+)
+from DashAI.back.models.RAG.chunking_models.base_chunking_model import (
+    BaseChunkingModel,
+)
+
+
+class CharacterChunkModelSchema(BaseSchema):
+    """Schema for character-based chunking model."""
+
+    chunk_size: schema_field(
+        int_field(gt=1),
+        placeholder=400,
+        description="Size of each chunk in characters.",
+    )  # type: ignore
+
+    chunk_overlap: schema_field(
+        int_field(gt=0),
+        placeholder=40,
+        description=(
+            "Number of characters to overlap between chunks. "
+            "Must be less than chunk_size."
+        ),
+    )  # type: ignore
+
+    @field_validator("chunk_overlap", mode="after")
+    @classmethod
+    def validate_chunk_overlap(cls, v, info):
+        """Validate that chunk_overlap is less than chunk_size."""
+        chunk_size = info.data.get("chunk_size")
+        if chunk_size is not None and v >= chunk_size:
+            raise ValueError(
+                f"chunk_overlap must be less than chunk_size. "
+                f"Got chunk_overlap={v} and chunk_size={chunk_size}"
+            )
+        return v
+
+
+class CharacterChunkModel(BaseChunkingModel):
+    """
+    Chunking model that splits documents into chunks based on character count.
+    This model is useful for processing large text documents into manageable pieces.
+    """
+
+    SCHEMA = CharacterChunkModelSchema
+
+    def __init__(self, **kwargs):
+        """
+        Initialize the character chunking model with the specified chunk size and
+        overlap.
+
+        Args:
+            chunk_size (int): Size of each chunk in characters.
+            chunk_overlap (int): Number of characters to overlap between chunks.
+        """
+        self.chunk_size = kwargs["chunk_size"]
+        self.chunk_overlap = kwargs["chunk_overlap"]
+        super().__init__(**kwargs)
+
+    def chunk_text(self, text: str) -> List[str]:
+        """
+        Chunk the input text into smaller segments based on character count.
+
+        Args:
+            text (str): The input text to be chunked.
+
+        Returns:
+            List[str]: A list of text chunks.
+        """
+        chunks = []
+        start = 0
+        text_length = len(text)
+
+        while start < text_length:
+            end = min(start + self.chunk_size, text_length)
+            chunk = text[start:end]
+            chunks.append(chunk)
+            if end == text_length:
+                break
+            start += self.chunk_size - self.chunk_overlap
+
+        return chunks
diff --git a/DashAI/back/models/RAG/chunking_models/chunking_model_factory.py b/DashAI/back/models/RAG/chunking_models/chunking_model_factory.py
new file mode 100644
index 000000000..dc6c367c9
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/chunking_model_factory.py
@@ -0,0 +1,177 @@
+"""Factory for chunking models with full lifecycle encapsulation.
+
+Single-call interface: create() handles DB-record resolution,
+model instantiation, document chunking, and chunk persistence.
+"""
+
+from dataclasses import dataclass
+from typing import Any, Dict
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import (
+    Chunk as ChunkDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGChunkingModel as ChunkingModelDBModel,
+)
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.RAG.chunking_models.base_chunking_model import (
+    BaseChunkingModel,
+)
+from DashAI.back.models.RAG.documents.base_document import BaseDocument
+from DashAI.back.models.RAG.documents.chunk import Chunk
+
+
+@dataclass(frozen=True)
+class ChunkingFactoryResult:
+    """Result of chunking model creation via ChunkingModelFactory."""
+
+    db_record_id: int
+    model: BaseChunkingModel
+    chunks: Dict[int, Dict[int, Chunk]]
+
+
+class ChunkingModelFactory:
+    """Creates a chunking model, chunks documents, and persists everything.
+
+    Encapsulates the full lifecycle: DB-record resolution, model
+    instantiation, document splitting, and chunk persistence.
+    The caller calls ``create()`` once — no post-call required.
+    """
+
+    def __init__(
+        self,
+        db: Session,
+        registry: ComponentRegistry,
+        documents: Dict[int, BaseDocument],
+        chunk_set_id: int,
+    ):
+        self._db = db
+        self._registry = registry
+        self._documents = documents
+        self._chunk_set_id = chunk_set_id
+
+    def create(
+        self,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> ChunkingFactoryResult:
+        """Full lifecycle: resolve DB record → instantiate → chunk → persist.
+
+        Returns:
+            ChunkingFactoryResult with record_id, model, and populated chunks.
+        """
+        sorted_params: Dict[str, Any] = dict(sorted(params.items()))
+
+        existing_record = self._resolve_db_record(component_name, sorted_params)
+        if existing_record is not None:
+            params["documents"] = self._documents
+            model_class = self._registry[component_name]["class"]
+            model = model_class(**params)
+            model.set_id(existing_record.id)
+            self._update_chunk_ids(model)
+            return ChunkingFactoryResult(
+                db_record_id=existing_record.id,
+                model=model,
+                chunks=model.get_chunks(),
+            )
+
+        model_class = self._registry[component_name]["class"]
+        params["documents"] = self._documents
+        model = model_class(**params)
+        record_id = self._save_db_record(model)
+
+        self._persist_chunks(model)
+
+        return ChunkingFactoryResult(
+            db_record_id=record_id,
+            model=model,
+            chunks=model.get_chunks(),
+        )
+
+    def _resolve_db_record(
+        self, class_name: str, sorted_params: Dict[str, Any]
+    ) -> ChunkingModelDBModel | None:
+        return (
+            self._db.query(ChunkingModelDBModel)
+            .filter_by(class_name=class_name, parameters=sorted_params)
+            .first()
+        )
+
+    def _save_db_record(self, model: BaseChunkingModel) -> int:
+        sorted_params = dict(sorted(model.parameters.items()))
+        record = ChunkingModelDBModel(
+            class_name=model.__class__.__name__,
+            parameters=sorted_params,
+        )
+        self._db.add(record)
+        self._db.commit()
+        self._db.refresh(record)
+        model.set_id(record.id)
+        return record.id
+
+    def _fetch_chunks_from_db(self) -> Dict[int, Dict[int, Chunk]]:
+        chunks: Dict[int, Dict[int, Chunk]] = {}
+        db_chunks = (
+            self._db.query(ChunkDBModel)
+            .filter_by(chunk_set_id=self._chunk_set_id)
+            .all()
+        )
+        for db_chunk in db_chunks:
+            if db_chunk.document_id not in chunks:
+                chunks[db_chunk.document_id] = {}
+            chunk = Chunk(
+                id=db_chunk.id,
+                document_id=db_chunk.document_id,
+                document_position=db_chunk.chunk_index,
+                text=db_chunk.text,
+            )
+            chunks[db_chunk.document_id][db_chunk.chunk_index] = chunk
+        return chunks
+
+    def _create_chunks_in_db(self, chunks: Dict[int, Dict[int, Chunk]]) -> None:
+        for document_id, document_chunks in chunks.items():
+            for idx, chunk in document_chunks.items():
+                self._db.add(
+                    ChunkDBModel(
+                        document_id=document_id,
+                        chunk_index=idx,
+                        chunk_set_id=self._chunk_set_id,
+                        text=chunk.text,
+                    )
+                )
+        self._db.commit()
+        self._db.flush()
+
+    def _update_chunk_ids(self, model: BaseChunkingModel) -> None:
+        for document_id, document_chunks in model.get_chunks().items():
+            for idx, chunk in document_chunks.items():
+                db_chunk = (
+                    self._db.query(ChunkDBModel)
+                    .filter_by(
+                        document_id=document_id,
+                        chunk_index=idx,
+                        chunk_set_id=self._chunk_set_id,
+                    )
+                    .first()
+                )
+                if db_chunk is not None:
+                    model.chunks[document_id][idx].id = db_chunk.id
+
+    def _persist_chunks(self, model: BaseChunkingModel) -> None:
+        existing_chunks = self._fetch_chunks_from_db()
+        chunks_to_create: Dict[int, Dict[int, Chunk]] = {}
+        for document_id, document_chunks in model.get_chunks().items():
+            if document_id not in existing_chunks:
+                chunks_to_create[document_id] = document_chunks
+            else:
+                for idx, chunk in document_chunks.items():
+                    if idx not in existing_chunks[document_id]:
+                        if document_id not in chunks_to_create:
+                            chunks_to_create[document_id] = {}
+                        chunks_to_create[document_id][idx] = chunk
+
+        self._create_chunks_in_db(chunks_to_create)
+
+        self._update_chunk_ids(model)
diff --git a/DashAI/back/models/RAG/chunking_models/recursive_character_chunk_model.py b/DashAI/back/models/RAG/chunking_models/recursive_character_chunk_model.py
new file mode 100644
index 000000000..40f1571f8
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/recursive_character_chunk_model.py
@@ -0,0 +1,129 @@
+from typing import List
+
+from pydantic import field_validator
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    int_field,
+    list_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.chunking_models.base_chunking_model import (
+    BaseChunkingModel,
+)
+
+
+class RecursiveCharacterChunkModelSchema(BaseSchema):
+    chunk_size: schema_field(
+        int_field(gt=1),
+        placeholder=1000,
+        description=MultilingualString(
+            en="Size of each chunk in characters.",
+            es="Tamaño de cada fragmento en caracteres.",
+        ),
+    )  # type: ignore
+
+    chunk_overlap: schema_field(
+        int_field(ge=0),
+        placeholder=100,
+        description=MultilingualString(
+            en=(
+                "Number of characters to overlap between chunks. "
+                "Must be less than chunk_size."
+            ),
+            es=(
+                "Número de caracteres a solapar entre fragmentos. "
+                "Debe ser menor que chunk_size."
+            ),
+        ),
+    )  # type: ignore
+
+    separators: schema_field(
+        list_field(string_field(), min_items=1),
+        placeholder=["\n\n", "\n", ".", " ", ""],
+        description=MultilingualString(
+            en="Ordered list of separators to use for splitting.",
+            es="Lista ordenada de separadores a usar para dividir.",
+        ),
+    )  # type: ignore
+
+    @field_validator("chunk_overlap", mode="after")
+    @classmethod
+    def validate_chunk_overlap(cls, v, info):
+        chunk_size = info.data.get("chunk_size")
+        if chunk_size is not None and v >= chunk_size:
+            raise ValueError(
+                f"chunk_overlap must be less than chunk_size. "
+                f"Got chunk_overlap={v} and chunk_size={chunk_size}"
+            )
+        return v
+
+
+class RecursiveCharacterChunkModel(BaseChunkingModel):
+    SCHEMA = RecursiveCharacterChunkModelSchema
+
+    DISPLAY_NAME: str = MultilingualString(
+        en="Recursive Character Chunk Model",
+        es="Modelo de Fragmentación Recursiva por Caracteres",
+    )
+
+    FLAGS: list[str] = []
+
+    def __init__(self, **kwargs):
+        self.chunk_size = kwargs["chunk_size"]
+        self.chunk_overlap = kwargs["chunk_overlap"]
+        self.separators = kwargs["separators"]
+        super().__init__(**kwargs)
+
+    def chunk_text(self, text: str) -> List[str]:
+        if len(text) <= self.chunk_size:
+            return [text]
+        return self._split_text(text, 0)
+
+    def _split_text(self, text: str, separator_idx: int) -> List[str]:
+        if separator_idx >= len(self.separators):
+            return self._force_split(text)
+
+        separator = self.separators[separator_idx]
+
+        if separator == "" or separator not in text:
+            return self._split_text(text, separator_idx + 1)
+
+        splits = text.split(separator)
+        return self._recursive_split(splits, separator_idx + 1)
+
+    def _recursive_split(self, splits: List[str], next_sep_idx: int) -> List[str]:
+        result: List[str] = []
+        for split in splits:
+            if len(split) <= self.chunk_size:
+                result.append(split)
+                continue
+
+            resolved = False
+            for sep_idx in range(next_sep_idx, len(self.separators)):
+                separator = self.separators[sep_idx]
+                if separator == "" or separator in split:
+                    result.extend(self._split_text(split, sep_idx))
+                    resolved = True
+                    break
+
+            if not resolved:
+                result.extend(self._force_split(split))
+
+        return result
+
+    def _force_split(self, text: str) -> List[str]:
+        chunks: List[str] = []
+        start = 0
+        text_length = len(text)
+
+        while start < text_length:
+            end = min(start + self.chunk_size, text_length)
+            chunks.append(text[start:end])
+            if end == text_length:
+                break
+            start += self.chunk_size - self.chunk_overlap
+
+        return chunks
diff --git a/DashAI/back/models/RAG/chunking_models/token_chunk_model.py b/DashAI/back/models/RAG/chunking_models/token_chunk_model.py
new file mode 100644
index 000000000..07b13c642
--- /dev/null
+++ b/DashAI/back/models/RAG/chunking_models/token_chunk_model.py
@@ -0,0 +1,88 @@
+from typing import TYPE_CHECKING, List, Optional
+
+from pydantic import field_validator
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    int_field,
+    schema_field,
+)
+from DashAI.back.models.RAG.chunking_models.base_chunking_model import (
+    BaseChunkingModel,
+)
+
+
+class TokenChunkModelSchema(BaseSchema):
+    tokenizer_name: schema_field(
+        enum_field(
+            enum=[
+                "intfloat/e5-mistral-7b-instruct",
+                "dccuchile/bert-base-spanish-wwm-uncased",
+            ],
+        ),
+        placeholder="intfloat/e5-mistral-7b-instruct",
+        description="The tokenizer model to use for tokenizing the text.",
+    )  # type: ignore
+
+    chunk_size: schema_field(
+        int_field(gt=1),
+        placeholder=400,
+        description="The size of each chunk in tokens.",
+    )  # type: ignore
+
+    chunk_overlap: schema_field(
+        int_field(ge=0),
+        placeholder=40,
+        description=(
+            "The number of overlapping tokens between chunks. "
+            "Must be less than chunk_size."
+        ),
+    )  # type: ignore
+
+    @field_validator("chunk_overlap")
+    @classmethod
+    def validate_chunk_overlap(cls, v, info):
+        """Validate that chunk_overlap is less than chunk_size."""
+        chunk_size = info.data.get("chunk_size")
+        if chunk_size is not None and v >= chunk_size:
+            raise ValueError(
+                f"chunk_overlap must be less than chunk_size. "
+                f"Got chunk_overlap={v} and chunk_size={chunk_size}"
+            )
+        return v
+
+
+if TYPE_CHECKING:
+    from transformers import AutoTokenizer
+
+
+class TokenChunkModel(BaseChunkingModel):
+    SCHEMA = TokenChunkModelSchema
+
+    def __init__(self, **kwargs):
+        self.parameters = kwargs
+        self.chunk_size = self.parameters["chunk_size"]
+        self.chunk_overlap = self.parameters["chunk_overlap"]
+        self.tokenizer_name = self.parameters["tokenizer_name"]
+        self._tokenizer: Optional["AutoTokenizer"] = None
+        super().__init__(**kwargs)
+
+    @property
+    def tokenizer(self):
+        if self._tokenizer is None:
+            from transformers import AutoTokenizer
+
+            self._tokenizer = AutoTokenizer.from_pretrained(self.tokenizer_name)
+        return self._tokenizer
+
+    def chunk_text(self, text: str) -> List[str]:
+        tokens = self.tokenizer.tokenize(text)
+
+        token_chunks = []
+        while len(tokens) > 0:
+            chunk = tokens[: self.chunk_size]
+            token_chunks.append(self.tokenizer.convert_tokens_to_string(chunk))
+            tokens = tokens[self.chunk_size - self.chunk_overlap :]
+
+        return token_chunks
diff --git a/DashAI/back/models/RAG/document_loader.py b/DashAI/back/models/RAG/document_loader.py
new file mode 100644
index 000000000..9ce48ea52
--- /dev/null
+++ b/DashAI/back/models/RAG/document_loader.py
@@ -0,0 +1,68 @@
+"""Loads Document rows from the database and hydrates in-memory
+BaseDocument instances.  Keeps the pipeline constructor free of raw
+SQL queries.
+"""
+
+from typing import Dict, List
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import Document as DBDocument
+from DashAI.back.models.RAG.documents import BaseDocument, PDFDocument, TxtDocument
+
+_DOCUMENT_CLASSES: Dict[str, type[BaseDocument]] = {
+    "txt": TxtDocument,
+    "pdf": PDFDocument,
+    "md": TxtDocument,
+    "rst": TxtDocument,
+    "tex": TxtDocument,
+    "csv": TxtDocument,
+}
+
+
+class DocumentLoader:
+    """Resolves a list of document IDs to in-memory BaseDocument objects."""
+
+    def __init__(self, db: Session):
+        self.db = db
+
+    def load(self, document_ids: List[int]) -> Dict[int, BaseDocument]:
+        """Load and hydrate documents.
+
+        Parameters
+        ----------
+        document_ids : list[int]
+            IDs present in the ``document`` table.
+
+        Returns
+        -------
+        dict[int, BaseDocument]
+            Mapping from document ID to the hydrated document object.
+
+        Raises
+        ------
+        ValueError
+            If any document ID is not found in the database.
+        """
+        documents: Dict[int, BaseDocument] = {}
+        for doc_id in document_ids:
+            db_doc: DBDocument | None = (
+                self.db.query(DBDocument).filter(DBDocument.id == doc_id).first()
+            )
+            if db_doc is None:
+                raise ValueError(f"Document with ID {doc_id} not found in database.")
+            try:
+                doc_class = _DOCUMENT_CLASSES[db_doc.file_type]
+            except KeyError:
+                raise ValueError(
+                    f"Unsupported file type '{db_doc.file_type}'. "
+                    f"Supported types: txt, pdf, md, rst, tex, csv."
+                )
+            documents[doc_id] = doc_class(
+                id=db_doc.id,
+                file_name=db_doc.file_name,
+                file_path=db_doc.file_path,
+                created=db_doc.created,
+                optional_metadata=db_doc.optional_metadata,
+            )
+        return documents
diff --git a/DashAI/back/models/RAG/documents/__init__.py b/DashAI/back/models/RAG/documents/__init__.py
new file mode 100644
index 000000000..d88c9bd6b
--- /dev/null
+++ b/DashAI/back/models/RAG/documents/__init__.py
@@ -0,0 +1,4 @@
+from .base_document import BaseDocument
+from .pdf_document import PDFDocument
+from .txt_document import TxtDocument
+from .chunk import Chunk
\ No newline at end of file
diff --git a/DashAI/back/models/RAG/documents/base_document.py b/DashAI/back/models/RAG/documents/base_document.py
new file mode 100644
index 000000000..424146d89
--- /dev/null
+++ b/DashAI/back/models/RAG/documents/base_document.py
@@ -0,0 +1,111 @@
+from abc import ABC
+from typing import Any, Dict, Optional
+
+
+class BaseDocument(ABC):
+    """
+    Base class for documents.
+    """
+
+    SUPPORTED_FILE_TYPES = ["pdf", "txt"]
+
+    def __init__(
+        self,
+        id: int,
+        file_name: str,
+        file_path: str,
+        file_hash: str,
+        created: Optional[str] = None,
+        optional_metadata: Optional[Dict[str, Any]] = None,
+    ):
+        """
+        Initialize the document.
+        Args (from database):
+            id (int): The unique identifier of the document.
+            file_name (str): The name of the file.
+            file_path (str): The path to the file.
+            file_hash (str): A hash of the file content.
+            created (Optional[str]): The creation date of the document.
+            optional_metadata (Optional[Dict[str, Any]]): Additional metadata for the document.
+        """
+        self.id = id
+        self.file_name = file_name
+        self.file_path = file_path
+        self.file_hash = file_hash
+        self.created = created if created else None
+        self.optional_metadata = optional_metadata if optional_metadata else None
+
+    def get_text(self) -> str:
+        """
+        Get the text content of the document.
+
+        Returns:
+            str: The text content of the document.
+        """
+        raise NotImplementedError("This method should be implemented by subclasses.")
+
+    def get_text_length(self) -> int:
+        """
+        Get the length of the text content of the document.
+
+        Returns:
+            int: The length of the text content of the document.
+        """
+        return len(self.get_text())
+
+    def get_metadata(self) -> Dict[str, Any]:
+        """
+        Get the metadata of the document.
+
+        Returns:
+            Dict[str, Any]: The metadata of the document.
+        """
+        raise NotImplementedError("This method should be implemented by subclasses.")
+
+    def get_id(self) -> int:
+        """
+        Get the unique identifier of the document.
+
+        Returns:
+            int: The unique identifier of the document.
+        """
+        return self.id
+
+    def get_file_name(self) -> Optional[str]:
+        """
+        Get the filename of the document.
+
+        Returns:
+            Optional[str]: The filename of the document, or None if not applicable.
+        """
+        return self.file_name
+
+    def get_file_path(self) -> Optional[str]:
+        """
+        Get the file path of the document.
+
+        Returns:
+            Optional[str]: The file path of the document, or None if not applicable.
+        """
+        return self.file_path
+
+    def get_file_hash(self) -> str:
+        """
+        Get the hash of the document content.
+
+        Returns:
+            str: A hash string representing the document content.
+        """
+        return self.file_hash
+
+    def get_file_type(self) -> Optional[str]:
+        """
+        Get the filetype of the document.
+
+        Returns:
+            Optional[str]: The filetype of the document, or None if not applicable.
+        """
+        return self.file_path.split(".")[-1].lower() if self.file_path else None
+
+    def __repr__(self):
+        return f"BaseDocument(id={self.id}, filename='{self.get_file_name()}', content='{self.get_text()[:50]}...', metadata={self.get_metadata()})"
diff --git a/DashAI/back/models/RAG/documents/chunk.py b/DashAI/back/models/RAG/documents/chunk.py
new file mode 100644
index 000000000..4bab49839
--- /dev/null
+++ b/DashAI/back/models/RAG/documents/chunk.py
@@ -0,0 +1,18 @@
+from DashAI.back.models.RAG.utils import hash_function
+
+
+class Chunk:
+    """A class representing a chunk of a document."""
+
+    def __init__(
+        self,
+        id: int,
+        document_id: str,
+        document_position: int,
+        text: str,
+    ):
+        self.id = id
+        self.document_id = document_id
+        self.document_position = document_position
+        self.text = text
+        self.hash = hash_function(text)
diff --git a/DashAI/back/models/RAG/documents/pdf_document.py b/DashAI/back/models/RAG/documents/pdf_document.py
new file mode 100644
index 000000000..ed62d31d7
--- /dev/null
+++ b/DashAI/back/models/RAG/documents/pdf_document.py
@@ -0,0 +1,82 @@
+from typing import Any, Dict, Optional
+
+from DashAI.back.models.RAG.documents.base_document import BaseDocument
+from DashAI.back.models.RAG.utils import hash_function
+
+
+def _clean_textract_output(text: str) -> str:
+    """Clean textract output by removing control characters and normalizing whitespace."""
+    import re
+
+    text = re.sub(r"[\x00-\x1f\x7f]", " ", text)
+    text = re.sub(r"\s+", " ", text)
+    return text.strip()
+
+
+class PDFDocument(BaseDocument):
+    """
+    Class representing a PDF document.
+    """
+
+    def __init__(
+        self,
+        id: int,
+        file_name: str,
+        file_path: str,
+        created: Optional[str] = None,
+        optional_metadata: Optional[Dict[str, Any]] = None,
+        **kwargs,
+    ):
+        """
+        Initialize the document.
+        Args (from database):
+            id (int): The unique identifier of the document.
+            file_name (str): The name of the file.
+            file_path (str): The path to the file.
+            created (Optional[str]): The creation date of the document.
+            optional_metadata (Optional[Dict[str, Any]]): Additional metadata for the document.
+        """
+        self.PARSER = kwargs.get("parser", "textract")
+        file_hash = hash_function(file_path)
+        super().__init__(
+            id=id,
+            file_name=file_name,
+            file_path=file_path,
+            file_hash=file_hash,
+            created=created,
+            optional_metadata=optional_metadata,
+        )
+
+    def get_text(self) -> str:
+        if self.PARSER == "PyPDF2":
+            from PyPDF2 import PdfReader
+
+            reader = PdfReader(self.file_path)
+            if not reader.pages:
+                raise ValueError(
+                    f"The PDF file {self.file_path} is empty or not valid."
+                )
+
+            text = ""
+            for page in reader.pages:
+                text += page.extract_text() or ""
+
+            return text.strip()
+        elif self.PARSER == "textract":
+            import textract
+
+            try:
+                text = textract.process(self.file_path, output_encoding="utf-8").decode(
+                    "utf-8"
+                )
+                text = _clean_textract_output(text)
+                return text.strip()
+            except Exception as e:
+                raise ValueError(
+                    f"Error extracting text from PDF file {self.file_path}: {str(e)}"
+                )
+        else:
+            raise ValueError(f"Unsupported parser: {self.PARSER}")
+
+    def get_metadata(self):
+        return self.optional_metadata if self.optional_metadata else {}
diff --git a/DashAI/back/models/RAG/documents/txt_document.py b/DashAI/back/models/RAG/documents/txt_document.py
new file mode 100644
index 000000000..0b94be276
--- /dev/null
+++ b/DashAI/back/models/RAG/documents/txt_document.py
@@ -0,0 +1,58 @@
+from typing import Any, Dict, Optional
+
+from DashAI.back.models.RAG.documents.base_document import BaseDocument
+from DashAI.back.models.RAG.utils import hash_function
+
+
+class TxtDocument(BaseDocument):
+    """
+    Class representing a .txt document.
+    """
+
+    def __init__(
+        self,
+        id: int,
+        file_name: str,
+        file_path: str,
+        created: Optional[str] = None,
+        optional_metadata: Optional[Dict[str, Any]] = None,
+    ):
+        """
+        Initialize the document.
+        Args (from database):
+            id (int): The unique identifier of the document.
+            file_name (str): The name of the file.
+            file_path (str): The path to the file.
+            file_hash (str): A hash of the file content.
+            created (Optional[str]): The creation date of the document.
+            optional_metadata (Optional[Dict[str, Any]]): Additional metadata for the document.
+        """
+        file_hash = hash_function(file_path)
+        super().__init__(
+            id=id,
+            file_name=file_name,
+            file_path=file_path,
+            file_hash=file_hash,
+            created=created,
+            optional_metadata=optional_metadata,
+        )
+
+    def get_text(self) -> str:
+        """
+        Get the text content of the document.
+
+        Returns:
+            str: The text content of the document.
+        """
+        with open(self.file_path, "r", encoding="utf-8") as file:
+            text = file.read()
+        return text.strip()
+
+    def get_metadata(self) -> Dict[str, Any]:
+        """
+        Get the metadata of the document.
+
+        Returns:
+            Dict[str, Any]: The metadata of the document.
+        """
+        return self.optional_metadata if self.optional_metadata else {}
diff --git a/DashAI/back/models/RAG/embeddings/__init__.py b/DashAI/back/models/RAG/embeddings/__init__.py
new file mode 100644
index 000000000..f0f81ee2a
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/__init__.py
@@ -0,0 +1,35 @@
+"""Embedding model exports."""
+
+from DashAI.back.models.RAG.embeddings.dense.bert_embedding import BERTEmbedding
+from DashAI.back.models.RAG.embeddings.dense.distilbert_embedding import (
+    DistilBERTEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense.e5_embedding import E5Embedding
+from DashAI.back.models.RAG.embeddings.dense.gemma_embedding import GemmaEmbedding
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import (
+    HuggingFaceEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense.instructor_embedding import (
+    InstructorEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense.labse_embedding import LaBSEmbedding
+from DashAI.back.models.RAG.embeddings.dense.openai_embedding import OpenAIEmbedding
+from DashAI.back.models.RAG.embeddings.dense.roberta_embedding import RoBERTaEmbedding
+from DashAI.back.models.RAG.embeddings.dense.sentence_transformer_embedding import (
+    SentenceTransformerEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+__all__ = [
+    "DenseEmbedding",
+    "BERTEmbedding",
+    "DistilBERTEmbedding",
+    "E5Embedding",
+    "GemmaEmbedding",
+    "HuggingFaceEmbedding",
+    "InstructorEmbedding",
+    "LaBSEmbedding",
+    "OpenAIEmbedding",
+    "RoBERTaEmbedding",
+    "SentenceTransformerEmbedding",
+]
diff --git a/DashAI/back/models/RAG/embeddings/dense/__init__.py b/DashAI/back/models/RAG/embeddings/dense/__init__.py
new file mode 100644
index 000000000..e15b168ba
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/__init__.py
@@ -0,0 +1,10 @@
+from DashAI.back.models.RAG.embeddings.dense.bert_embedding import BERTEmbedding
+from DashAI.back.models.RAG.embeddings.dense.distilbert_embedding import DistilBERTEmbedding
+from DashAI.back.models.RAG.embeddings.dense.e5_embedding import E5Embedding
+from DashAI.back.models.RAG.embeddings.dense.gemma_embedding import GemmaEmbedding
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import HuggingFaceEmbedding
+from DashAI.back.models.RAG.embeddings.dense.instructor_embedding import InstructorEmbedding
+from DashAI.back.models.RAG.embeddings.dense.labse_embedding import LaBSEmbedding
+from DashAI.back.models.RAG.embeddings.dense.openai_embedding import OpenAIEmbedding
+from DashAI.back.models.RAG.embeddings.dense.roberta_embedding import RoBERTaEmbedding
+from DashAI.back.models.RAG.embeddings.dense.sentence_transformer_embedding import SentenceTransformerEmbedding
diff --git a/DashAI/back/models/RAG/embeddings/dense/_bert_embedding.py b/DashAI/back/models/RAG/embeddings/dense/_bert_embedding.py
new file mode 100644
index 000000000..0aee9915a
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_bert_embedding.py
@@ -0,0 +1,70 @@
+from typing import Dict
+
+import torch
+
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    OverfloatHandler,
+)
+
+CLS = "cls"
+MEAN = "mean"
+MAX = "max"
+CONCAT_2 = "concat_2"
+CONCAT_3 = "concat_3"
+CONCAT_4 = "concat_4"
+
+POOLING_STRATEGIES: Dict[str, str] = {
+    CLS: "CLS token",
+    MEAN: "Mean pooling",
+    MAX: "Max pooling",
+    CONCAT_2: "Concat last 2 layers",
+    CONCAT_3: "Concat last 3 layers",
+    CONCAT_4: "Concat last 4 layers",
+}
+
+
+class _BERTEmbedding(OverfloatHandler):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        model_max_length: int,
+        overflow_strategy: str,
+        pooling_strategy: str,
+    ):
+        super().__init__(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_max_length,
+            overflow_strategy=overflow_strategy,
+        )
+        self.pooling_strategy = pooling_strategy
+        self.params["pooling_strategy"] = pooling_strategy
+
+    def load(self):
+        from transformers import AutoModel, AutoTokenizer
+        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+        self.model = AutoModel.from_pretrained(
+            self.model_name,
+            output_hidden_states=self.pooling_strategy in (CONCAT_2, CONCAT_3, CONCAT_4),
+        ).to(self.device)
+
+    def _pool(self, model_output, attention_mask):
+        if self.pooling_strategy == CLS:
+            return model_output[0][:, 0]
+        if self.pooling_strategy == MAX:
+            token_embeddings = model_output[0]
+            input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+            token_embeddings[input_mask_expanded == 0] = -1e9
+            return torch.max(token_embeddings, 1)[0]
+        if self.pooling_strategy in (CONCAT_2, CONCAT_3, CONCAT_4):
+            hidden_states = model_output.hidden_states
+            layer_count = int(self.pooling_strategy.split("_")[1])
+            selected = [hidden_states[-(i + 1)][:, 0, :] for i in range(layer_count)]
+            return torch.cat(selected, dim=1)
+        token_embeddings = model_output[0]
+        input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+        pooled = torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(
+            input_mask_expanded.sum(1), min=1e-9
+        )
+        return pooled
diff --git a/DashAI/back/models/RAG/embeddings/dense/_e5_embedding.py b/DashAI/back/models/RAG/embeddings/dense/_e5_embedding.py
new file mode 100644
index 000000000..ac98a39b0
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_e5_embedding.py
@@ -0,0 +1,42 @@
+from typing import List
+
+import numpy as np
+import torch.nn.functional as F
+
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    OverfloatHandler,
+)
+
+QUERY_PREFIX = "query: "
+PASSAGE_PREFIX = "passage: "
+
+
+class _E5Embedding(OverfloatHandler):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        model_max_length: int,
+        overflow_strategy: str,
+    ):
+        super().__init__(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_max_length,
+            overflow_strategy=overflow_strategy,
+        )
+
+    def _pool(self, model_output, attention_mask):
+        last_hidden = model_output[0]
+        mask = attention_mask[..., None].bool()
+        last_hidden = last_hidden.masked_fill(~mask, 0.0)
+        pooled = last_hidden.sum(dim=1) / attention_mask.sum(dim=1)[..., None]
+        return F.normalize(pooled, p=2, dim=1)
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        prefixed = [f"{PASSAGE_PREFIX}{t}" for t in texts]
+        return self._batch_encode_impl(prefixed)
+
+    def encode(self, text: str) -> np.ndarray:
+        result = self._batch_encode_impl([f"{QUERY_PREFIX}{text}"])
+        return result.squeeze()
diff --git a/DashAI/back/models/RAG/embeddings/dense/_gemma_embedding.py b/DashAI/back/models/RAG/embeddings/dense/_gemma_embedding.py
new file mode 100644
index 000000000..bb59ff694
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_gemma_embedding.py
@@ -0,0 +1,55 @@
+from typing import List
+
+import numpy as np
+
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import (
+    HuggingFaceEmbedding,
+)
+
+TASK_PROMPTS = {
+    "search_result": "task: search result | query: ",
+    "question_answering": "task: question answering | query: ",
+    "fact_checking": "task: fact checking | query: ",
+    "classification": "task: classification | query: ",
+    "clustering": "task: clustering | query: ",
+    "sentence_similarity": "task: sentence similarity | query: ",
+    "code_retrieval": "task: code retrieval | query: ",
+}
+
+DOCUMENT_PROMPT = "title: none | text: "
+
+
+class _GemmaEmbedding(HuggingFaceEmbedding):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        model_max_length: int,
+        overflow_strategy: str,
+        task_type: str,
+    ):
+        super().__init__(model_name=model_name, device=device)
+        self.model_max_length = model_max_length
+        self.overflow_strategy = overflow_strategy
+        self.task_type = task_type
+        self.params["model_max_length"] = model_max_length
+        self.params["overflow_strategy"] = overflow_strategy
+        self.params["task_type"] = task_type
+
+    def _pool(self, model_output, attention_mask):
+        raise NotImplementedError("Gemma uses SentenceTransformer API, _pool is unused.")
+
+    def load(self):
+        from sentence_transformers import SentenceTransformer
+
+        self.model = SentenceTransformer(self.model_name, device=self.device)
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        prompted = [DOCUMENT_PROMPT + t for t in texts]
+        return self.model.encode(prompted, normalize_embeddings=True, show_progress_bar=False)
+
+    def encode(self, text: str) -> np.ndarray:
+        query_prompt = TASK_PROMPTS[self.task_type]
+        return self.model.encode(
+            [query_prompt + text], normalize_embeddings=True, show_progress_bar=False
+        ).squeeze()
diff --git a/DashAI/back/models/RAG/embeddings/dense/_instructor_embedding.py b/DashAI/back/models/RAG/embeddings/dense/_instructor_embedding.py
new file mode 100644
index 000000000..4fe3651a5
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_instructor_embedding.py
@@ -0,0 +1,35 @@
+from typing import List
+
+import numpy as np
+
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import (
+    HuggingFaceEmbedding,
+)
+
+
+class _InstructorEmbedding(HuggingFaceEmbedding):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        instruction: str,
+    ):
+        super().__init__(model_name=model_name, device=device)
+        self.instruction = instruction
+        self.params["instruction"] = instruction
+
+    def _pool(self, model_output, attention_mask):
+        raise NotImplementedError("INSTRUCTOR uses custom encoding API, _pool is unused.")
+
+    def load(self):
+        from InstructorEmbedding import INSTRUCTOR
+        self.model = INSTRUCTOR(self.model_name)
+        self.model._text_length = self.model._input_length
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        pairs = [[self.instruction, text] for text in texts]
+        return self.model.encode(pairs, show_progress_bar=False)
+
+    def encode(self, text: str) -> np.ndarray:
+        result = self.model.encode([[self.instruction, text]], show_progress_bar=False)
+        return result.squeeze()
diff --git a/DashAI/back/models/RAG/embeddings/dense/_overflow_handler.py b/DashAI/back/models/RAG/embeddings/dense/_overflow_handler.py
new file mode 100644
index 000000000..2173c6483
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_overflow_handler.py
@@ -0,0 +1,62 @@
+from abc import abstractmethod
+from typing import List
+
+import numpy as np
+import torch
+
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import (
+    HuggingFaceEmbedding,
+)
+
+TRUNCATE = "truncate"
+AGGREGATE = "aggregate"
+
+
+class OverfloatHandler(HuggingFaceEmbedding):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        model_max_length: int,
+        overflow_strategy: str,
+    ):
+        super().__init__(model_name=model_name, device=device)
+        self.model_max_length = model_max_length
+        self.overflow_strategy = overflow_strategy
+        self.params["model_max_length"] = model_max_length
+        self.params["overflow_strategy"] = overflow_strategy
+
+    @abstractmethod
+    def _pool(self, model_output, attention_mask):
+        raise NotImplementedError
+
+    def _batch_encode_impl(self, texts: List[str]) -> np.ndarray:
+        encoded = self.tokenizer(
+            texts,
+            padding=True,
+            truncation=True,
+            max_length=self.model_max_length,
+            return_tensors="pt",
+            return_overflowing_tokens=self.overflow_strategy == AGGREGATE,
+            stride=0,
+        )
+        num_texts = len(texts)
+        if self.overflow_strategy == AGGREGATE and "overflow_to_sample_mapping" in encoded:
+            mapping = encoded.pop("overflow_to_sample_mapping")
+            encoded = {k: v.to(self.device) for k, v in encoded.items()}
+            with torch.no_grad():
+                outputs = self.model(**encoded)
+            segment_embeddings = self._pool(outputs, encoded["attention_mask"]).cpu()
+            result = []
+            for i in range(num_texts):
+                segment_indices = (mapping == i).nonzero(as_tuple=True)[0]
+                if len(segment_indices) == 1:
+                    result.append(segment_embeddings[segment_indices[0]])
+                else:
+                    result.append(torch.mean(segment_embeddings[segment_indices], dim=0))
+            return torch.stack(result).numpy()
+        encoded = {k: v.to(self.device) for k, v in encoded.items()}
+        with torch.no_grad():
+            outputs = self.model(**encoded)
+        embeddings = self._pool(outputs, encoded["attention_mask"])
+        return embeddings.cpu().numpy()
diff --git a/DashAI/back/models/RAG/embeddings/dense/_sentence_transformer_embedding.py b/DashAI/back/models/RAG/embeddings/dense/_sentence_transformer_embedding.py
new file mode 100644
index 000000000..3d10497a4
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/_sentence_transformer_embedding.py
@@ -0,0 +1,56 @@
+import torch
+import torch.nn.functional as F
+
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    OverfloatHandler,
+)
+
+TRUNCATE = "truncate"
+AGGREGATE = "aggregate"
+
+
+class _SentenceTransformerEmbedding(OverfloatHandler):
+    def __init__(
+        self,
+        model_name: str,
+        device: str,
+        model_max_length: int,
+        overflow_strategy: str,
+        normalize: bool,
+        pooling: str = "mean",
+    ):
+        super().__init__(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_max_length,
+            overflow_strategy=overflow_strategy,
+        )
+        self.normalize = normalize
+        self.pooling = pooling
+        self.params["normalize"] = normalize
+        self.params["pooling"] = pooling
+
+    def _pool(self, model_output, attention_mask):
+        if self.pooling == "last_token":
+            return self._last_token_pool(model_output, attention_mask)
+        token_embeddings = model_output[0]
+        input_mask_expanded = attention_mask.unsqueeze(-1).expand(token_embeddings.size()).float()
+        pooled = torch.sum(token_embeddings * input_mask_expanded, 1) / torch.clamp(
+            input_mask_expanded.sum(1), min=1e-9
+        )
+        if self.normalize:
+            pooled = F.normalize(pooled, p=2, dim=1)
+        return pooled
+
+    def _last_token_pool(self, model_output, attention_mask):
+        last_hidden_states = model_output[0]
+        left_padding = (attention_mask[:, -1].sum() == attention_mask.shape[0])
+        if left_padding:
+            pooled = last_hidden_states[:, -1]
+        else:
+            sequence_lengths = attention_mask.sum(dim=1) - 1
+            batch_size = last_hidden_states.shape[0]
+            pooled = last_hidden_states[torch.arange(batch_size, device=last_hidden_states.device), sequence_lengths]
+        if self.normalize:
+            pooled = F.normalize(pooled, p=2, dim=1)
+        return pooled
diff --git a/DashAI/back/models/RAG/embeddings/dense/bert_embedding.py b/DashAI/back/models/RAG/embeddings/dense/bert_embedding.py
new file mode 100644
index 000000000..12a4c1b27
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/bert_embedding.py
@@ -0,0 +1,146 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._bert_embedding import (
+    CLS,
+    CONCAT_2,
+    CONCAT_3,
+    CONCAT_4,
+    MAX,
+    MEAN,
+    _BERTEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+BERT_POOLING_STRATEGIES = [CLS, MEAN, MAX, CONCAT_2, CONCAT_3, CONCAT_4]
+
+BERT_MODELS: Dict[str, dict] = {
+    "google-bert/bert-base-cased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "google-bert/bert-base-uncased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "google-bert/bert-large-cased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "google-bert/bert-large-uncased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "google-bert/bert-base-multilingual-cased": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "google-bert/bert-base-multilingual-uncased": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+}
+
+BERT_MODEL_NAMES = list(BERT_MODELS.keys())
+
+
+class BERTEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(BERT_MODEL_NAMES),
+        placeholder="google-bert/bert-base-cased",
+        description=MultilingualString(
+            en="BERT model for embedding generation.",
+            es="Modelo BERT para generación de embeddings.",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+    pooling_strategy: schema_field(
+        enum_field(BERT_POOLING_STRATEGIES),
+        placeholder=MEAN,
+        description=MultilingualString(
+            en="Pooling strategy to aggregate token embeddings.",
+            es="Estrategia de pooling para agregar embeddings de tokens.",
+        ),
+    )  # type: ignore
+
+
+class BERTEmbedding(DenseEmbedding):
+    SCHEMA = BERTEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:bert", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="BERT Embedding",
+        es="Embedding BERT",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using BERT models with configurable pooling (CLS, mean, max, concat layers).",
+        es="Embeddings densos usando modelos BERT con pooling configurable (CLS, mean, max, concat layers).",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        pooling_strategy = self.params["pooling_strategy"]
+        model_info = BERT_MODELS[model_name]
+        self._embedding = _BERTEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            pooling_strategy=pooling_strategy,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/distilbert_embedding.py b/DashAI/back/models/RAG/embeddings/dense/distilbert_embedding.py
new file mode 100644
index 000000000..ec056e266
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/distilbert_embedding.py
@@ -0,0 +1,132 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._bert_embedding import (
+    CLS,
+    CONCAT_2,
+    CONCAT_3,
+    CONCAT_4,
+    MAX,
+    MEAN,
+    _BERTEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+DISTILBERT_POOLING_STRATEGIES = [CLS, MEAN, MAX, CONCAT_2, CONCAT_3, CONCAT_4]
+
+DISTILBERT_MODELS: Dict[str, dict] = {
+    "distilbert/distilbert-base-cased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "distilbert/distilbert-base-uncased": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "distilbert/distilbert-base-multilingual-cased": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "distilbert/distilbert-roberta-base": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+}
+
+DISTILBERT_MODEL_NAMES = list(DISTILBERT_MODELS.keys())
+
+
+class DistilBERTEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(DISTILBERT_MODEL_NAMES),
+        placeholder="distilbert/distilbert-base-cased",
+        description=MultilingualString(
+            en="DistilBERT model for embedding generation.",
+            es="Modelo DistilBERT para generación de embeddings.",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+    pooling_strategy: schema_field(
+        enum_field(DISTILBERT_POOLING_STRATEGIES),
+        placeholder=MEAN,
+        description=MultilingualString(
+            en="Pooling strategy to aggregate token embeddings.",
+            es="Estrategia de pooling para agregar embeddings de tokens.",
+        ),
+    )  # type: ignore
+
+
+class DistilBERTEmbedding(DenseEmbedding):
+    SCHEMA = DistilBERTEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:distilbert", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="DistilBERT Embedding",
+        es="Embedding DistilBERT",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using DistilBERT models with configurable pooling.",
+        es="Embeddings densos usando modelos DistilBERT con pooling configurable.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        pooling_strategy = self.params["pooling_strategy"]
+        model_info = DISTILBERT_MODELS[model_name]
+        self._embedding = _BERTEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            pooling_strategy=pooling_strategy,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/e5_embedding.py b/DashAI/back/models/RAG/embeddings/dense/e5_embedding.py
new file mode 100644
index 000000000..21f5678af
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/e5_embedding.py
@@ -0,0 +1,131 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._e5_embedding import _E5Embedding
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+E5_MODELS: Dict[str, dict] = {
+    "intfloat/e5-small-v2": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "intfloat/e5-large-v2": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "intfloat/multilingual-e5-large": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "intfloat/multilingual-e5-base": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "intfloat/multilingual-e5-small": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "intfloat/e5-mistral-7b-instruct": {
+        "languages": ["en"],
+        "max_seq_length": 4096,
+    },
+}
+
+E5_MODEL_NAMES = list(E5_MODELS.keys())
+
+
+class E5EmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(E5_MODEL_NAMES),
+        placeholder="intfloat/e5-small-v2",
+        description=MultilingualString(
+            en="E5 model for embedding generation (uses query/passage prefixes).",
+            es="Modelo E5 para generación de embeddings (usa prefijos query/passage).",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+
+class E5Embedding(DenseEmbedding):
+    SCHEMA = E5EmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:e5", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="E5 Embedding",
+        es="Embedding E5",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using E5 models with average pooling + L2 normalization + query/passage prefixes.",
+        es="Embeddings densos usando modelos E5 con average pooling + normalización L2 + prefijos query/passage.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        model_info = E5_MODELS[model_name]
+        self._embedding = _E5Embedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/fasttext_embedding.py b/DashAI/back/models/RAG/embeddings/dense/fasttext_embedding.py
new file mode 100644
index 000000000..af4db87c1
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/fasttext_embedding.py
@@ -0,0 +1,74 @@
+from typing import List
+
+import fasttext
+import numpy as np
+from huggingface_hub import hf_hub_download
+
+from DashAI.back.core.schema_fields import enum_field
+from DashAI.back.core.schema_fields.base_schema import BaseSchema
+from DashAI.back.core.schema_fields.schema_field import schema_field
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+
+class FastTextEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(["facebook/fasttext-es-vectors", "facebook/fasttext-en-vectors"]),
+        "facebook/fasttext-en-vectors",
+        "Name of the pre-trained model to use",
+    )  # type: ignore
+
+    pooling_strategy: schema_field(
+        enum_field(["mean", "max"]),
+        "mean",
+        "Pooling strategy to use",
+    )  # type: ignore
+
+
+class FastTextEmbedding(DenseEmbedding):
+    """FastText embedding"""
+
+    FLAGS: list[str] = ["FAMILY:fasttext"]
+    SCHEMA = FastTextEmbeddingSchema
+    DISPLAY_NAME = MultilingualString(
+        en="FastText Embedding",
+        es="Embedding FastText",
+    )
+    DESCRIPTION = MultilingualString(
+        en="Convert text to embeddings using FastText.",
+        es="Convierte texto a embeddings usando FastText.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        self.model_name = self.params["model_name"]
+        self.pooling_strategy = self.params["pooling_strategy"]
+        pooling_functions = {"mean": np.mean, "max": np.max}
+        self.pooling_function = pooling_functions[self.pooling_strategy]
+        self.model = None
+
+    def load(self):
+        """Load the FastText model."""
+        if self.model is not None:
+            return
+        model_path = hf_hub_download(repo_id=self.model_name, filename="model.bin")
+        self.model = fasttext.load_model(model_path)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        pass
+
+    def encode(self, text: str) -> np.ndarray:
+        """Encode text into an embedding."""
+        token_embeddings = [self.model.get_word_vector(word) for word in text.split()]
+        return self.pooling_function(np.array(token_embeddings), axis=0)
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        """Encode a batch of texts into embeddings."""
+        all_embeddings = []
+        for text in texts:
+            embedding = self.encode(text)
+            all_embeddings.append(embedding)
+        return np.array(all_embeddings)
diff --git a/DashAI/back/models/RAG/embeddings/dense/gemma_embedding.py b/DashAI/back/models/RAG/embeddings/dense/gemma_embedding.py
new file mode 100644
index 000000000..bd1ec013e
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/gemma_embedding.py
@@ -0,0 +1,109 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._gemma_embedding import (
+    TASK_PROMPTS,
+    _GemmaEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+TASK_TYPES = list(TASK_PROMPTS.keys())
+
+GEMMA_MODELS: Dict[str, dict] = {
+    "google/embeddinggemma-300m": {
+        "languages": ["en"],
+        "max_seq_length": 8192,
+    },
+}
+
+GEMMA_MODEL_NAMES = list(GEMMA_MODELS.keys())
+
+
+class GemmaEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(GEMMA_MODEL_NAMES),
+        placeholder="google/embeddinggemma-300m",
+        description=MultilingualString(
+            en="Gemma embedding model (uses SentenceTransformers API).",
+            es="Modelo de embedding Gemma (usa API SentenceTransformers).",
+        ),
+    )  # type: ignore
+
+    task_type: schema_field(
+        enum_field(TASK_TYPES),
+        placeholder="search_result",
+        description=MultilingualString(
+            en="Task type that optimizes the query embedding for a specific use case.",
+            es="Tipo de tarea que optimiza el embedding de consulta para un caso de uso específico.",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+
+class GemmaEmbedding(DenseEmbedding):
+    SCHEMA = GemmaEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:gemma", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="Gemma Embedding",
+        es="Embedding Gemma",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using Gemma models via SentenceTransformers API.",
+        es="Embeddings densos usando modelos Gemma mediante API SentenceTransformers.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        task_type = self.params["task_type"]
+        model_info = GEMMA_MODELS[model_name]
+        self._embedding = _GemmaEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            task_type=task_type,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/huggingface_embedding.py b/DashAI/back/models/RAG/embeddings/dense/huggingface_embedding.py
new file mode 100644
index 000000000..e3e6d07e7
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/huggingface_embedding.py
@@ -0,0 +1,56 @@
+from abc import abstractmethod
+from typing import List
+
+import numpy as np
+import torch
+from transformers import AutoModel, AutoTokenizer
+
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+
+class HuggingFaceEmbedding(DenseEmbedding):
+    FLAGS: list[str] = ["abstract", "huggingface"]
+
+    def __init__(self, model_name: str, device: str):
+        self.model_name = model_name
+        self.device = device
+        self.params = {
+            "model_name": model_name,
+            "device": device,
+        }
+        self.model = None
+        self.tokenizer = None
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
+
+    @abstractmethod
+    def _pool(self, model_output, attention_mask):
+        raise NotImplementedError
+
+    def load(self):
+        self.tokenizer = AutoTokenizer.from_pretrained(self.model_name)
+        self.model = AutoModel.from_pretrained(self.model_name).to(self.device)
+
+    def _batch_encode_impl(self, texts: List[str]) -> np.ndarray:
+        encoded = self.tokenizer(
+            texts,
+            padding=True,
+            truncation=True,
+            return_tensors="pt",
+        )
+        encoded = {k: v.to(self.device) for k, v in encoded.items()}
+        with torch.no_grad():
+            outputs = self.model(**encoded)
+        embeddings = self._pool(outputs, encoded["attention_mask"])
+        return embeddings.cpu().numpy()
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        return self._batch_encode_impl(texts)
+
+    def encode(self, text: str) -> np.ndarray:
+        embeddings = self.batch_encode([text])
+        return embeddings.squeeze()
diff --git a/DashAI/back/models/RAG/embeddings/dense/instructor_embedding.py b/DashAI/back/models/RAG/embeddings/dense/instructor_embedding.py
new file mode 100644
index 000000000..b33f598ee
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/instructor_embedding.py
@@ -0,0 +1,95 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._instructor_embedding import (
+    _InstructorEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+INSTRUCTOR_MODELS: Dict[str, dict] = {
+    "hkunlp/instructor-base": {
+        "languages": ["en"],
+    },
+    "hkunlp/instructor-large": {
+        "languages": ["en"],
+    },
+    "hkunlp/instructor-xl": {
+        "languages": ["en"],
+    },
+}
+
+INSTRUCTOR_MODEL_NAMES = list(INSTRUCTOR_MODELS.keys())
+
+
+class InstructorEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(INSTRUCTOR_MODEL_NAMES),
+        placeholder="hkunlp/instructor-base",
+        description=MultilingualString(
+            en="INSTRUCTOR model for instruction-tuned embedding generation.",
+            es="Modelo INSTRUCTOR para generación de embeddings ajustados por instrucción.",
+        ),
+    )  # type: ignore
+
+    instruction: schema_field(
+        string_field(),
+        placeholder="Represent the document for retrieval:",
+        description=MultilingualString(
+            en="Instruction text that guides the embedding model.",
+            es="Texto de instrucción que guía al modelo de embedding.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+
+class InstructorEmbedding(DenseEmbedding):
+    SCHEMA = InstructorEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:instructor", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="INSTRUCTOR Embedding",
+        es="Embedding INSTRUCTOR",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using INSTRUCTOR instruction-tuned models.",
+        es="Embeddings densos usando modelos INSTRUCTOR ajustados por instrucción.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        instruction = self.params["instruction"]
+        self._embedding = _InstructorEmbedding(
+            model_name=model_name,
+            device=device,
+            instruction=instruction,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/labse_embedding.py b/DashAI/back/models/RAG/embeddings/dense/labse_embedding.py
new file mode 100644
index 000000000..a38a8d33a
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/labse_embedding.py
@@ -0,0 +1,102 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense._sentence_transformer_embedding import (
+    _SentenceTransformerEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+LABSE_MODELS: Dict[str, dict] = {
+    "sentence-transformers/LaBSE": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "gu", "ka", "ku", "my", "sq", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+}
+
+LABSE_MODEL_NAMES = list(LABSE_MODELS.keys())
+
+
+class LaBSEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(LABSE_MODEL_NAMES),
+        placeholder="sentence-transformers/LaBSE",
+        description=MultilingualString(
+            en="LaBSE model for multilingual embedding generation (109 languages).",
+            es="Modelo LaBSE para generación de embeddings multilingües (109 idiomas).",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+
+class LaBSEmbedding(DenseEmbedding):
+    SCHEMA = LaBSEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:labse", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="LaBSE Embedding",
+        es="Embedding LaBSE",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using LaBSE multilingual model (109 languages).",
+        es="Embeddings densos usando el modelo multilingüe LaBSE (109 idiomas).",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        model_info = LABSE_MODELS[model_name]
+        self._embedding = _SentenceTransformerEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            normalize=True,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/openai_embedding.py b/DashAI/back/models/RAG/embeddings/dense/openai_embedding.py
new file mode 100644
index 000000000..d08d3fb2a
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/openai_embedding.py
@@ -0,0 +1,73 @@
+from typing import List
+
+import numpy as np
+from openai import OpenAI
+
+from DashAI.back.core.schema_fields import BaseSchema
+from DashAI.back.core.schema_fields.enum_field import enum_field
+from DashAI.back.core.schema_fields.schema_field import schema_field
+from DashAI.back.core.schema_fields.string_field import string_field
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+
+class OpenAIEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(["text-embedding-ada-002", "text-embedding-3-small", "text-embedding-3-large"]),
+        placeholder="text-embedding-3-small",
+        description=MultilingualString(
+            en="OpenAI embedding model to use.",
+            es="Modelo de embedding de OpenAI a utilizar.",
+        ),
+    )  # type: ignore
+
+    api_key: schema_field(
+        string_field(),
+        placeholder="",
+        description=MultilingualString(
+            en="OpenAI API key.",
+            es="Clave API de OpenAI.",
+        ),
+    )  # type: ignore
+
+
+class OpenAIEmbedding(DenseEmbedding):
+    FLAGS: list[str] = ["FAMILY:openai", "remote"]
+    DISPLAY_NAME = MultilingualString(
+        en="OpenAI Embedding",
+        es="Embedding OpenAI",
+    )
+    DESCRIPTION = MultilingualString(
+        en="OpenAI text embeddings",
+        es="Embeddings de texto de OpenAI",
+    )
+    SCHEMA = OpenAIEmbeddingSchema
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        self.model_name = self.params["model_name"]
+        self.api_key = self.params["api_key"]
+        self.client = OpenAI(api_key=self.api_key)
+
+    def load(self):
+        pass
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
+
+    def encode(self, text: str) -> np.ndarray:
+        response = self.client.embeddings.create(
+            model=self.model_name,
+            input=text,
+        )
+        return np.array(response.data[0].embedding)
+
+    def batch_encode(self, texts: List[str]) -> np.ndarray:
+        response = self.client.embeddings.create(
+            model=self.model_name,
+            input=texts,
+        )
+        return np.array([d.embedding for d in response.data])
diff --git a/DashAI/back/models/RAG/embeddings/dense/roberta_embedding.py b/DashAI/back/models/RAG/embeddings/dense/roberta_embedding.py
new file mode 100644
index 000000000..d916ea55d
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/roberta_embedding.py
@@ -0,0 +1,134 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._bert_embedding import (
+    MAX,
+    MEAN,
+    _BERTEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+ROBERTA_POOLING_STRATEGIES = [MEAN, MAX]
+
+ROBERTA_MODELS: Dict[str, dict] = {
+    "FacebookAI/roberta-base": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "FacebookAI/roberta-large": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+    },
+    "FacebookAI/xlm-roberta-base": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+    "FacebookAI/xlm-roberta-large": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+    },
+}
+
+ROBERTA_MODEL_NAMES = list(ROBERTA_MODELS.keys())
+
+
+class RoBERTaEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(ROBERTA_MODEL_NAMES),
+        placeholder="FacebookAI/roberta-base",
+        description=MultilingualString(
+            en="RoBERTa / XLM-RoBERTa model for embedding generation.",
+            es="Modelo RoBERTa / XLM-RoBERTa para generación de embeddings.",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+    pooling_strategy: schema_field(
+        enum_field(ROBERTA_POOLING_STRATEGIES),
+        placeholder=MEAN,
+        description=MultilingualString(
+            en="Pooling strategy to aggregate token embeddings. RoBERTa CLS token is not trained for similarity.",
+            es="Estrategia de pooling para agregar embeddings de tokens. El token CLS de RoBERTa no está entrenado para similitud.",
+        ),
+    )  # type: ignore
+
+
+class RoBERTaEmbedding(DenseEmbedding):
+    SCHEMA = RoBERTaEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:roberta", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="RoBERTa Embedding",
+        es="Embedding RoBERTa",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using RoBERTa / XLM-RoBERTa models with mean/max pooling.",
+        es="Embeddings densos usando modelos RoBERTa / XLM-RoBERTa con pooling mean/max.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        pooling_strategy = self.params["pooling_strategy"]
+        model_info = ROBERTA_MODELS[model_name]
+        self._embedding = _BERTEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            pooling_strategy=pooling_strategy,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense/sentence_transformer_embedding.py b/DashAI/back/models/RAG/embeddings/dense/sentence_transformer_embedding.py
new file mode 100644
index 000000000..3e0cfa432
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense/sentence_transformer_embedding.py
@@ -0,0 +1,266 @@
+from typing import Dict, List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    bool_field,
+    enum_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense._overflow_handler import (
+    AGGREGATE,
+    TRUNCATE,
+)
+from DashAI.back.models.RAG.embeddings.dense._sentence_transformer_embedding import (
+    _SentenceTransformerEmbedding,
+)
+from DashAI.back.models.RAG.embeddings.dense_embedding import DenseEmbedding
+
+ST_MODELS: Dict[str, dict] = {
+    "microsoft/harrier-oss-v1-270m": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "last_token",
+    },
+    "microsoft/harrier-oss-v1-0.6b": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "last_token",
+    },
+    "microsoft/harrier-oss-v1-27b": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "last_token",
+    },
+    "Qwen/Qwen3-Embedding-0.6B": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "mean",
+    },
+    "Qwen/Qwen3-Embedding-4B": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "mean",
+    },
+    "Qwen/Qwen3-Embedding-8B": {
+        "languages": ["multi"],
+        "max_seq_length": 32768,
+        "pooling": "mean",
+    },
+    "sentence-transformers/all-MiniLM-L6-v2": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/all-MiniLM-L12-v2": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/all-mpnet-base-v2": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/all-distilroberta-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/multi-qa-mpnet-base-dot-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+        "normalize_default": False,
+    },
+    "sentence-transformers/multi-qa-mpnet-base-cos-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/multi-qa-distilbert-dot-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+        "normalize_default": False,
+    },
+    "sentence-transformers/multi-qa-distilbert-cos-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/multi-qa-MiniLM-L6-dot-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+        "normalize_default": False,
+    },
+    "sentence-transformers/multi-qa-MiniLM-L6-cos-v1": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/msmarco-bert-base-dot-v5": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+        "normalize_default": False,
+    },
+    "sentence-transformers/msmarco-distilbert-dot-v5": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+        "normalize_default": False,
+    },
+    "sentence-transformers/msmarco-distilbert-base-tas-b": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/msmarco-distilbert-cos-v5": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/msmarco-MiniLM-L12-cos-v5": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/msmarco-MiniLM-L6-cos-v5": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/paraphrase-multilingual-mpnet-base-v2": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/distiluse-base-multilingual-cased-v2": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "pt", "nl", "pl", "ca", "fi",
+            "ar", "zh", "ja", "ko", "ru", "tr", "hi", "sv", "da", "no",
+            "cs", "ro", "el", "he", "hu", "th", "vi", "id", "ms", "bg",
+            "hr", "sk", "sl", "sr", "uk", "et", "lv", "lt", "fa", "ur",
+            "mk", "af", "bn", "multi",
+        ],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/distiluse-base-multilingual-cased-v1": {
+        "languages": [
+            "en", "es", "fr", "de", "it", "nl", "pt", "ar", "zh", "ja",
+            "ko", "pl", "ru", "tr", "multi",
+        ],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+    "sentence-transformers/allenai-specter": {
+        "languages": ["en"],
+        "max_seq_length": 512,
+        "pooling": "mean",
+    },
+}
+
+ST_MODEL_NAMES = list(ST_MODELS.keys())
+
+
+class SentenceTransformerEmbeddingSchema(BaseSchema):
+    model_name: schema_field(
+        enum_field(ST_MODEL_NAMES),
+        placeholder="microsoft/harrier-oss-v1-0.6b",
+        description=MultilingualString(
+            en="Sentence Transformer model for embedding generation.",
+            es="Modelo Sentence Transformer para generación de embeddings.",
+        ),
+    )  # type: ignore
+
+    overflow_strategy: schema_field(
+        enum_field([TRUNCATE, AGGREGATE]),
+        placeholder=TRUNCATE,
+        description=MultilingualString(
+            en="Strategy for chunks exceeding model max sequence length.",
+            es="Estrategia para fragmentos que exceden la longitud máxima del modelo.",
+        ),
+    )  # type: ignore
+
+    normalize: schema_field(
+        bool_field(),
+        placeholder=True,
+        description=MultilingualString(
+            en="Whether to L2-normalize the output embeddings.",
+            es="Si normalizar con L2 los embeddings de salida.",
+        ),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(["cpu", "cuda"]),
+        placeholder="cpu",
+        description=MultilingualString(
+            en="Device to run the model on.",
+            es="Dispositivo para ejecutar el modelo.",
+        ),
+    )  # type: ignore
+
+
+class SentenceTransformerEmbedding(DenseEmbedding):
+    SCHEMA = SentenceTransformerEmbeddingSchema
+    FLAGS: list[str] = ["FAMILY:sentence_transformer", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="Sentence Transformer Embedding",
+        es="Embedding Sentence Transformer",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense embeddings using Sentence Transformer models with mean pooling and L2 normalization.",
+        es="Embeddings densos usando modelos Sentence Transformer con mean pooling y normalización L2.",
+    )
+
+    def __init__(self, **kwargs):
+        self.params = self.validate_and_transform(kwargs)
+        model_name = self.params["model_name"]
+        device = self.params["device"]
+        normalize = self.params["normalize"]
+        overflow_strategy = self.params.get("overflow_strategy", "truncate")
+        model_info = ST_MODELS[model_name]
+        pooling = model_info.get("pooling", "mean")
+        self._embedding = _SentenceTransformerEmbedding(
+            model_name=model_name,
+            device=device,
+            model_max_length=model_info["max_seq_length"],
+            overflow_strategy=overflow_strategy,
+            normalize=normalize if model_info.get("normalize_default", True) else normalize,
+            pooling=pooling,
+        )
+
+    def load(self):
+        self._embedding.load()
+
+    def encode(self, text: str):
+        return self._embedding.encode(text)
+
+    def batch_encode(self, texts: List[str]):
+        return self._embedding.batch_encode(texts)
+
+    def save(self):
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/embeddings/dense_embedding.py b/DashAI/back/models/RAG/embeddings/dense_embedding.py
new file mode 100644
index 000000000..aee70f370
--- /dev/null
+++ b/DashAI/back/models/RAG/embeddings/dense_embedding.py
@@ -0,0 +1,46 @@
+from abc import abstractmethod
+from typing import Any, List
+
+import numpy as np
+
+from DashAI.back.models.base_model import BaseModel
+
+
+class DenseEmbedding(BaseModel):
+    """
+    Base class for all encoding (embedding) models.
+    This class should be inherited by any specific encoding (embedding) model implementation.
+    """
+
+    embedding_dim: int
+
+    def __init__(self, **kwargs: Any):
+        """
+        Initialize the encoding (embedding) model with the given documents.
+        """
+        raise NotImplementedError("Subclasses must implement the __init__ method.")
+
+    def encode(self, text: str) -> List[float] | np.ndarray:
+        """
+        Method to generate encodings (embeddings) for the given text.
+        This method should be implemented by subclasses.
+
+        :param text: The input text to encode (embed).
+        :return: A list representing the encodings (embeddings) of the input text.
+        """
+        raise NotImplementedError("Subclasses must implement this method.")
+
+    def batch_encode(self, texts: List[str]) -> List[List[float]] | np.ndarray:
+        """
+        Method to generate encodings (embeddings) for a batch of texts.
+        This method should be implemented by subclasses.
+
+        :param texts: A list of input texts to encode (embed).
+        :return: A list of lists representing the encodings (embeddings) of the input texts.
+        """
+        raise NotImplementedError("Subclasses must implement this method.")
+
+    @abstractmethod
+    def train(self, **kwargs):
+        """Train the embedding model."""
+        return
diff --git a/DashAI/back/models/RAG/embeddings/sparse/__init__.py b/DashAI/back/models/RAG/embeddings/sparse/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/DashAI/back/models/RAG/exceptions.py b/DashAI/back/models/RAG/exceptions.py
new file mode 100644
index 000000000..abdbb49ee
--- /dev/null
+++ b/DashAI/back/models/RAG/exceptions.py
@@ -0,0 +1,2 @@
+class RAGWorkflowError(Exception):
+    """Custom exception for RAG workflow errors."""
diff --git a/DashAI/back/models/RAG/llm_factory.py b/DashAI/back/models/RAG/llm_factory.py
new file mode 100644
index 000000000..169adb8bd
--- /dev/null
+++ b/DashAI/back/models/RAG/llm_factory.py
@@ -0,0 +1,62 @@
+"""Factory for text-to-text generation models (LLMs).
+
+Resolves a component name + parameters into an instantiated model
+via the component registry, with lookup-or-create DB semantics.
+"""
+
+from dataclasses import dataclass
+from typing import Any, Dict
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import (
+    RAGGenerationModel as GenerationDBModel,
+)
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+
+
+@dataclass(frozen=True)
+class LLMFactoryResult:
+    """Result of LLM instantiation via LLMFactory."""
+
+    db_record_id: int
+    model: TextToTextGenerationTaskModel
+
+
+class LLMFactory:
+    """Creates LLM instances with lookup-or-create DB semantics."""
+
+    def __init__(self, db: Session, component_registry: ComponentRegistry):
+        self._db = db
+        self._registry = component_registry
+
+    def create(self, component_name: str, params: Dict[str, Any]) -> LLMFactoryResult:
+        sorted_params = dict(sorted(params.items()))
+        existing = (
+            self._db.query(GenerationDBModel)
+            .filter_by(class_name=component_name, parameters=sorted_params)
+            .first()
+        )
+        if existing is not None:
+            model_class = self._registry[existing.class_name]["class"]
+            return LLMFactoryResult(
+                db_record_id=existing.id,
+                model=model_class(**existing.parameters),
+            )
+
+        db_record = GenerationDBModel(
+            class_name=component_name,
+            parameters=sorted_params,
+        )
+        self._db.add(db_record)
+        self._db.commit()
+        self._db.refresh(db_record)
+
+        model_class = self._registry[component_name]["class"]
+        return LLMFactoryResult(
+            db_record_id=db_record.id,
+            model=model_class(**params),
+        )
diff --git a/DashAI/back/models/RAG/pipeline_repository.py b/DashAI/back/models/RAG/pipeline_repository.py
new file mode 100644
index 000000000..7a2bf26ed
--- /dev/null
+++ b/DashAI/back/models/RAG/pipeline_repository.py
@@ -0,0 +1,62 @@
+"""Persistence layer for the RAGPipeline DB record.
+
+Keeps DB access out of the pipeline's __init__ so the pipeline can
+focus on orchestration.
+"""
+
+from typing import Optional
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import RAGPipeline as PipelineDBModel
+
+
+class PipelineRepository:
+    """Manages the rag_pipeline DB record for a given session."""
+
+    def __init__(self, db: Session):
+        self._db = db
+
+    def ensure_db_record(self, session_id: int) -> PipelineDBModel:
+        """Return the existing pipeline record or create a placeholder.
+
+        The placeholder has filler values for the FK columns; call
+        ``update_db_record`` once the real component IDs are known.
+        """
+        existing: Optional[PipelineDBModel] = (
+            self._db.query(PipelineDBModel).filter_by(session_id=session_id).first()
+        )
+        if existing is not None:
+            return existing
+
+        record = PipelineDBModel(
+            session_id=session_id,
+            name="",
+            description=None,
+            parameters=None,
+            chunking_model_id=0,
+            prompt_id=0,
+            generation_model_id=0,
+        )
+        self._db.add(record)
+        self._db.commit()
+        self._db.refresh(record)
+        return record
+
+    def update_db_record(
+        self,
+        session_id: int,
+        chunking_model_id: int,
+        prompt_id: int,
+        generation_model_id: int,
+    ) -> None:
+        """Patch the pipeline record with real component IDs."""
+        pipeline_record: Optional[PipelineDBModel] = (
+            self._db.query(PipelineDBModel).filter_by(session_id=session_id).first()
+        )
+        if pipeline_record is None:
+            raise ValueError(f"No pipeline record for session {session_id}")
+        pipeline_record.chunking_model_id = chunking_model_id
+        pipeline_record.prompt_id = prompt_id
+        pipeline_record.generation_model_id = generation_model_id
+        self._db.commit()
diff --git a/DashAI/back/models/RAG/prompts/__init__.py b/DashAI/back/models/RAG/prompts/__init__.py
new file mode 100644
index 000000000..a2fdf1136
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/__init__.py
@@ -0,0 +1,12 @@
+from DashAI.back.models.RAG.prompts.prompt import Prompt
+from DashAI.back.models.RAG.prompts.augmentation import (
+    AugmentationPrompt,
+    DefaultAugmentationPrompt,
+    CustomAugmentationPrompt
+)
+from DashAI.back.models.RAG.prompts.generation import (
+    RAGGenerationPrompt,
+    CustomRAGGenerationPrompt,
+    DefaultRAGGenerationPrompt,
+    DefaultQARAGGenerationPrompt
+)
diff --git a/DashAI/back/models/RAG/prompts/augmentation/__init__.py b/DashAI/back/models/RAG/prompts/augmentation/__init__.py
new file mode 100644
index 000000000..e5a807072
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/augmentation/__init__.py
@@ -0,0 +1,3 @@
+from DashAI.back.models.RAG.prompts.augmentation.augmentation_prompt import AugmentationPrompt
+from DashAI.back.models.RAG.prompts.augmentation.custom_augmentation_prompt import CustomAugmentationPrompt
+from DashAI.back.models.RAG.prompts.augmentation.default_augmentation_prompt import DefaultAugmentationPrompt
diff --git a/DashAI/back/models/RAG/prompts/augmentation/augmentation_prompt.py b/DashAI/back/models/RAG/prompts/augmentation/augmentation_prompt.py
new file mode 100644
index 000000000..895c4e03d
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/augmentation/augmentation_prompt.py
@@ -0,0 +1,29 @@
+from typing import Any
+
+from DashAI.back.models.RAG.prompts.prompt import Prompt
+
+
+class AugmentationPrompt(Prompt):
+    """
+    AugmentationPrompt class for generating augmented retrieval prompts,
+    it uses the language model to generate keywords or phrases that can be used
+    to augment the input.
+    """
+
+    required_placeholders = ["{input}", "{n_search_terms}"]
+    optional_placeholders = []
+
+    def __init__(self, **kwargs: Any):
+        self.template = kwargs.pop("template")
+
+    def format(self, input: str, n_search_terms: int, **kwargs: Any) -> str:
+        """
+        Instantiate and format the prompt for augmentation.
+        Args:
+            input (str): The input to be formatted.
+            n_search_terms (int): The number of search terms to generate.
+            **kwargs: Additional keyword arguments for formatting.
+        Returns:
+            str: The formatted prompt.
+        """
+        raise NotImplementedError("Subclasses must implement this method.")
diff --git a/DashAI/back/models/RAG/prompts/augmentation/custom_augmentation_prompt.py b/DashAI/back/models/RAG/prompts/augmentation/custom_augmentation_prompt.py
new file mode 100644
index 000000000..4b1d8dfdb
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/augmentation/custom_augmentation_prompt.py
@@ -0,0 +1,51 @@
+from typing import Any
+
+from DashAI.back.models.RAG.prompts.augmentation import AugmentationPrompt
+
+
+class CustomAugmentationPrompt(AugmentationPrompt):
+    """
+    User-defined augmentation prompt template for generating augmented retrieval
+    prompts.
+    It uses the language model to generate keywords or phrases that can be used
+    to augment the input.
+    """
+
+    metadata = {
+        "name": "Custom Augmentation Prompt",
+        "description": "User-defined prompt template for generating augmented "
+        "retrieval prompts.",
+        "type": "augmentation",
+        "required_placeholders": AugmentationPrompt.required_placeholders,
+        "optional_placeholders": AugmentationPrompt.optional_placeholders,
+        "placeholder_descriptions": {
+            "{input}": "The user input message.",
+            "{n_search_terms}": "The number of search terms to generate.",
+        },
+    }
+
+    required_placeholders = ["{input}", "{n_search_terms}"]
+    optional_placeholders = []
+
+    def __init__(self, **kwargs):
+        self.template = kwargs.pop("template")
+
+    def format(
+        self,
+        input: str,
+        n_search_terms: int,
+        **kwargs: Any,
+    ) -> str:
+        """
+        Instantiate and format the prompt for augmentation.
+        Args:
+            input (str): The input to be formatted.
+            n_search_terms (int): The number of search terms to generate.
+            **kwargs: Additional keyword arguments for formatting.
+        Returns:
+            str: The formatted prompt.
+        """
+        buffer = self.template
+        buffer = buffer.replace("{input}", input)
+        buffer = buffer.replace("{n_search_terms}", str(n_search_terms))
+        return buffer
diff --git a/DashAI/back/models/RAG/prompts/augmentation/default_augmentation_prompt.py b/DashAI/back/models/RAG/prompts/augmentation/default_augmentation_prompt.py
new file mode 100644
index 000000000..91b53d8a8
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/augmentation/default_augmentation_prompt.py
@@ -0,0 +1,153 @@
+from typing import Any
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.prompts.augmentation.augmentation_prompt import (
+    AugmentationPrompt,
+)
+
+TEMPLATES = {
+    "en": (
+        "You are an intelligent and insightful assistant. Your task is to generate "
+        "keywords or phrases to search for relevant information based on the input "
+        "provided.\n"
+        "The keywords or phrases should be relevant to the input and should help in "
+        "retrieving useful information to improve the precision of the response.\n"
+        "\n"
+        "User input:\n"
+        "{input}\n"
+        "\n"
+        "Generate {n_search_terms} keywords or phrases that can be used to search for "
+        "relevant information. The keywords or phrases should be concise and to the "
+        "point.\n"
+        "\n"
+        "You must respond with a JSON object following this exact format:\n"
+        "{'keywords': ['keyword_1', 'keyword_2', ..., 'keyword_n']}"
+    ),
+    "es": (
+        "Eres un asistente inteligente y perspicaz. Tu tarea es generar palabras clave "
+        "o frases para buscar información relevante basada en la entrada "
+        "proporcionada.\n"
+        "Las palabras clave o frases deben ser relevantes para la entrada y ayudar a "
+        "recuperar información útil para mejorar la precisión de la respuesta.\n"
+        "\n"
+        "Entrada del usuario:\n"
+        "{input}\n"
+        "\n"
+        "Genera {n_search_terms} palabras clave o frases que puedan usarse para buscar "
+        "información relevante. Las palabras clave o frases deben ser concisas y "
+        "directas.\n"
+        "\n"
+        "Debes responder con un objeto JSON siguiendo este formato exacto:\n"
+        "{'keywords': ['palabra_clave_1', 'palabra_clave_2', ..., 'palabra_clave_n']}"
+    ),
+    "pt": (
+        "Você é um assistente inteligente e perspicaz. Sua tarefa é gerar "
+        "palavras-chave ou frases para buscar informações relevantes com base na "
+        "entrada fornecida.\n"
+        "As palavras-chave ou frases devem ser relevantes para a entrada e ajudar a "
+        "recuperar informações úteis para melhorar a precisão da resposta.\n"
+        "\n"
+        "Entrada do usuário:\n"
+        "{input}\n"
+        "\n"
+        "Gere {n_search_terms} palavras-chave ou frases que possam ser usadas para "
+        "buscar informações relevantes. As palavras-chave ou frases devem ser concisas "
+        "e diretas.\n"
+        "\n"
+        "Você deve responder com um objeto JSON seguindo este formato exato:\n"
+        "{'keywords': ['palavra_chave_1', 'palavra_chave_2', ..., 'palavra_chave_n']}"
+    ),
+}
+
+
+class DefaultAugmentationPromptSchema(BaseSchema):
+    language: schema_field(
+        enum_field(enum=["en", "es", "pt"]),
+        placeholder="en",
+        description=MultilingualString(
+            en="Language for the generated response.",
+            es="Idioma de la respuesta generada.",
+            pt="Idioma da resposta gerada.",
+        ),
+    )
+    template: schema_field(
+        string_field(),
+        placeholder="",
+        description="The prompt template with placeholders.",
+    )
+
+
+class DefaultAugmentationPrompt(AugmentationPrompt):
+    """
+    AugmentationPrompt class for generating augmented retrieval prompts,
+    it uses the language model to generate keywords or phrases that can be used
+    to augment the input.
+    """
+
+    SCHEMA = DefaultAugmentationPromptSchema
+    DESCRIPTION: str = MultilingualString(
+        en="Default prompt template for generating augmented retrieval queries.",
+        es="Plantilla de prompt predeterminada para generar consultas de "
+        "recuperación aumentadas.",
+        pt="Modelo de prompt padrão para gerar consultas de recuperação aumentadas.",
+    )
+    DISPLAY_NAME: str = MultilingualString(
+        en="Default Augmentation Prompt",
+        es="Prompt de Aumento Predeterminado",
+        pt="Prompt de Aumento Padrão",
+    )
+
+    required_placeholders = ["{input}", "{n_search_terms}"]
+    optional_placeholders: list[str] = []
+
+    metadata = {
+        "name": "Default Augmentation Prompt",
+        "description": "Default prompt template for generating augmented retrieval "
+        "prompts.",
+        "type": "augmentation",
+        "required_placeholders": required_placeholders,
+        "optional_placeholders": optional_placeholders,
+        "placeholder_descriptions": {
+            "{input}": "The user input message.",
+            "{n_search_terms}": "The number of search terms to generate.",
+        },
+        "templates": TEMPLATES,
+    }
+
+    def __init__(self, **kwargs: Any) -> None:
+        """Initialize the augmentation prompt with a language and resolved template.
+
+        Args:
+            language: Language code to select the appropriate template
+                (one of "en", "es", "pt").
+            template: Override template string. If not provided, the
+                default template for the selected language is used.
+        """
+        self.language = kwargs.pop("language")
+        self.template = kwargs.pop("template") or TEMPLATES.get(self.language, "")
+
+    def format(
+        self,
+        input: str,
+        n_search_terms: int,
+        **kwargs: Any,
+    ) -> str:
+        """Render the augmentation prompt by replacing all placeholders.
+
+        Args:
+            input: The user's input message.
+            n_search_terms: Number of search terms the LLM should generate.
+
+        Returns:
+            Fully formatted prompt string ready for LLM input.
+        """
+        buffer = self.template
+        buffer = buffer.replace("{input}", input)
+        buffer = buffer.replace("{n_search_terms}", str(n_search_terms))
+        return buffer
diff --git a/DashAI/back/models/RAG/prompts/generation/__init__.py b/DashAI/back/models/RAG/prompts/generation/__init__.py
new file mode 100644
index 000000000..2867bb7a7
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/generation/__init__.py
@@ -0,0 +1,4 @@
+from DashAI.back.models.RAG.prompts.generation.rag_generation_prompt import RAGGenerationPrompt
+from DashAI.back.models.RAG.prompts.generation.custom_rag_generation_prompt import CustomRAGGenerationPrompt
+from DashAI.back.models.RAG.prompts.generation.default_rag_generation_prompt import DefaultRAGGenerationPrompt
+from DashAI.back.models.RAG.prompts.generation.default_QA_rag_generation_prompt import DefaultQARAGGenerationPrompt
diff --git a/DashAI/back/models/RAG/prompts/generation/custom_rag_generation_prompt.py b/DashAI/back/models/RAG/prompts/generation/custom_rag_generation_prompt.py
new file mode 100644
index 000000000..a540f5e84
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/generation/custom_rag_generation_prompt.py
@@ -0,0 +1,44 @@
+from typing import Any
+
+from DashAI.back.models.RAG.prompts.generation.rag_generation_prompt import (
+    RAGGenerationPrompt,
+)
+
+
+class CustomRAGGenerationPrompt(RAGGenerationPrompt):
+    """
+    CustomRAGGenerationPrompt class for user-defined prompt templates used in the language generation step of RAG.
+    """
+
+    metadata = {
+        "name": "Custom RAG Generation Prompt",
+        "description": "User-defined prompt template used in the language generation step of RAG.",
+        "type": "generation",
+        "required_placeholders": RAGGenerationPrompt.required_placeholders,
+        "optional_placeholders": RAGGenerationPrompt.optional_placeholders,
+        "placeholder_descriptions": {
+            "{input}": "The user input message.",
+            "{chunks}": "The document chunks to be included in the context.",
+        },
+    }
+
+    def __init__(self, **kwargs: Any):
+        self.template = kwargs.pop("template")
+
+    def format(self, input: str, chunks: str, **kwargs: Any) -> str:
+        """
+        Format the prompt using the provided template.
+
+        Args:
+            input (str): The user input message.
+            history (List[Tuple[str, str]]): The chat history to be included in the context.
+            chunks (List[str]): The document chunks to be included in the context.
+            **kwargs: Additional keyword arguments for formatting.
+
+        Returns:
+            str: The formatted prompt.
+        """
+        buffer = self.template
+        buffer = buffer.replace("{input}", input)
+        buffer = buffer.replace("{chunks}", chunks)
+        return buffer
diff --git a/DashAI/back/models/RAG/prompts/generation/default_QA_rag_generation_prompt.py b/DashAI/back/models/RAG/prompts/generation/default_QA_rag_generation_prompt.py
new file mode 100644
index 000000000..80f5e4872
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/generation/default_QA_rag_generation_prompt.py
@@ -0,0 +1,129 @@
+from typing import Any
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.prompts.generation.rag_generation_prompt import (
+    RAGGenerationPrompt,
+)
+
+TEMPLATES = {
+    "en": (
+        "You are a question-answering assistant with access to a set of "
+        "retrieved documents.\n"
+        "Your task is to answer the user's question using ONLY the "
+        "information found in the documents below. If the documents do "
+        "not contain enough information to answer the question, respond "
+        'with "I don\'t know" and explain what information is missing.\n'
+        "Cite relevant passages from the documents to support your answer.\n"
+        "\n"
+        "User question:\n"
+        "{input}\n"
+        "\n"
+        "Retrieved documents:\n"
+        "{chunks}\n"
+        "\n"
+        "Answer:"
+    ),
+    "es": (
+        "Eres un asistente de preguntas y respuestas con acceso a un "
+        "conjunto de documentos recuperados.\n"
+        "Tu tarea es responder la pregunta del usuario usando SOLAMENTE "
+        "la información encontrada en los documentos a continuación. Si "
+        "los documentos no contienen suficiente información para responder "
+        'la pregunta, responde con "No lo sé" y explica qué información '
+        "falta.\n"
+        "Cita pasajes relevantes de los documentos para respaldar tu "
+        "respuesta.\n"
+        "\n"
+        "Pregunta del usuario:\n"
+        "{input}\n"
+        "\n"
+        "Documentos recuperados:\n"
+        "{chunks}\n"
+        "\n"
+        "Respuesta:"
+    ),
+    "pt": (
+        "Você é um assistente de perguntas e respostas com acesso a um "
+        "conjunto de documentos recuperados.\n"
+        "Sua tarefa é responder à pergunta do usuário usando SOMENTE as "
+        "informações encontradas nos documentos abaixo. Se os documentos "
+        "não contiverem informações suficientes para responder à pergunta, "
+        'responda com "Não sei" e explique quais informações estão '
+        "faltando.\n"
+        "Cite passagens relevantes dos documentos para apoiar sua "
+        "resposta.\n"
+        "\n"
+        "Pergunta do usuário:\n"
+        "{input}\n"
+        "\n"
+        "Documentos recuperados:\n"
+        "{chunks}\n"
+        "\n"
+        "Resposta:"
+    ),
+}
+
+
+class DefaultQARAGGenerationPromptSchema(BaseSchema):
+    language: schema_field(
+        enum_field(enum=["en", "es", "pt"]),
+        placeholder="en",
+        description=MultilingualString(
+            en="Language for the generated response.",
+            es="Idioma de la respuesta generada.",
+            pt="Idioma da resposta gerada.",
+        ),
+    )
+    template: schema_field(
+        string_field(),
+        placeholder="",
+        description="The prompt template with placeholders.",
+    )
+
+
+class DefaultQARAGGenerationPrompt(RAGGenerationPrompt):
+    """
+    Default generation prompt for Question Answering tasks.
+    This prompt is designed to guide the language model in generating answers based on provided context chunks.
+    """
+
+    SCHEMA = DefaultQARAGGenerationPromptSchema
+    DESCRIPTION: str = MultilingualString(
+        en="Default prompt template used in the language generation step of RAG for Question Answering tasks.",
+        es="Plantilla de prompt predeterminada utilizada en el paso de generación de lenguaje de RAG para tareas de preguntas y respuestas.",
+        pt="Modelo de prompt padrão usado na etapa de geração de linguagem do RAG para tarefas de perguntas e respostas.",
+    )
+    DISPLAY_NAME: str = MultilingualString(
+        en="Default QA RAG Generation Prompt",
+        es="Prompt de Generación RAG de Preguntas y Respuestas Predeterminado",
+        pt="Prompt de Geração RAG de Perguntas e Respostas Padrão",
+    )
+
+    metadata = {
+        "name": "Default QA RAG Generation Prompt",
+        "description": "Default prompt template used in the language generation step of RAG for Question Answering tasks.",
+        "type": "generation",
+        "required_placeholders": RAGGenerationPrompt.required_placeholders,
+        "optional_placeholders": RAGGenerationPrompt.optional_placeholders,
+        "placeholder_descriptions": {
+            "{input}": "The user input message.",
+            "{chunks}": "The document chunks to be included in the context.",
+        },
+        "templates": TEMPLATES,
+    }
+
+    def __init__(self, **kwargs):
+        self.language = kwargs.pop("language")
+        self.template = kwargs.pop("template") or TEMPLATES.get(self.language, "")
+
+    def format(self, input: str, chunks: str, **kwargs: Any) -> str:
+        buffer = self.template
+        buffer = buffer.replace("{input}", input)
+        buffer = buffer.replace("{chunks}", chunks)
+        return buffer
diff --git a/DashAI/back/models/RAG/prompts/generation/default_rag_generation_prompt.py b/DashAI/back/models/RAG/prompts/generation/default_rag_generation_prompt.py
new file mode 100644
index 000000000..37e5b4c94
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/generation/default_rag_generation_prompt.py
@@ -0,0 +1,125 @@
+from typing import Any
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.prompts.generation.rag_generation_prompt import (
+    RAGGenerationPrompt,
+)
+
+TEMPLATES = {
+    "en": (
+        "You are a helpful AI assistant with access to a set of retrieved "
+        "documents.\n"
+        "Use the documents below to support your answer. Prioritize "
+        "information from the documents over your own knowledge. "
+        "If the documents do not contain relevant information, say so "
+        "clearly rather than guessing.\n"
+        "\n"
+        "User message:\n"
+        "{input}\n"
+        "\n"
+        "Retrieved documents:\n"
+        "{chunks}\n"
+        "\n"
+        "Answer the user's message based on the documents above."
+    ),
+    "es": (
+        "Eres un asistente de IA con acceso a un conjunto de documentos "
+        "recuperados.\n"
+        "Usa los documentos a continuación para fundamentar tu respuesta. "
+        "Prioriza la información de los documentos sobre tu propio "
+        "conocimiento. Si los documentos no contienen información relevante, "
+        "indícalo claramente en lugar de adivinar.\n"
+        "\n"
+        "Mensaje del usuario:\n"
+        "{input}\n"
+        "\n"
+        "Documentos recuperados:\n"
+        "{chunks}\n"
+        "\n"
+        "Responde al mensaje del usuario basándote en los documentos "
+        "anteriores."
+    ),
+    "pt": (
+        "Você é um assistente de IA com acesso a um conjunto de documentos "
+        "recuperados.\n"
+        "Use os documentos abaixo para fundamentar sua resposta. Priorize "
+        "as informações dos documentos sobre seu próprio conhecimento. Se "
+        "os documentos não contiverem informações relevantes, indique isso "
+        "claramente em vez de adivinhar.\n"
+        "\n"
+        "Mensagem do usuário:\n"
+        "{input}\n"
+        "\n"
+        "Documentos recuperados:\n"
+        "{chunks}\n"
+        "\n"
+        "Responda à mensagem do usuário com base nos documentos acima."
+    ),
+}
+
+
+class DefaultRAGGenerationPromptSchema(BaseSchema):
+    language: schema_field(
+        enum_field(enum=["en", "es", "pt"]),
+        placeholder="en",
+        description=MultilingualString(
+            en="Language for the generated response.",
+            es="Idioma de la respuesta generada.",
+            pt="Idioma da resposta gerada.",
+        ),
+    )
+    template: schema_field(
+        string_field(),
+        placeholder="",
+        description="The prompt template with placeholders.",
+    )
+
+
+class DefaultRAGGenerationPrompt(RAGGenerationPrompt):
+    """
+    Default prompt template used in the language generation step of RAG.
+    """
+
+    SCHEMA = DefaultRAGGenerationPromptSchema
+    DESCRIPTION: str = MultilingualString(
+        en="Default prompt template used in the language generation step of RAG.",
+        es="Plantilla de prompt predeterminada utilizada en el paso de generación de lenguaje de RAG.",
+        pt="Modelo de prompt padrão usado na etapa de geração de linguagem do RAG.",
+    )
+    DISPLAY_NAME: str = MultilingualString(
+        en="Default RAG Generation Prompt",
+        es="Prompt de Generación RAG Predeterminado",
+        pt="Prompt de Geração RAG Padrão",
+    )
+
+    metadata = {
+        "name": "Default RAG Generation Prompt",
+        "description": "Default prompt template used in the language generation step of RAG.",
+        "type": "generation",
+        "required_placeholders": RAGGenerationPrompt.required_placeholders,
+        "optional_placeholders": RAGGenerationPrompt.optional_placeholders,
+        "placeholder_descriptions": {
+            "{input}": "The user input message.",
+            "{chunks}": "The document chunks to be included in the context.",
+        },
+        "templates": TEMPLATES,
+    }
+
+    required_placeholders = ["{input}", "{chunks}"]
+    optional_placeholders = []
+
+    def __init__(self, **kwargs):
+        self.language = kwargs.pop("language")
+        self.template = kwargs.pop("template") or TEMPLATES.get(self.language, "")
+
+    def format(self, input: str, chunks: str, **kwargs: Any) -> str:
+        buffer = self.template
+        buffer = buffer.replace("{input}", input)
+        buffer = buffer.replace("{chunks}", chunks)
+        return buffer
diff --git a/DashAI/back/models/RAG/prompts/generation/rag_generation_prompt.py b/DashAI/back/models/RAG/prompts/generation/rag_generation_prompt.py
new file mode 100644
index 000000000..3e1e0aaa2
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/generation/rag_generation_prompt.py
@@ -0,0 +1,21 @@
+from typing import Any, Tuple
+
+from DashAI.back.models.RAG.prompts.prompt import Prompt
+
+
+class RAGGenerationPrompt(Prompt):
+    """
+    RAGGenerationPrompt class for formatting prompts used in the language generation step of RAG.
+    """
+
+    required_placeholders = ["{input}", "{chunks}"]
+    optional_placeholders = []
+
+    @staticmethod
+    def format(
+        input: str,
+        chunks: str,
+        history: Tuple[Tuple[str, str], ...] | None = None,
+        **kwargs: Any,
+    ) -> str:
+        raise NotImplementedError("Subclasses must implement this method.")
diff --git a/DashAI/back/models/RAG/prompts/prompt.py b/DashAI/back/models/RAG/prompts/prompt.py
new file mode 100644
index 000000000..f01308d14
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/prompt.py
@@ -0,0 +1,86 @@
+from typing import Any, Dict, List
+
+from DashAI.back.core.schema_fields import BaseSchema, schema_field, string_field
+from DashAI.back.models.base_model import BaseModel
+
+
+class PromptSchema(BaseSchema):
+    template: str = schema_field(
+        string_field(),
+        placeholder="",
+        description="The prompt template with placeholders.",
+    )
+
+
+class Prompt(BaseModel):
+    """
+    Base class for all RAG prompt templates.
+    This class defines the interface for creating and formatting prompts.
+    """
+
+    SCHEMA = PromptSchema
+    DESCRIPTION: str = "Base class for RAG prompts."
+    DISPLAY_NAME: str = "Base RAG Prompt"
+    id: int
+    name: str
+    REQUIRED_EXTRA_KWARGS = []
+
+    def load(self, **kwargs: Any) -> None:
+        pass
+
+    def save(self) -> None:
+        pass
+
+    def train(self, **kwargs: Any) -> None:
+        pass
+
+    @classmethod
+    def get_metadata(cls) -> Dict[str, Any]:
+        metadata = cls.metadata if hasattr(cls, "metadata") else {}
+        return metadata
+
+    @classmethod
+    def get_required_placeholders(cls) -> List[str]:
+        """
+        Get the list of required placeholders for the prompt template.
+        Returns:
+            List[str]: List of required placeholders.
+        """
+        assert hasattr(cls, "required_placeholders"), (
+            "Subclasses must define 'required_placeholders' class attribute."
+        )
+        return cls.required_placeholders
+
+    def get_optional_placeholders(self) -> List[str]:
+        """
+        Get the list of optional placeholders for the prompt template.
+        Returns:
+            List[str]: List of optional placeholders.
+        """
+        assert hasattr(self, "optional_placeholders"), (
+            "Subclasses must define 'optional_placeholders' class attribute."
+        )
+        return self.optional_placeholders
+
+    @classmethod
+    def validate_template(cls, template: str) -> bool:
+        """
+        Validate that the template contains all required placeholders.
+        Args:
+            template (str): The prompt template to be validated.
+        Returns:
+            bool: True if the template is valid, False otherwise.
+        """
+        return all(placeholder in template for placeholder in cls.required_placeholders)
+
+    @staticmethod
+    def format(input: str, **kwargs: Any) -> str:
+        """
+        Instantiate and format the prompt.
+        Args:
+            input (str): The input to be formatted.
+            **kwargs: Additional keyword arguments for formatting.
+        Returns:
+            str: The formatted prompt.
+        """
+        raise NotImplementedError("Subclasses must implement this method.")
diff --git a/DashAI/back/models/RAG/prompts/prompt_factory.py b/DashAI/back/models/RAG/prompts/prompt_factory.py
new file mode 100644
index 000000000..5d292ed78
--- /dev/null
+++ b/DashAI/back/models/RAG/prompts/prompt_factory.py
@@ -0,0 +1,60 @@
+"""Factory for prompt templates with lookup-or-create semantics.
+
+Resolves a component name + parameters into a Prompt instance,
+creating a RAGPrompt DB record when no matching record exists.
+"""
+
+from dataclasses import dataclass
+from typing import Any, Dict
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import RAGPrompt as PromptDBModel
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.RAG.prompts import Prompt
+
+
+@dataclass(frozen=True)
+class PromptFactoryResult:
+    """Result of prompt instantiation via PromptFactory."""
+
+    db_record_id: int
+    model: Prompt
+
+
+class PromptFactory:
+    """Creates Prompt instances with automatic DB record reuse."""
+
+    def __init__(self, db: Session, component_registry: ComponentRegistry):
+        self._db = db
+        self._registry = component_registry
+
+    def create(
+        self, component_name: str, params: Dict[str, Any]
+    ) -> PromptFactoryResult:
+        sorted_params: Dict[str, Any] = dict(sorted(params.items()))
+        existing: PromptDBModel | None = (
+            self._db.query(PromptDBModel)
+            .filter_by(class_name=component_name, parameters=sorted_params)
+            .first()
+        )
+        if existing is not None:
+            prompt_class = self._registry[existing.class_name]["class"]
+            return PromptFactoryResult(
+                db_record_id=existing.id,
+                model=prompt_class(**existing.parameters),
+            )
+
+        db_record = PromptDBModel(
+            class_name=component_name,
+            parameters=sorted_params,
+        )
+        self._db.add(db_record)
+        self._db.commit()
+        self._db.refresh(db_record)
+
+        prompt_class = self._registry[component_name]["class"]
+        return PromptFactoryResult(
+            db_record_id=db_record.id,
+            model=prompt_class(**params),
+        )
diff --git a/DashAI/back/models/RAG/rag_models_factory.py b/DashAI/back/models/RAG/rag_models_factory.py
new file mode 100644
index 000000000..46470ec49
--- /dev/null
+++ b/DashAI/back/models/RAG/rag_models_factory.py
@@ -0,0 +1,105 @@
+"""Abstract Factory (GoF) for RAG component creation.
+
+RAGModelsFactory provides a unified interface for creating the four
+RAG component types: prompts, chunking models, retrievers, and LLMs.
+
+Each ``create_*`` method delegates to a specialised sub-factory that
+encapsulates the component's full lifecycle (DB-record resolution,
+instantiation, and persistence).
+"""
+
+from typing import Any, Dict
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.RAG.chunking_models.chunking_model_factory import (
+    ChunkingFactoryResult,
+    ChunkingModelFactory,
+)
+from DashAI.back.models.RAG.documents.base_document import BaseDocument
+from DashAI.back.models.RAG.documents.chunk import Chunk
+from DashAI.back.models.RAG.llm_factory import (
+    LLMFactory,
+    LLMFactoryResult,
+)
+from DashAI.back.models.RAG.prompts.prompt_factory import (
+    PromptFactory,
+    PromptFactoryResult,
+)
+from DashAI.back.models.RAG.retrievers.retriever_factory import (
+    RetrieverFactory,
+    RetrieverFactoryResult,
+)
+
+
+class RAGModelsFactory:
+    """Abstract Factory for the family of RAG components.
+
+    Provides four ``create_*`` methods, each delegating to a
+    specialised sub-factory. Shared dependencies (db, registry,
+    env_rag_path) are injected once via the constructor.
+    """
+
+    def __init__(
+        self,
+        db: Session,
+        registry: ComponentRegistry,
+        env_rag_path: str,
+    ):
+        self._db = db
+        self._registry = registry
+        self._env_rag_path = env_rag_path
+
+    def create_prompt(
+        self,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> PromptFactoryResult:
+        """Create a prompt with lookup-or-create semantics."""
+        factory = PromptFactory(self._db, self._registry)
+        return factory.create(component_name, params)
+
+    def create_chunking_model(
+        self,
+        documents: Dict[int, BaseDocument],
+        chunk_set_id: int,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> ChunkingFactoryResult:
+        """Create a chunking model, chunk documents, and persist chunks."""
+        factory = ChunkingModelFactory(
+            self._db,
+            self._registry,
+            documents,
+            chunk_set_id,
+        )
+        return factory.create(component_name, params)
+
+    def create_retriever(
+        self,
+        pipeline_id: int,
+        chunks: Dict[int, Dict[int, Chunk]],
+        chunk_set_id: int,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> RetrieverFactoryResult:
+        """Create a retriever with full persistence lifecycle."""
+        factory = RetrieverFactory(
+            self._db,
+            pipeline_id,
+            self._registry,
+            self._env_rag_path,
+            chunks,
+            chunk_set_id,
+        )
+        return factory.create(component_name, params)
+
+    def create_llm(
+        self,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> LLMFactoryResult:
+        """Create an LLM with lookup-or-create DB semantics."""
+        factory = LLMFactory(self._db, self._registry)
+        return factory.create(component_name, params)
diff --git a/DashAI/back/models/RAG/retrievers/__init__.py b/DashAI/back/models/RAG/retrievers/__init__.py
new file mode 100644
index 000000000..929f5addb
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/__init__.py
@@ -0,0 +1,32 @@
+from DashAI.back.models.RAG.retrievers.composite import (
+    CompositeRetriever,
+    MMRRerankerRetriever,
+    ParallelRetriever,
+    SequentialRetriever,
+)
+from DashAI.back.models.RAG.retrievers.dense import (
+    DenseEmbeddingRetriever,
+    DenseRetriever,
+    HuggingFaceDenseRetriever,
+)
+from DashAI.back.models.RAG.retrievers.enums import MergeStrategy
+from DashAI.back.models.RAG.retrievers.exceptions import (
+    CompositeValidationError,
+    ExtraKwargsMissingError,
+    MissingParameterError,
+    RetrieverError,
+    UnitRetrieverChildError,
+)
+from DashAI.back.models.RAG.retrievers.retriever_factory import (
+    RetrieverFactory,
+    RetrieverFactoryResult,
+)
+from DashAI.back.models.RAG.retrievers.retriever_model import RetrieverModel
+from DashAI.back.models.RAG.retrievers.sparse import (
+    BM25Retriever,
+    BM25VectorizerModel,
+    SparseRetriever,
+    TFIDFRetriever,
+    TFIDFVectorizerModel,
+)
+from DashAI.back.models.RAG.retrievers.unit_retriever import UnitRetriever
diff --git a/DashAI/back/models/RAG/retrievers/composite/__init__.py b/DashAI/back/models/RAG/retrievers/composite/__init__.py
new file mode 100644
index 000000000..347e2a803
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/composite/__init__.py
@@ -0,0 +1,12 @@
+from DashAI.back.models.RAG.retrievers.composite.composite_retriever import (
+    CompositeRetriever,
+)
+from DashAI.back.models.RAG.retrievers.composite.mmr_reranker_retriever import (
+    MMRRerankerRetriever,
+)
+from DashAI.back.models.RAG.retrievers.composite.parallel_retriever import (
+    ParallelRetriever,
+)
+from DashAI.back.models.RAG.retrievers.composite.sequential_retriever import (
+    SequentialRetriever,
+)
diff --git a/DashAI/back/models/RAG/retrievers/composite/composite_retriever.py b/DashAI/back/models/RAG/retrievers/composite/composite_retriever.py
new file mode 100644
index 000000000..31fc4dd39
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/composite/composite_retriever.py
@@ -0,0 +1,47 @@
+from abc import ABC, abstractmethod
+from typing import Final, List
+
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.retriever_model import RetrieverModel
+
+
+class CompositeRetriever(RetrieverModel, ABC):
+    """
+    Composite: abstract base for retrievers that contain child retrievers.
+
+    Implements the Composite role in the Composite design pattern (GoF).
+    """
+
+    TYPE: Final[str] = "RetrieverModel"
+    FLAGS: list[str] = ["abstract"]
+    REQUIRED_EXTRA_KWARGS: list = []
+
+    def __init__(self, **kwargs):
+        children_data = kwargs.pop("children")
+        if not isinstance(children_data, list):
+            raise TypeError(
+                f"'children' must be a list, got {type(children_data).__name__}"
+            )
+        if children_data and not isinstance(children_data[0], RetrieverModel):
+            raise TypeError(
+                "'children' must contain RetrieverModel instances, "
+                f"got {type(children_data[0]).__name__}"
+            )
+        self._children: List[RetrieverModel] = children_data
+        super().__init__(**kwargs)
+
+    def add(self, child: RetrieverModel) -> None:
+        self._children.append(child)
+
+    def remove(self, child: RetrieverModel) -> None:
+        self._children.remove(child)
+
+    def get_children(self) -> List[RetrieverModel]:
+        return list(self._children)
+
+    @abstractmethod
+    def retrieve(self, query, **kwargs) -> List[Chunk]:
+        raise NotImplementedError
+
+    def score_chunks(self, chunk_ids: List[int], query: str) -> list:
+        return []
diff --git a/DashAI/back/models/RAG/retrievers/composite/mmr_reranker_retriever.py b/DashAI/back/models/RAG/retrievers/composite/mmr_reranker_retriever.py
new file mode 100644
index 000000000..750ddd6d6
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/composite/mmr_reranker_retriever.py
@@ -0,0 +1,130 @@
+from typing import List
+
+import numpy as np
+from sklearn.metrics.pairwise import cosine_similarity
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    component_field,
+    float_field,
+    int_field,
+    list_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.composite.composite_retriever import (
+    CompositeRetriever,
+)
+
+
+class MMRRerankerRetrieverSchema(BaseSchema):
+    mmr_lambda: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.5,
+        description=MultilingualString(
+            en="Trade-off between relevance (1.0) and diversity (0.0).",
+            es="Compromiso entre relevancia (1.0) y diversidad (0.0).",
+        ),
+    )  # type: ignore
+
+    retrieval_factor: schema_field(
+        int_field(gt=1),
+        placeholder=3,
+        description=MultilingualString(
+            en="Multiplier for initial retrieval size.",
+            es="Multiplicador para el tamaño de recuperación inicial.",
+        ),
+    )  # type: ignore
+
+    top_k: schema_field(
+        int_field(gt=0),
+        placeholder=5,
+        description=MultilingualString(
+            en="Final number of chunks to select.",
+            es="Número final de fragmentos a seleccionar.",
+        ),
+    )  # type: ignore
+
+    children: schema_field(
+        list_field(component_field(parent="RetrieverModel"), min_items=1, max_items=1),
+        placeholder=[],
+        description=MultilingualString(
+            en="The child retriever whose results will be re-ranked.",
+            es="El recuperador hijo cuyos resultados serán reordenados.",
+        ),
+    )  # type: ignore
+
+
+class MMRRerankerRetriever(CompositeRetriever):
+    FLAGS: list[str] = ["FAMILY:mmr", "composite", "reranker"]
+    SCHEMA = MMRRerankerRetrieverSchema
+    DISPLAY_NAME: str = MultilingualString(
+        en="MMR Reranker",
+        es="Reordenador MMR",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Re-ranks retrieval results using Maximum Marginal Relevance for diversity.",
+        es="Reordena resultados de recuperación usando Maximum Marginal Relevance "
+        "para diversidad.",
+    )
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.mmr_lambda = self.params.pop("mmr_lambda")
+        self.retrieval_factor = self.params.pop("retrieval_factor")
+        self._top_k = self.params.pop("top_k")
+
+    @property
+    def retrieval_top_k(self) -> int:
+        return self._top_k
+
+    def retrieve(self, query, **kwargs) -> List[Chunk]:
+        child = self._children[0]
+        expanded_k = self._top_k * self.retrieval_factor
+        candidates = child.retrieve(query, top_k=expanded_k)
+        if len(candidates) <= self._top_k:
+            return candidates
+
+        chunk_ids = [c.id for c in candidates]
+        scored = child.score_chunks(chunk_ids, query)
+        id_to_dist = dict(scored)
+
+        ordered = [c for c in candidates if c.id in id_to_dist]
+        if len(ordered) <= self._top_k:
+            return ordered[: self._top_k]
+
+        relevance = np.array([1.0 - id_to_dist[c.id] for c in ordered])
+        ordered_ids = [c.id for c in ordered]
+        try:
+            vectors = child.get_chunk_vectors(ordered_ids)
+        except ValueError:
+            return ordered[: self._top_k]
+
+        pairwise = cosine_similarity(vectors)
+        selected = self._mmr_select(relevance, pairwise)
+        return [ordered[i] for i in selected]
+
+    def _mmr_select(
+        self,
+        relevance: np.ndarray,
+        pairwise: np.ndarray,
+    ) -> List[int]:
+        n = len(relevance)
+        best = int(np.argmax(relevance))
+        selected = [best]
+        remaining = [i for i in range(n) if i != best]
+
+        while len(selected) < self._top_k and remaining:
+            scores = np.empty(len(remaining))
+            for j, cand in enumerate(remaining):
+                max_pair = max(pairwise[cand][s] for s in selected)
+                scores[j] = (
+                    self.mmr_lambda * relevance[cand]
+                    - (1.0 - self.mmr_lambda) * max_pair
+                )
+            best_idx = int(np.argmax(scores))
+            selected.append(remaining[best_idx])
+            remaining.pop(best_idx)
+
+        return selected
diff --git a/DashAI/back/models/RAG/retrievers/composite/parallel_retriever.py b/DashAI/back/models/RAG/retrievers/composite/parallel_retriever.py
new file mode 100644
index 000000000..e8952b82f
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/composite/parallel_retriever.py
@@ -0,0 +1,112 @@
+from typing import List, Tuple
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    component_field,
+    enum_field,
+    list_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.composite.composite_retriever import (
+    CompositeRetriever,
+)
+from DashAI.back.models.RAG.retrievers.enums import MergeStrategy
+
+
+class ParallelRetrieverSchema(BaseSchema):
+    children: schema_field(
+        list_field(component_field(parent="RetrieverModel"), min_items=2),
+        placeholder=[],
+        description=MultilingualString(
+            en="List of child retrievers queried in parallel.",
+            es="Lista de recuperadores hijos consultados en paralelo.",
+        ),
+    )  # type: ignore
+
+    merge_strategy: schema_field(
+        enum_field(enum=[s.value for s in MergeStrategy]),
+        placeholder=MergeStrategy.ROUND_ROBIN.value,
+        description=MultilingualString(
+            en=(
+                f"'{MergeStrategy.ROUND_ROBIN.value}': alternates results from each retriever. "
+                f"'{MergeStrategy.INTERLEAVE.value}': merges preserving internal order."
+            ),
+            es=(
+                f"'{MergeStrategy.ROUND_ROBIN.value}': alterna resultados de cada recuperador. "
+                f"'{MergeStrategy.INTERLEAVE.value}': fusiona preservando el orden interno."
+            ),
+        ),
+    )  # type: ignore
+
+
+class ParallelRetriever(CompositeRetriever):
+    FLAGS: list[str] = ["composite", "parallel"]
+    SCHEMA = ParallelRetrieverSchema
+    DISPLAY_NAME: str = MultilingualString(
+        en="Parallel Retriever",
+        es="Recuperador Paralelo",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Queries multiple retrievers in parallel and merges their results.",
+        es="Consulta múltiples recuperadores en paralelo y fusiona sus resultados.",
+    )
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self.merge_strategy = MergeStrategy(self.params.pop("merge_strategy"))
+
+    @property
+    def retrieval_top_k(self) -> int:
+        return sum(c.retrieval_top_k for c in self._children)
+
+    def retrieve(self, query, **kwargs) -> List[Chunk]:
+        all_child_results = []
+        for child in self._children:
+            all_child_results.append(child.retrieve(query, **kwargs))
+
+        total_k = sum(c.retrieval_top_k for c in self._children)
+
+        if self.merge_strategy == MergeStrategy.ROUND_ROBIN:
+            return self._merge_round_robin(all_child_results, total_k)
+        return self._merge_interleave(all_child_results, total_k)
+
+    @staticmethod
+    def _chunk_key(chunk: Chunk) -> Tuple[str, int]:
+        return (chunk.document_id, chunk.document_position)
+
+    def _merge_round_robin(
+        self, child_results_list: List[List[Chunk]], total_k: int
+    ) -> List[Chunk]:
+        results = []
+        seen_keys: set[Tuple[str, int]] = set()
+        max_len = max(len(r) for r in child_results_list) if child_results_list else 0
+
+        for i in range(max_len):
+            for child_results in child_results_list:
+                if i < len(child_results):
+                    chunk = child_results[i]
+                    key = self._chunk_key(chunk)
+                    if key not in seen_keys:
+                        seen_keys.add(key)
+                        results.append(chunk)
+                        if len(results) >= total_k:
+                            return results
+        return results
+
+    def _merge_interleave(
+        self, child_results_list: List[List[Chunk]], total_k: int
+    ) -> List[Chunk]:
+        results = []
+        seen_keys: set[Tuple[str, int]] = set()
+
+        for child_results in child_results_list:
+            for chunk in child_results:
+                key = self._chunk_key(chunk)
+                if key not in seen_keys:
+                    seen_keys.add(key)
+                    results.append(chunk)
+                    if len(results) >= total_k:
+                        return results
+        return results
diff --git a/DashAI/back/models/RAG/retrievers/composite/sequential_retriever.py b/DashAI/back/models/RAG/retrievers/composite/sequential_retriever.py
new file mode 100644
index 000000000..6a84b9b9b
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/composite/sequential_retriever.py
@@ -0,0 +1,89 @@
+from typing import List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    component_field,
+    list_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.composite.composite_retriever import (
+    CompositeRetriever,
+)
+from DashAI.back.models.RAG.retrievers.exceptions import (
+    CompositeValidationError,
+    UnitRetrieverChildError,
+)
+from DashAI.back.models.RAG.retrievers.unit_retriever import UnitRetriever
+
+
+class SequentialRetrieverSchema(BaseSchema):
+    children: schema_field(
+        list_field(component_field(parent="RetrieverModel"), min_items=2),
+        placeholder=[],
+        description=MultilingualString(
+            en="Ordered list of child retrievers. The first child retrieves broadly; "
+            "each subsequent child re-ranks and tightens the results.",
+            es="Lista ordenada de recuperadores hijos. El primero recupera ampliamente; "
+            "cada hijo subsiguiente reordena y ajusta los resultados.",
+        ),
+    )  # type: ignore
+
+
+class SequentialRetriever(CompositeRetriever):
+    FLAGS: list[str] = ["composite", "sequential"]
+    SCHEMA = SequentialRetrieverSchema
+    DISPLAY_NAME: str = MultilingualString(
+        en="Sequential Retriever",
+        es="Recuperador Secuencial",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Queries multiple retrievers in sequence, re-ranking at each step.",
+        es="Consulta múltiples recuperadores en secuencia, reordenando en cada paso.",
+    )
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+        self._validate()
+
+    def _validate(self) -> None:
+        children = getattr(self, "_children", [])
+        if len(children) < 2:
+            return
+
+        children_k = [c.retrieval_top_k for c in children]
+
+        for _i, child in enumerate(children):
+            if not isinstance(child, UnitRetriever):
+                raise UnitRetrieverChildError(
+                    child_class=child.__class__.__name__,
+                    strategy="cascade",
+                )
+
+        for i in range(1, len(children_k)):
+            if children_k[i] >= children_k[i - 1]:
+                raise CompositeValidationError(
+                    f"Cascade requires strictly decreasing top_k. "
+                    f"Child {i} top_k={children_k[i]} >= "
+                    f"child {i - 1} top_k={children_k[i - 1]}"
+                )
+
+    def retrieve(self, query, **kwargs) -> List[Chunk]:
+        first = self._children[0]
+        results = first.retrieve(query, **kwargs)
+
+        for child in self._children[1:]:
+            chunk_ids = [c.id for c in results]
+            scored = child.score_chunks(chunk_ids, query)
+            scored = scored[: child.retrieval_top_k]
+            id_to_chunk = {c.id: c for c in results}
+            results = [id_to_chunk[cid] for cid, _ in scored]
+
+        return results
+
+    @property
+    def retrieval_top_k(self) -> int:
+        if self._children:
+            return self._children[-1].retrieval_top_k
+        return 1
diff --git a/DashAI/back/models/RAG/retrievers/dense/__init__.py b/DashAI/back/models/RAG/retrievers/dense/__init__.py
new file mode 100644
index 000000000..14a5985ab
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/__init__.py
@@ -0,0 +1,7 @@
+from DashAI.back.models.RAG.retrievers.dense.dense_embedding_retriever import (
+    DenseEmbeddingRetriever,
+)
+from DashAI.back.models.RAG.retrievers.dense.dense_retriever import DenseRetriever
+from DashAI.back.models.RAG.retrievers.dense.huggingface_dense_retriever import (
+    HuggingFaceDenseRetriever,
+)
diff --git a/DashAI/back/models/RAG/retrievers/dense/_hf_language_utils.py b/DashAI/back/models/RAG/retrievers/dense/_hf_language_utils.py
new file mode 100644
index 000000000..03aa92f16
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/_hf_language_utils.py
@@ -0,0 +1,91 @@
+from typing import Dict, List, Tuple
+
+LANGUAGE_LABELS: Dict[str, str] = {
+    "af": "Afr",
+    "ar": "Ara",
+    "bg": "Bul",
+    "bn": "Ben",
+    "ca": "Cat",
+    "cs": "Cze",
+    "da": "Dan",
+    "de": "Deu",
+    "el": "Gre",
+    "en": "Eng",
+    "es": "Esp",
+    "et": "Est",
+    "fa": "Per",
+    "fi": "Fin",
+    "fr": "Fra",
+    "he": "Heb",
+    "hi": "Hin",
+    "hr": "Cro",
+    "hu": "Hun",
+    "id": "Ind",
+    "it": "Ita",
+    "ja": "Jpn",
+    "ko": "Kor",
+    "lt": "Lit",
+    "lv": "Lav",
+    "mk": "Mac",
+    "ms": "May",
+    "multi": "Multi",
+    "nl": "Dut",
+    "no": "Nor",
+    "pl": "Pol",
+    "pt": "Por",
+    "ro": "Rom",
+    "ru": "Rus",
+    "sk": "Slo",
+    "sl": "Svn",
+    "sr": "Srp",
+    "sv": "Swe",
+    "th": "Tha",
+    "tr": "Tur",
+    "uk": "Ukr",
+    "ur": "Urd",
+    "vi": "Vie",
+    "zh": "Chi",
+}
+
+MAX_DISPLAY_LANGUAGES: int = 3
+
+
+def compute_language_summary(languages: List[str]) -> Tuple[str, int]:
+    if not languages:
+        return "", 0
+    labels = [LANGUAGE_LABELS.get(lang, lang.title()) for lang in languages]
+    if len(labels) <= MAX_DISPLAY_LANGUAGES:
+        return ", ".join(labels), len(languages)
+    shown = labels[:MAX_DISPLAY_LANGUAGES]
+    overflow = len(languages) - MAX_DISPLAY_LANGUAGES
+    return f"{', '.join(shown)} +{overflow}", len(languages)
+
+
+def build_model_language_summaries(
+    models: Dict[str, dict],
+) -> Dict[str, Dict[str, object]]:
+    result: Dict[str, Dict[str, object]] = {}
+    for model_name, info in models.items():
+        summary, count = compute_language_summary(info.get("languages", []))
+        result[model_name] = {
+            "summary": summary,
+            "count": count,
+            "labels": [
+                LANGUAGE_LABELS.get(lang, lang.title())
+                for lang in info.get("languages", [])
+            ],
+        }
+    return result
+
+
+def build_family_language_summary(
+    models: Dict[str, dict],
+) -> Tuple[str, int]:
+    all_languages: List[str] = []
+    seen: set[str] = set()
+    for info in models.values():
+        for lang in info.get("languages", []):
+            if lang not in seen:
+                seen.add(lang)
+                all_languages.append(lang)
+    return compute_language_summary(all_languages)
diff --git a/DashAI/back/models/RAG/retrievers/dense/_hf_metadata_utils.py b/DashAI/back/models/RAG/retrievers/dense/_hf_metadata_utils.py
new file mode 100644
index 000000000..daca8d7a9
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/_hf_metadata_utils.py
@@ -0,0 +1,22 @@
+from typing import Dict
+
+from DashAI.back.models.RAG.retrievers.dense._hf_language_utils import (
+    build_family_language_summary,
+    build_model_language_summaries,
+)
+
+
+def build_retriever_metadata(
+    models: Dict[str, dict],
+    family_name: str,
+    model_count: int,
+) -> Dict[str, object]:
+    family_summary, family_count = build_family_language_summary(models)
+    model_languages = build_model_language_summaries(models)
+    return {
+        "family": family_name,
+        "language_summary": family_summary,
+        "language_count": family_count,
+        "model_count": model_count,
+        "model_languages": model_languages,
+    }
diff --git a/DashAI/back/models/RAG/retrievers/dense/dense_embedding_retriever.py b/DashAI/back/models/RAG/retrievers/dense/dense_embedding_retriever.py
new file mode 100644
index 000000000..23e5bb347
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/dense_embedding_retriever.py
@@ -0,0 +1,75 @@
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    component_field,
+    enum_field,
+    int_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.retrievers.dense.dense_retriever import DenseRetriever
+
+METRICS = ["cityblock", "cosine", "euclidean", "l1", "l2", "manhattan"]
+
+
+class DenseEmbeddingRetrieverSchema(BaseSchema):
+    embedding_model: schema_field(
+        component_field(parent="DenseEmbedding"),
+        placeholder={"component": "SentenceTransformerEmbedding", "params": {}},
+        description=MultilingualString(
+            en="Embedding model to use for encoding chunks.",
+            es="Modelo de embedding a usar para codificar fragmentos.",
+        ),
+    )  # type: ignore
+
+    similarity_metric: schema_field(
+        enum_field(enum=METRICS),
+        placeholder="cosine",
+        description=MultilingualString(
+            en="Distance metric for comparing dense vectors.",
+            es="Métrica de distancia para comparar vectores densos.",
+        ),
+    )  # type: ignore
+
+    top_k: schema_field(
+        int_field(gt=0),
+        placeholder=5,
+        description=MultilingualString(
+            en="Number of chunks to select.",
+            es="Número de fragmentos a seleccionar.",
+        ),
+    )  # type: ignore
+
+
+class DenseEmbeddingRetriever(DenseRetriever):
+    """Concrete dense retriever that accepts any DenseEmbedding component."""
+
+    FLAGS: list[str] = ["dense", "dense_embedding"]
+    SCHEMA = DenseEmbeddingRetrieverSchema
+
+    DISPLAY_NAME: str = MultilingualString(
+        en="Dense Embedding Retriever",
+        es="Recuperador por Embeddings Densos",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense retriever using any registered DenseEmbedding for similarity search.",
+        es="Recuperador denso que usa cualquier DenseEmbedding registrado para búsqueda por similitud.",
+    )
+
+    def __init__(self, **kwargs):
+        embedding_instance = kwargs.pop("embedding_model")
+        super().__init__(**kwargs)
+        self.params["embedding_model"] = {
+            "component": embedding_instance.__class__.__name__,
+            "params": dict(sorted(embedding_instance.params.items())),
+        }
+        self._embedding_instance = embedding_instance
+
+    def init_model(self) -> None:
+        """Load the embedding model, then initialise the similarity matrix.
+
+        The embedding instance is created by :meth:`fill_objects` during
+        factory construction but its heavy resources (tokenizer, weights)
+        are only acquired on the explicit ``load()`` call.
+        """
+        self._embedding_instance.load()
+        self._init_embedding(self._embedding_instance)
diff --git a/DashAI/back/models/RAG/retrievers/dense/dense_retriever.py b/DashAI/back/models/RAG/retrievers/dense/dense_retriever.py
new file mode 100644
index 000000000..63745336d
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/dense_retriever.py
@@ -0,0 +1,179 @@
+import os
+from typing import Dict, List, Tuple
+
+import numpy as np
+from sklearn.metrics.pairwise import pairwise_distances
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    int_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.embeddings import DenseEmbedding
+from DashAI.back.models.RAG.retrievers.unit_retriever import UnitRetriever
+
+# NOTE: The entire chunk index (similarity_matrix) is loaded into memory,
+# which is fine for typical use with tens to low hundreds of documents but
+# may be a bottleneck for very large collections.
+
+
+class DenseRetrieverSchema(BaseSchema):
+    similarity_metric: schema_field(
+        enum_field(enum=["cityblock", "cosine", "euclidean", "l1", "l2", "manhattan"]),
+        placeholder="cosine",
+        description=MultilingualString(
+            en="Distance metric for comparing dense vectors.",
+            es="Métrica de distancia para comparar vectores densos.",
+        ),
+    )  # type: ignore
+
+    top_k: schema_field(
+        int_field(gt=0),
+        placeholder=5,
+        description=MultilingualString(
+            en="Number of chunks to select.",
+            es="Número de fragmentos a seleccionar.",
+        ),
+    )  # type: ignore
+
+
+class DenseRetriever(UnitRetriever):
+    FLAGS: list[str] = ["abstract"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="Embedding Retriever",
+        es="Recuperador por Embeddings",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Embedding retriever using vector embeddings for similarity search.",
+        es="Recuperador por embeddings que usa representaciones vectoriales para búsqueda por similitud.",
+    )
+
+    SCHEMA = DenseRetrieverSchema
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        self.similarity_metric = self.params.pop("similarity_metric")
+        self._top_k = self.params.pop("top_k")
+
+    def _init_embedding(self, embedding_model):
+        if not isinstance(embedding_model, DenseEmbedding):
+            raise TypeError(
+                f"Expected DenseEmbedding instance, "
+                f"got {type(embedding_model).__name__}"
+            )
+        self.embedding_model = embedding_model
+        encoding_class_name = embedding_model.__class__.__name__
+        encoding_params = dict(sorted(embedding_model.params.items()))
+
+        self.params["encoding_model"] = {
+            "class_name": encoding_class_name,
+            "parameters": encoding_params,
+        }
+
+        self.compute_missing_embeddings()
+        self.init_similarity_matrix()
+
+    def compute_missing_embeddings(self):
+        for doc_id, doc_chunks in self.chunks.items():
+            matrix_dir = self._persistence.matrix_dirs.get(doc_id)
+            if matrix_dir is None:
+                continue
+            matrix_path = os.path.join(matrix_dir, "embeddings.npy")
+            if os.path.exists(matrix_path):
+                continue
+            chunk_texts = [chunk.text for chunk in doc_chunks.values()]
+            embeddings = self.embedding_model.batch_encode(chunk_texts)
+            os.makedirs(matrix_dir, exist_ok=True)
+            np.save(matrix_path, embeddings)
+
+    def init_similarity_matrix(self):
+        self.similarity_matrix = None
+        self.matrix_row_to_chunk_id = {}
+        self.chunk_id_to_doc_id = {}
+
+        all_embeddings = []
+        row_index = 0
+        for doc_id, chunks in self.chunks.items():
+            matrix_dir = self._persistence.matrix_dirs.get(doc_id)
+            if matrix_dir is None:
+                continue
+            matrix_path = os.path.join(matrix_dir, "embeddings.npy")
+            if not os.path.exists(matrix_path):
+                continue
+            for chunk_id in chunks:
+                self.matrix_row_to_chunk_id[row_index] = chunk_id
+                self.chunk_id_to_doc_id[chunk_id] = doc_id
+                row_index += 1
+            all_embeddings.append(np.load(matrix_path))
+
+        if all_embeddings:
+            self.similarity_matrix = np.vstack(all_embeddings)
+
+    @property
+    def retrieval_top_k(self) -> int:
+        return self._top_k
+
+    def retrieve(self, query: str, top_k: int | None = None) -> List[Chunk]:
+        self._check_infra()
+        if self.similarity_matrix is None:
+            raise ValueError("Similarity matrix not initialized.")
+        k = top_k if top_k is not None else self._top_k
+        query_embedding = self.embedding_model.encode(query)
+        # NOTE: pairwise_distances computes O(n×dim) per query with no
+        # FAISS/HNSW indexing. Low priority since current use cases have
+        # small document sets.
+        distances = pairwise_distances(
+            query_embedding.reshape(1, -1),
+            self.similarity_matrix,
+            metric=self.similarity_metric,
+        )[0]
+        top_indices = np.argsort(distances)[:k]
+        results = []
+        for idx in top_indices:
+            chunk_id = self.matrix_row_to_chunk_id[idx]
+            doc_id = self.chunk_id_to_doc_id[chunk_id]
+            results.append(self.chunks[doc_id][chunk_id])
+        return results
+
+    def score_chunks(self, chunk_ids: List[int], query: str) -> List[Tuple[int, float]]:
+        self._check_infra()
+        if self.similarity_matrix is None:
+            raise ValueError("Similarity matrix not initialized.")
+        query_embedding = self.embedding_model.encode(query)
+        chunk_id_to_row = {
+            self.matrix_row_to_chunk_id[r]: r for r in self.matrix_row_to_chunk_id
+        }
+        valid_ids = [cid for cid in chunk_ids if cid in chunk_id_to_row]
+        if not valid_ids:
+            return []
+        rows = [chunk_id_to_row[cid] for cid in valid_ids]
+        distances = pairwise_distances(
+            query_embedding.reshape(1, -1),
+            self.similarity_matrix[rows],
+            metric=self.similarity_metric,
+        )[0]
+        scored = list(zip(valid_ids, distances.tolist()))
+        scored.sort(key=lambda x: x[1])
+        return scored
+
+    def get_chunk_vectors(self, chunk_ids: List[int]) -> np.ndarray:
+        if self.similarity_matrix is None:
+            raise ValueError("Similarity matrix not initialized.")
+        chunk_id_to_row: Dict[int, int] = {
+            self.matrix_row_to_chunk_id[r]: r for r in self.matrix_row_to_chunk_id
+        }
+        rows = []
+        for cid in chunk_ids:
+            row = chunk_id_to_row.get(cid)
+            if row is not None:
+                rows.append(row)
+        if not rows:
+            raise ValueError(
+                f"None of the provided chunk_ids {chunk_ids} were found "
+                "in the similarity matrix."
+            )
+        return self.similarity_matrix[rows]
diff --git a/DashAI/back/models/RAG/retrievers/dense/huggingface_dense_retriever.py b/DashAI/back/models/RAG/retrievers/dense/huggingface_dense_retriever.py
new file mode 100644
index 000000000..e46526f18
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/dense/huggingface_dense_retriever.py
@@ -0,0 +1,33 @@
+from abc import abstractmethod
+
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.embeddings.dense.huggingface_embedding import (
+    HuggingFaceEmbedding,
+)
+from DashAI.back.models.RAG.retrievers.dense.dense_retriever import DenseRetriever
+
+METRICS = ["cityblock", "cosine", "euclidean", "l1", "l2", "manhattan"]
+
+
+class HuggingFaceDenseRetriever(DenseRetriever):
+    FLAGS: list[str] = ["abstract", "huggingface"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="HuggingFace Embedding Retriever",
+        es="Recuperador por Embeddings HuggingFace",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Dense retriever using HuggingFace embeddings for similarity search.",
+        es="Recuperador denso que usa embeddings HuggingFace para búsqueda por similitud.",
+    )
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+    @abstractmethod
+    def _create_embedding(self) -> HuggingFaceEmbedding:
+        raise NotImplementedError
+
+    def init_model(self) -> None:
+        embedding = self._create_embedding()
+        embedding.load()
+        self._init_embedding(embedding)
diff --git a/DashAI/back/models/RAG/retrievers/enums.py b/DashAI/back/models/RAG/retrievers/enums.py
new file mode 100644
index 000000000..33f721321
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/enums.py
@@ -0,0 +1,6 @@
+from enum import Enum
+
+
+class MergeStrategy(str, Enum):
+    ROUND_ROBIN = "round_robin"
+    INTERLEAVE = "interleave"
diff --git a/DashAI/back/models/RAG/retrievers/exceptions.py b/DashAI/back/models/RAG/retrievers/exceptions.py
new file mode 100644
index 000000000..b9cba1448
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/exceptions.py
@@ -0,0 +1,42 @@
+class RetrieverError(Exception):
+    """Base exception for all retriever-related errors."""
+
+
+class MissingParameterError(RetrieverError):
+    """A required parameter was not provided to the retriever."""
+
+    def __init__(self, param_name: str, retriever_name: str):
+        super().__init__(
+            f"Missing required parameter '{param_name}' "
+            f"in retriever '{retriever_name}'."
+        )
+        self.param_name = param_name
+        self.retriever_name = retriever_name
+
+
+class ExtraKwargsMissingError(RetrieverError):
+    """One or more required infrastructure kwargs are missing."""
+
+    def __init__(self, missing_keys: set, retriever_name: str):
+        super().__init__(
+            f"Missing required extra kwargs {missing_keys} "
+            f"in retriever '{retriever_name}'."
+        )
+        self.missing_keys = missing_keys
+        self.retriever_name = retriever_name
+
+
+class CompositeValidationError(RetrieverError):
+    """A composite retriever has an invalid configuration."""
+
+
+class UnitRetrieverChildError(RetrieverError):
+    """A composite retriever received a child that is not a UnitRetriever."""
+
+    def __init__(self, child_class: str, strategy: str):
+        super().__init__(
+            f"Cascade strategy '{strategy}' requires UnitRetriever children, "
+            f"but child '{child_class}' is a composite."
+        )
+        self.child_class = child_class
+        self.strategy = strategy
diff --git a/DashAI/back/models/RAG/retrievers/persistence.py b/DashAI/back/models/RAG/retrievers/persistence.py
new file mode 100644
index 000000000..171e38a7d
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/persistence.py
@@ -0,0 +1,25 @@
+from dataclasses import dataclass
+
+
+@dataclass
+class SparsePersistence:
+    """Reference to a sparse retriever's on-disk dump.
+
+    model_dir is the absolute path to the directory containing .pkl files.
+    If None, no previous dump exists and the retriever must train from scratch.
+    """
+
+    model_dir: str | None
+
+
+@dataclass
+class DensePersistence:
+    """References to embedding matrices for a dense retriever.
+
+    matrix_dirs maps document_id to the absolute path of the directory
+    containing embeddings.npy for that document.
+    embedding_model_id identifies the embedding model that produced the vectors.
+    """
+
+    matrix_dirs: dict[int, str]
+    embedding_model_id: int
diff --git a/DashAI/back/models/RAG/retrievers/retriever_factory.py b/DashAI/back/models/RAG/retrievers/retriever_factory.py
new file mode 100644
index 000000000..a30a60eba
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/retriever_factory.py
@@ -0,0 +1,335 @@
+"""Factory for retriever instances with full lifecycle encapsulation.
+
+Single-call ``create()`` interface encapsulates DB-record resolution,
+persistence planning, constructor injection, and composite child recursion.
+"""
+
+import os
+from dataclasses import dataclass
+from typing import Any, Dict
+from uuid import uuid4
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.core.schema_fields.utils import fill_objects, normalize_payload
+from DashAI.back.dependencies.registry.component_registry import ComponentRegistry
+from DashAI.back.models.RAG.documents.chunk import Chunk
+from DashAI.back.models.RAG.retrievers.composite.composite_retriever import (
+    CompositeRetriever,
+)
+from DashAI.back.models.RAG.retrievers.dense.dense_retriever import DenseRetriever
+from DashAI.back.models.RAG.retrievers.persistence import (
+    DensePersistence,
+    SparsePersistence,
+)
+from DashAI.back.models.RAG.retrievers.retriever_model import RetrieverModel
+from DashAI.back.models.RAG.retrievers.retriever_repository import (
+    RetrieverRepository,
+)
+from DashAI.back.models.RAG.retrievers.sparse.sparse_retriever import SparseRetriever
+
+
+@dataclass(frozen=True)
+class RetrieverFactoryResult:
+    db_record_id: int
+    model: RetrieverModel
+
+
+_ENCODING_MODEL_KEY = "encoding_model"
+
+
+class RetrieverFactory:
+    """Creates retriever instances with full persistence lifecycle.
+
+    All DB access delegated to ``RetrieverRepository``.
+    """
+
+    def __init__(
+        self,
+        db: Session,
+        pipeline_id: int,
+        registry: ComponentRegistry,
+        env_rag_path: str,
+        chunks: Dict[int, Dict[int, Chunk]],
+        chunk_set_id: int,
+    ):
+        self._db = db
+        self._pipeline_id = pipeline_id
+        self._registry = registry
+        self._env_rag_path = env_rag_path
+        self._chunks = chunks
+        self._chunk_set_id = chunk_set_id
+        self._repo = RetrieverRepository(db)
+
+    def create(
+        self,
+        component_name: str,
+        params: Dict[str, Any],
+    ) -> RetrieverFactoryResult:
+        params = normalize_payload(params)
+        model_class = self._registry[component_name]["class"]
+
+        if issubclass(model_class, CompositeRetriever):
+            return self._create_composite(model_class, params)
+
+        return self._create_unit(model_class, params)
+
+    def _create_composite(self, model_class, params):
+        existing = self._load_composite_from_db(model_class, params)
+        if existing is not None:
+            return RetrieverFactoryResult(
+                db_record_id=existing.get_id(),
+                model=existing,
+            )
+
+        children_configs = params.pop("children")
+        children_instances = [self._create_child_retriever(c) for c in children_configs]
+        params["children"] = children_instances
+        model = model_class(**params)
+        bridge_id = self._save_composite(model)
+        model.set_id(bridge_id)
+        return RetrieverFactoryResult(db_record_id=bridge_id, model=model)
+
+    def _create_child_retriever(self, child_config):
+        return self.create(
+            component_name=child_config["component"],
+            params=child_config["params"],
+        ).model
+
+    def _load_composite_from_db(self, model_class, params):
+        bridge_record = self._repo.find_composite(
+            self._pipeline_id,
+            model_class.__name__,
+        )
+        if bridge_record is None:
+            return None
+        child_links = self._repo.find_composite_children(bridge_record.id)
+        if not child_links:
+            return None
+
+        children = []
+        for link in child_links:
+            child_bridge = self._repo.get_bridge_by_id(link.child_id)
+            if child_bridge is None:
+                return None
+            child_params = self._load_child_params(
+                child_bridge.id, child_bridge.class_name
+            )
+            if child_params is None:
+                return None
+            children.append(
+                self.create(
+                    component_name=child_bridge.class_name,
+                    params=child_params,
+                ).model,
+            )
+
+        model = model_class(
+            children=children,
+            **{k: v for k, v in params.items() if k != "children"},
+        )
+        model.set_id(bridge_record.id)
+        return model
+
+    def _load_child_params(
+        self,
+        bridge_id: int,
+        class_name: str,
+    ) -> Dict[str, Any] | None:
+        sparse = self._repo.find_sparse_by_bridge_id(bridge_id)
+        if sparse is not None and sparse.parameters is not None:
+            return dict(sorted(sparse.parameters.items()))
+        dense = self._repo.find_dense_by_bridge_id(bridge_id)
+        if dense is not None and dense.parameters is not None:
+            return dict(sorted(dense.parameters.items()))
+        return None
+
+    def _save_composite(self, model):
+        child_ids = []
+        for child in model.get_children():
+            cid = child.get_id()
+            if cid is None:
+                raise ValueError(f"Child {child.__class__.__name__} has no bridge ID.")
+            child_ids.append(cid)
+        return self._repo.save_composite(
+            model.__class__.__name__,
+            self._pipeline_id,
+            child_ids,
+        ).id
+
+    def _create_unit(self, model_class, params):
+        sorted_params = dict(sorted(params.items()))
+
+        loaded = self._load_unit_from_db(
+            model_class,
+            model_class.__name__,
+            sorted_params,
+        )
+        if loaded is not None:
+            return RetrieverFactoryResult(
+                db_record_id=loaded.get_id(),
+                model=loaded,
+            )
+
+        validated = model_class.SCHEMA.model_validate(params)
+        resolved = fill_objects(validated, self._registry)
+        model = model_class(**resolved)
+
+        if issubclass(model_class, DenseRetriever):
+            model.inject_infra(
+                self._env_rag_path,
+                self._chunks,
+                DensePersistence(matrix_dirs={}, embedding_model_id=0),
+            )
+            model.init_model()
+            self._finalize_dense(model, sorted_params)
+        elif issubclass(model_class, SparseRetriever):
+            model.inject_infra(
+                self._env_rag_path,
+                self._chunks,
+                SparsePersistence(model_dir=None),
+            )
+            model.init_model()
+        else:
+            raise ValueError(f"Unsupported unit retriever: {model_class.__name__}")
+
+        bridge_id = self._save_unit(model, sorted_params)
+        model.set_id(bridge_id)
+        return RetrieverFactoryResult(db_record_id=bridge_id, model=model)
+
+    def _finalize_dense(self, model, sorted_params):
+        """After init_model() for a new dense model: persist embedding
+        model, build persistence with the proper ID, compute matrices."""
+        enc = model.params.get(_ENCODING_MODEL_KEY)
+        if not enc:
+            return
+        emb_record = self._repo.find_or_create_embedding_model(
+            enc["class_name"],
+            dict(sorted(enc["parameters"].items())),
+        )
+        model._persistence = self._dense_persistence(emb_record.id)
+        model.compute_missing_embeddings()
+        model.init_similarity_matrix()
+
+    def _dense_persistence(self, embedding_model_id: int) -> DensePersistence:
+        """Build a DensePersistence with matrix dirs for every known doc."""
+        matrix_dirs = {}
+        for doc_id in self._chunks:
+            folder = (
+                f"doc_id-{doc_id}__"
+                f"chunk_set_id-{self._chunk_set_id}__"
+                f"embedding_model_id-{embedding_model_id}"
+            )
+            matrix_dirs[doc_id] = os.path.join(
+                self._env_rag_path,
+                "embeddings",
+                folder,
+            )
+        return DensePersistence(
+            matrix_dirs=matrix_dirs,
+            embedding_model_id=embedding_model_id,
+        )
+
+    def _load_unit_from_db(self, model_class, class_name, sorted_params):
+        if issubclass(model_class, DenseRetriever):
+            return self._load_dense(class_name, sorted_params)
+        if issubclass(model_class, SparseRetriever):
+            return self._load_sparse(class_name, sorted_params)
+        raise ValueError(f"Unsupported unit retriever: {model_class.__name__}")
+
+    def _load_dense(self, class_name, sorted_params):
+        record = self._repo.find_dense(class_name, sorted_params, self._chunk_set_id)
+        if record is None:
+            return None
+        persistence = self._dense_persistence(record.embedding_model_id)
+        model_class = self._registry[class_name]["class"]
+
+        validated = model_class.SCHEMA.model_validate(dict(sorted_params))
+        resolved = fill_objects(validated, self._registry)
+        model = model_class(**resolved)
+        model.inject_infra(self._env_rag_path, self._chunks, persistence)
+        model.init_model()
+        bridge = self._repo.find_bridge_for_sub_table(record, class_name)
+        if bridge is not None:
+            model.set_id(bridge.id)
+        return model
+
+    def _load_sparse(self, class_name, sorted_params):
+        record = self._repo.find_sparse(class_name, sorted_params, self._chunk_set_id)
+        if record is None:
+            return None
+        persistence = SparsePersistence(model_dir=record.storage_folder)
+        model_class = self._registry[class_name]["class"]
+
+        validated = model_class.SCHEMA.model_validate(dict(sorted_params))
+        resolved = fill_objects(validated, self._registry)
+        model = model_class(**resolved)
+        model.inject_infra(self._env_rag_path, self._chunks, persistence)
+        model.init_model()
+        bridge = self._repo.find_bridge_for_sub_table(record, class_name)
+        if bridge is not None:
+            model.set_id(bridge.id)
+        return model
+
+    def _save_unit(self, model, sorted_params):
+        if isinstance(model, DenseRetriever):
+            return self._save_dense(model, sorted_params)
+        if isinstance(model, SparseRetriever):
+            return self._save_sparse(model, sorted_params)
+        raise ValueError(f"Unsupported retriever: {type(model).__name__}")
+
+    def _save_dense(self, model, sorted_params):
+        import numpy as np
+
+        for doc_id, mdir in model._persistence.matrix_dirs.items():
+            path = os.path.join(mdir, "embeddings.npy")
+            if not os.path.exists(path):
+                continue
+            if self._repo.find_embedding_matrix(
+                doc_id,
+                self._chunk_set_id,
+                model._persistence.embedding_model_id,
+            ):
+                continue
+            self._repo.save_embedding_matrix(
+                doc_id,
+                self._chunk_set_id,
+                model._persistence.embedding_model_id,
+                mdir,
+                list(np.load(path).shape),
+            )
+
+        bridge = self._repo.create_bridge(
+            model.__class__.__name__,
+            self._pipeline_id,
+        )
+        self._repo.save_dense(
+            model.__class__.__name__,
+            sorted_params,
+            bridge.id,
+            self._chunk_set_id,
+            model._persistence.embedding_model_id,
+        )
+        return bridge.id
+
+    def _save_sparse(self, model, sorted_params):
+        folder_id = uuid4().hex
+        storage_folder = os.path.join(
+            self._env_rag_path,
+            "sparse_retrievers",
+            f"sparse_retriever_id-{folder_id}",
+        )
+        model._persistence.model_dir = storage_folder
+        model.save()
+        bridge = self._repo.create_bridge(
+            model.__class__.__name__,
+            self._pipeline_id,
+        )
+        self._repo.save_sparse(
+            model.__class__.__name__,
+            sorted_params,
+            storage_folder,
+            bridge.id,
+            self._chunk_set_id,
+        )
+        return bridge.id
diff --git a/DashAI/back/models/RAG/retrievers/retriever_model.py b/DashAI/back/models/RAG/retrievers/retriever_model.py
new file mode 100644
index 000000000..f4e1bb935
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/retriever_model.py
@@ -0,0 +1,128 @@
+import os
+from abc import ABC, abstractmethod
+from typing import Any, Dict, Final, List, Tuple
+
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.base_model import BaseModel
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.exceptions import RAGWorkflowError
+
+
+class RetrieverModel(BaseModel, ABC):
+    """
+    Component: abstract base class for all retriever models.
+
+    Implements the Component role in the Composite design pattern (GoF).
+    """
+
+    TYPE: Final[str] = "RetrieverModel"
+    FLAGS: list[str] = ["abstract"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="Retriever",
+        es="Recuperador",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Document retrieval component.",
+        es="Componente de recuperación de documentos.",
+    )
+    COLOR: str = "#9C27B0"
+    ICON: str = "Search"
+
+    env_rag_path: str | os.PathLike | None
+    chunks: Dict[int, Dict[int, Chunk]]
+    chunking_model_id: int | None
+    params: Dict[str, Any]
+
+    def __init__(self, **kwargs):
+        self._db_id: int | None = None
+        self.params = kwargs
+
+    def get_id(self) -> int | None:
+        return self._db_id
+
+    def set_id(self, id: int) -> None:
+        if self._db_id is not None:
+            raise RAGWorkflowError(
+                f"ID is already set to {self._db_id}, cannot reassign to {id}."
+            )
+        self._db_id = id
+
+    def _validate_chunks_dict(self) -> None:
+        if not isinstance(self.chunks, dict):
+            raise ValueError("Chunks must be a dictionary.")
+        for doc_id, doc_chunks in self.chunks.items():
+            if not isinstance(doc_id, int):
+                raise ValueError(f"Document ID {doc_id} must be an integer.")
+            if not isinstance(doc_chunks, dict):
+                raise ValueError(
+                    f"Chunks for document ID {doc_id} must be a dictionary."
+                )
+            for chunk_id, chunk in doc_chunks.items():
+                if not isinstance(chunk_id, int):
+                    raise ValueError(
+                        f"Chunk ID {chunk_id} in document ID {doc_id} must be an integer."
+                    )
+                if not isinstance(chunk, Chunk):
+                    raise ValueError(
+                        f"Chunk {chunk_id} in document ID {doc_id} must be an instance of Chunk."
+                    )
+                if chunk.document_id != doc_id:
+                    raise ValueError(
+                        f"Chunk {chunk_id} document_id {chunk.document_id} != doc ID {doc_id}."
+                    )
+
+    def inject_infra(
+        self,
+        env_rag_path: str | os.PathLike,
+        chunks: Dict[int, Dict[int, Chunk]],
+        persistence: Any,
+    ) -> None:
+        """Inject runtime infrastructure *after* schema validation.
+
+        Subclasses **must** store the three arguments as instance
+        attributes.  Raises ``TypeError`` if any argument has an
+        unexpected type.
+        """
+        self.env_rag_path = env_rag_path
+        self.chunks = chunks
+        self._persistence = persistence
+        self._validate_chunks_dict()
+
+    def init_model(self) -> None:
+        """Called by the factory **after** ``inject_infra()``.
+
+        Subclasses restore saved state (``load()``) or compute initial
+        state (``_fit()``, ``_init_embedding()``).  Default is a no-op.
+        """
+
+    @abstractmethod
+    def retrieve(self, query, **kwargs) -> List[Chunk]:
+        raise NotImplementedError
+
+    @abstractmethod
+    def score_chunks(self, chunk_ids: List[int], query: str) -> List[Tuple[int, float]]:
+        raise NotImplementedError
+
+    @property
+    def retrieval_top_k(self) -> int:
+        raise NotImplementedError
+
+    # ── Child management (Composite pattern) ────────────────────────
+
+    def add(self, child: "RetrieverModel") -> None:
+        raise NotImplementedError
+
+    def remove(self, child: "RetrieverModel") -> None:
+        raise NotImplementedError
+
+    def get_children(self) -> List["RetrieverModel"]:
+        raise NotImplementedError
+
+    def save(self, filename: str = "") -> None:
+        pass
+
+    def load(self, filename: str = "") -> None:
+        pass
+
+    def train(self, **kwargs):
+        return
diff --git a/DashAI/back/models/RAG/retrievers/retriever_repository.py b/DashAI/back/models/RAG/retrievers/retriever_repository.py
new file mode 100644
index 000000000..e3d3c9c0a
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/retriever_repository.py
@@ -0,0 +1,337 @@
+"""Repository for retriever-related database operations.
+
+Separates persistence logic (SQL queries, record creation) from
+orchestration logic (component construction, DI, composites).
+"""
+
+from typing import Dict, List, Optional
+
+from sqlalchemy.orm import Session
+
+from DashAI.back.dependencies.database.models import (
+    RAGDenseRetriever as DenseRetrieverDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGEmbeddingMatrix as EmbeddingMatrixDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGEmbeddingModel as EmbeddingDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGRetriever as RetrieverDBModel,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGRetrieverChild,
+)
+from DashAI.back.dependencies.database.models import (
+    RAGSparseRetriever as SparseRetrieverDBModel,
+)
+
+
+class RetrieverRepository:
+    """All database operations for retriever models.
+
+    This class owns every SQL query, INSERT, and DB-model construction
+    related to retrievers. Factories use it as a collaborator; retriever
+    instances never touch it.
+    """
+
+    def __init__(self, db: Session):
+        self.db = db
+
+    def create_bridge(self, class_name: str, pipeline_id: int) -> RetrieverDBModel:
+        """Insert a new canonical identity record in rag_retriever.
+
+        Every retriever (unit or composite) gets one bridge record.
+        Returns the persisted model with its auto-generated id.
+        """
+        bridge_record = RetrieverDBModel(
+            class_name=class_name,
+            pipeline_id=pipeline_id,
+        )
+        self.db.add(bridge_record)
+        self.db.commit()
+        self.db.refresh(bridge_record)
+        return bridge_record
+
+    def find_sparse(
+        self,
+        class_name: str,
+        parameters: Dict[str, object],
+        chunk_set_id: int,
+    ) -> Optional[SparseRetrieverDBModel]:
+        """Look up an existing sparse retriever record by its natural key.
+
+        Parameters
+        ----------
+        class_name : str
+            Concrete retriever class name (e.g. "TFIDFRetriever").
+        parameters : dict
+            Schema parameters, sorted for deterministic JSON serialisation.
+        chunk_set_id : int
+            Which chunk set this retriever was trained on.
+
+        Returns
+        -------
+        SparseRetrieverDBModel or None
+        """
+        parameters = dict(sorted(parameters.items()))
+        return (
+            self.db.query(SparseRetrieverDBModel)
+            .filter_by(
+                class_name=class_name,
+                parameters=parameters,
+                chunk_set_id=chunk_set_id,
+            )
+            .first()
+        )
+
+    def save_sparse(
+        self,
+        class_name: str,
+        parameters: Dict[str, object],
+        storage_folder: str,
+        bridge_id: int,
+        chunk_set_id: int,
+    ) -> SparseRetrieverDBModel:
+        """Persist a new sparse retriever record and link it to its bridge.
+
+        Returns the persisted model.
+        """
+        parameters = dict(sorted(parameters.items()))
+        record = SparseRetrieverDBModel(
+            bridge_id=bridge_id,
+            chunk_set_id=chunk_set_id,
+            class_name=class_name,
+            parameters=parameters,
+            storage_folder=storage_folder,
+        )
+        self.db.add(record)
+        self.db.commit()
+        self.db.refresh(record)
+        return record
+
+    def find_dense(
+        self,
+        class_name: str,
+        parameters: Dict[str, object],
+        chunk_set_id: int,
+    ) -> Optional[DenseRetrieverDBModel]:
+        """Look up an existing dense retriever record by its natural key."""
+        parameters = dict(sorted(parameters.items()))
+        return (
+            self.db.query(DenseRetrieverDBModel)
+            .filter_by(
+                class_name=class_name,
+                parameters=parameters,
+                chunk_set_id=chunk_set_id,
+            )
+            .first()
+        )
+
+    def save_dense(
+        self,
+        class_name: str,
+        parameters: Dict[str, object],
+        bridge_id: int,
+        chunk_set_id: int,
+        embedding_model_id: int,
+    ) -> DenseRetrieverDBModel:
+        """Persist a new dense retriever record and link it to its bridge."""
+        parameters = dict(sorted(parameters.items()))
+        record = DenseRetrieverDBModel(
+            bridge_id=bridge_id,
+            chunk_set_id=chunk_set_id,
+            class_name=class_name,
+            parameters=parameters,
+            embedding_model_id=embedding_model_id,
+        )
+        self.db.add(record)
+        self.db.commit()
+        self.db.refresh(record)
+        return record
+
+    def find_composite(
+        self, pipeline_id: int, class_name: str
+    ) -> Optional[RetrieverDBModel]:
+        """Find the composite bridge record for a given pipeline."""
+        return (
+            self.db.query(RetrieverDBModel)
+            .filter_by(
+                pipeline_id=pipeline_id,
+                class_name=class_name,
+            )
+            .first()
+        )
+
+    def find_composite_children(self, parent_id: int) -> List[RAGRetrieverChild]:
+        """Return child links ordered by child_order."""
+        return (
+            self.db.query(RAGRetrieverChild)
+            .filter_by(parent_id=parent_id)
+            .order_by(RAGRetrieverChild.child_order)
+            .all()
+        )
+
+    def save_composite(
+        self,
+        class_name: str,
+        pipeline_id: int,
+        child_bridge_ids: List[int],
+    ) -> RetrieverDBModel:
+        """Persist a composite bridge record and its child links.
+
+        Parameters
+        ----------
+        class_name : str
+            "SequentialRetriever" or "ParallelRetriever".
+        pipeline_id : int
+            Owning pipeline.
+        child_bridge_ids : list[int]
+            Ordered list of child bridge ids (rag_retriever.id).
+
+        Returns
+        -------
+        RetrieverDBModel
+            The persisted bridge record.
+        """
+        bridge_record = self.create_bridge(class_name, pipeline_id)
+        for order, child_id in enumerate(child_bridge_ids):
+            self.db.add(
+                RAGRetrieverChild(
+                    parent_id=bridge_record.id,
+                    child_id=child_id,
+                    child_order=order,
+                )
+            )
+        self.db.commit()
+        return bridge_record
+
+    def find_or_create_embedding_model(
+        self,
+        class_name: str,
+        parameters: Dict[str, object],
+    ) -> EmbeddingDBModel:
+        """Return an existing EmbeddingDBModel record or create one.
+
+        This is idempotent: the same (class_name, parameters) pair
+        always resolves to the same record.
+        """
+        parameters = dict(sorted(parameters.items()))
+        existing = (
+            self.db.query(EmbeddingDBModel)
+            .filter_by(
+                class_name=class_name,
+                parameters=parameters,
+            )
+            .first()
+        )
+        if existing is not None:
+            return existing
+        record = EmbeddingDBModel(
+            class_name=class_name,
+            parameters=parameters,
+        )
+        self.db.add(record)
+        self.db.commit()
+        self.db.refresh(record)
+        return record
+
+    # NOTE: All embedding matrices are loaded into memory via np.load()
+    # for simplicity — not suitable for very large document collections.
+    def find_embedding_matrices(
+        self,
+        doc_ids: List[int],
+        chunk_set_id: int,
+        embedding_model_id: int,
+    ) -> Dict[int, EmbeddingMatrixDBModel]:
+        """Return existing embedding matrices keyed by document_id.
+
+        Only matrices that actually exist in the database are returned.
+        """
+        result: Dict[int, EmbeddingMatrixDBModel] = {}
+        for doc_id in doc_ids:
+            matrix = (
+                self.db.query(EmbeddingMatrixDBModel)
+                .filter_by(
+                    document_id=doc_id,
+                    chunk_set_id=chunk_set_id,
+                    embedding_model_id=embedding_model_id,
+                )
+                .first()
+            )
+            if matrix is not None:
+                result[doc_id] = matrix
+        return result
+
+    def save_embedding_matrix(
+        self,
+        document_id: int,
+        chunk_set_id: int,
+        embedding_model_id: int,
+        storage_folder: str,
+        matrix_shape: List[int],
+    ) -> EmbeddingMatrixDBModel:
+        """Persist a new embedding matrix record.
+
+        Returns the persisted model.
+        """
+        record = EmbeddingMatrixDBModel(
+            document_id=document_id,
+            chunk_set_id=chunk_set_id,
+            embedding_model_id=embedding_model_id,
+            storage_folder=storage_folder,
+            matrix_shape=matrix_shape,
+        )
+        self.db.add(record)
+        self.db.commit()
+        self.db.refresh(record)
+        return record
+
+    def find_embedding_matrix(
+        self,
+        document_id: int,
+        chunk_set_id: int,
+        embedding_model_id: int,
+    ) -> Optional[EmbeddingMatrixDBModel]:
+        """Look up a single embedding matrix record."""
+        return (
+            self.db.query(EmbeddingMatrixDBModel)
+            .filter_by(
+                document_id=document_id,
+                chunk_set_id=chunk_set_id,
+                embedding_model_id=embedding_model_id,
+            )
+            .first()
+        )
+
+    def find_sparse_by_bridge_id(self, bridge_id: int) -> Optional[SparseRetrieverDBModel]:
+        """Look up a sparse retriever sub-table record by its bridge_id."""
+        return (
+            self.db.query(SparseRetrieverDBModel)
+            .filter_by(bridge_id=bridge_id)
+            .first()
+        )
+
+    def find_dense_by_bridge_id(self, bridge_id: int) -> Optional[DenseRetrieverDBModel]:
+        """Look up a dense retriever sub-table record by its bridge_id."""
+        return (
+            self.db.query(DenseRetrieverDBModel)
+            .filter_by(bridge_id=bridge_id)
+            .first()
+        )
+
+    def find_bridge_for_sub_table(
+        self,
+        sub_db_model: SparseRetrieverDBModel | DenseRetrieverDBModel,
+        class_name: str,
+    ) -> Optional[RetrieverDBModel]:
+        """Given a sub-table detail record, find its parent bridge record."""
+        bridge = self.db.query(RetrieverDBModel).get(sub_db_model.bridge_id)
+        if bridge is not None and bridge.class_name == class_name:
+            return bridge
+        return None
+
+    def get_bridge_by_id(self, bridge_id: int) -> Optional[RetrieverDBModel]:
+        """Fetch a bridge record by its primary key."""
+        return self.db.query(RetrieverDBModel).get(bridge_id)
diff --git a/DashAI/back/models/RAG/retrievers/sparse/__init__.py b/DashAI/back/models/RAG/retrievers/sparse/__init__.py
new file mode 100644
index 000000000..07e0f9285
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/sparse/__init__.py
@@ -0,0 +1,3 @@
+from DashAI.back.models.RAG.retrievers.sparse.sparse_retriever import SparseRetriever
+from DashAI.back.models.RAG.retrievers.sparse.tfidf_retriever import TFIDFRetriever, TFIDFVectorizerModel
+from DashAI.back.models.RAG.retrievers.sparse.bm25_retriever import BM25Retriever, BM25VectorizerModel
diff --git a/DashAI/back/models/RAG/retrievers/sparse/bm25_retriever.py b/DashAI/back/models/RAG/retrievers/sparse/bm25_retriever.py
new file mode 100644
index 000000000..4d5da12f1
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/sparse/bm25_retriever.py
@@ -0,0 +1,326 @@
+import logging
+import os
+import pickle
+from typing import Dict, List, Tuple
+
+import numpy as np
+from sklearn.metrics.pairwise import pairwise_distances
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    bool_field,
+    component_field,
+    enum_field,
+    float_field,
+    int_field,
+    list_field,
+    none_type,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.base_model import BaseModel
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.sparse.sparse_retriever import SparseRetriever
+
+log = logging.getLogger(__name__)
+
+
+class BM25VectorizerSchema(BaseSchema):
+    strip_accents: schema_field(
+        none_type(enum_field(enum=["ascii", "unicode"])),
+        placeholder=None,
+        description=MultilingualString(
+            en="Remove accents during preprocessing.",
+            es="Eliminar acentos durante el preprocesamiento.",
+        ),
+    )  # type: ignore
+
+    lowercase: schema_field(
+        bool_field(),
+        placeholder=True,
+        description=MultilingualString(
+            en="Convert all characters to lowercase.",
+            es="Convertir todos los caracteres a minúsculas.",
+        ),
+    )  # type: ignore
+
+    stop_words: schema_field(
+        none_type(list_field(string_field(), min_items=1)),
+        placeholder=None,
+        description=MultilingualString(
+            en="List of stop words.",
+            es="Lista de palabras vacías.",
+        ),
+    )  # type: ignore
+
+    max_df: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=1.0,
+        description=MultilingualString(
+            en="Ignore terms with document frequency above this threshold.",
+            es="Ignorar términos con frecuencia de documento superior a este umbral.",
+        ),
+    )  # type: ignore
+
+    min_df: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.0,
+        description=MultilingualString(
+            en="Ignore terms with document frequency below this threshold.",
+            es="Ignorar términos con frecuencia de documento inferior a este umbral.",
+        ),
+    )  # type: ignore
+
+    max_features: schema_field(
+        none_type(int_field(ge=1)),
+        placeholder=None,
+        description=MultilingualString(
+            en="Maximum number of features.",
+            es="Número máximo de características.",
+        ),
+    )  # type: ignore
+
+class BM25VectorizerModel(BaseModel):
+    DISPLAY_NAME: str = MultilingualString(
+        en="BM25 Vectorizer",
+        es="Vectorizador BM25",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Vectorizer for BM25Retriever using CountVectorizer with BM25-specific parameters.",
+        es="Vectorizador para BM25Retriever usando CountVectorizer con parámetros específicos de BM25.",
+    )
+
+    SCHEMA = BM25VectorizerSchema
+
+    def __init__(self, **kwargs):
+        from sklearn.feature_extraction.text import CountVectorizer
+
+        validated = self.SCHEMA.model_validate(kwargs)
+        self.params = dict(validated)
+        self.vectorizer = CountVectorizer(
+            strip_accents=self.params.pop("strip_accents"),
+            lowercase=self.params.pop("lowercase"),
+            stop_words=self.params.pop("stop_words"),
+            max_df=self.params.pop("max_df"),
+            min_df=self.params.pop("min_df"),
+            max_features=self.params.pop("max_features"),
+        )
+
+    def load(self):
+        pass
+
+    def save(self):
+        pass
+
+    def train(self):
+        pass
+
+class BM25RetrieverSchema(BaseSchema):
+    BM25Vectorizer: schema_field(
+        component_field(parent="BM25VectorizerModel"),
+        placeholder={"component": "BM25VectorizerModel", "params": {}},
+        description=MultilingualString(
+            en="BM25 Vectorizer parameters.",
+            es="Parámetros del vectorizador BM25.",
+        ),
+    )  # type: ignore
+
+    k1: schema_field(
+        float_field(ge=0.0),
+        placeholder=1.5,
+        description=MultilingualString(
+            en="BM25 k1 parameter: term frequency saturation.",
+            es="Parámetro k1 de BM25: saturación de frecuencia de término.",
+        ),
+    )  # type: ignore
+
+    b: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.75,
+        description=MultilingualString(
+            en="BM25 b parameter: length normalization.",
+            es="Parámetro b de BM25: normalización de longitud.",
+        ),
+    )  # type: ignore
+
+    delta: schema_field(
+        float_field(ge=0.0),
+        placeholder=0.0,
+        description=MultilingualString(
+            en="BM25 delta parameter for IDF smoothing.",
+            es="Parámetro delta de BM25 para suavizado de IDF.",
+        ),
+    )  # type: ignore
+
+    similarity_function: schema_field(
+        enum_field(["cityblock", "cosine", "euclidean", "l1", "l2", "manhattan"]),
+        placeholder="cosine",
+        description=MultilingualString(
+            en="Distance metric for comparing BM25-weighted vectors.",
+            es="Métrica de distancia para comparar vectores ponderados BM25.",
+        ),
+    )  # type: ignore
+
+    top_k: schema_field(
+        int_field(ge=1),
+        placeholder=5,
+        description=MultilingualString(
+            en="Number of chunks to select.",
+            es="Número de fragmentos a seleccionar.",
+        ),
+    )  # type: ignore
+
+
+class BM25Retriever(SparseRetriever):
+    FLAGS: list[str] = ["keyword", "sparse"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="BM25 Retriever",
+        es="Recuperador BM25",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Sparse retriever using BM25 (Okapi) ranking for document retrieval.",
+        es="Recuperador disperso que usa ranking BM25 (Okapi) para recuperar documentos.",
+    )
+
+    SCHEMA = BM25RetrieverSchema
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        self.k1 = self.params.pop("k1")
+        self.b = self.params.pop("b")
+        self.delta = self.params.pop("delta")
+        self.similarity_function_name = self.params.pop("similarity_function")
+        self._top_k = self.params.pop("top_k")
+
+        vectorizer_model = self.params.pop("BM25Vectorizer")
+        self._vectorizer = vectorizer_model.vectorizer
+
+    def init_model(self) -> None:
+        if not self.load():
+            self._fit()
+
+    def load(self) -> bool:
+        if self._persistence.model_dir is None:
+            return False
+        model_dir = self._persistence.model_dir
+        try:
+            with open(os.path.join(model_dir, "bm25_vectorizer.pkl"), "rb") as f:
+                self._vectorizer = pickle.load(f)
+            with open(os.path.join(model_dir, "bm25_tf_matrix.pkl"), "rb") as f:
+                self._tf_matrix = pickle.load(f)
+            with open(os.path.join(model_dir, "bm25_matrix.pkl"), "rb") as f:
+                self._bm25_matrix = pickle.load(f)
+            with open(os.path.join(model_dir, "bm25_row_to_chunk.pkl"), "rb") as f:
+                self.matrix_row_to_chunk_map = pickle.load(f)
+            return True
+        except Exception as e:
+            log.info("Error loading BM25 state: %s", e)
+            return False
+
+    def save(self) -> None:
+        model_dir = self._persistence.model_dir
+        if model_dir is None:
+            raise ValueError(
+                "Cannot save BM25Retriever: persistence.model_dir is None."
+            )
+        os.makedirs(model_dir, exist_ok=True)
+        to_save = {
+            "bm25_vectorizer.pkl": self._vectorizer,
+            "bm25_tf_matrix.pkl": self._tf_matrix,
+            "bm25_matrix.pkl": self._bm25_matrix,
+            "bm25_row_to_chunk.pkl": self.matrix_row_to_chunk_map,
+        }
+        for fname, obj in to_save.items():
+            with open(os.path.join(model_dir, fname), "wb") as f:
+                pickle.dump(obj, f)
+
+    def _fit(self):
+        chunk_texts = []
+        current_idx = 0
+        self.matrix_row_to_chunk_map: Dict[int, Chunk] = {}
+
+        for doc_chunks in self.chunks.values():
+            for chunk in doc_chunks.values():
+                chunk_texts.append(chunk.text)
+                self.matrix_row_to_chunk_map[current_idx] = chunk
+                current_idx += 1
+
+        self._vectorizer.fit(chunk_texts)
+        self._tf_matrix = self._vectorizer.transform(chunk_texts)
+
+        tf = self._tf_matrix.copy()
+        n_docs = tf.shape[0]
+        doc_lengths = np.array(tf.sum(axis=1)).flatten()
+        avgdl = np.mean(doc_lengths)
+
+        df = np.array((tf > 0).sum(axis=0)).flatten()
+        idf = np.log((n_docs - df + 0.5) / (df + 0.5) + 1.0)
+        idf += self.delta
+        idf[idf < 0] = 0
+
+        from scipy.sparse import lil_matrix
+
+        bm25 = lil_matrix(tf.shape)
+        for i in range(tf.shape[0]):
+            row = tf[i]
+            length_norm = 1 - self.b + self.b * (doc_lengths[i] / avgdl)
+            for j in range(row.indptr[0], row.indptr[1]):
+                col = row.indices[j]
+                tf_val = row.data[j]
+                numerator = tf_val * (self.k1 + 1)
+                denominator = tf_val + self.k1 * length_norm
+                bm25[i, col] = idf[col] * numerator / denominator
+
+        self._bm25_matrix = bm25.tocsr()
+
+    @property
+    def retrieval_top_k(self) -> int:
+        return self._top_k
+
+    def retrieve(self, query: str, top_k: int | None = None) -> List[Chunk]:
+        self._check_infra()
+        k = top_k if top_k is not None else self._top_k
+        query_vec = self._vectorizer.transform([query])
+        distances = pairwise_distances(
+            query_vec, self._bm25_matrix,
+            metric=self.similarity_function_name,
+        ).flatten()
+        top_indices = np.argsort(distances)[:k]
+        return [self.matrix_row_to_chunk_map[idx] for idx in top_indices]
+
+    def score_chunks(self, chunk_ids: List[int], query: str) -> List[Tuple[int, float]]:
+        self._check_infra()
+        query_vec = self._vectorizer.transform([query])
+        chunk_id_to_row = {c.id: r for r, c in self.matrix_row_to_chunk_map.items()}
+        rows, valid_ids = [], []
+        for cid in chunk_ids:
+            row = chunk_id_to_row.get(cid)
+            if row is not None:
+                rows.append(row)
+                valid_ids.append(cid)
+        if not rows:
+            return []
+        chunk_vectors = self._bm25_matrix[rows]
+        distances = pairwise_distances(
+            query_vec, chunk_vectors,
+            metric=self.similarity_function_name,
+        ).flatten()
+        scored = list(zip(valid_ids, distances.tolist()))
+        scored.sort(key=lambda x: x[1])
+        return scored
+
+    def get_chunk_vectors(self, chunk_ids: List[int]) -> np.ndarray:
+        chunk_id_to_row = {c.id: r for r, c in self.matrix_row_to_chunk_map.items()}
+        rows = []
+        for cid in chunk_ids:
+            row = chunk_id_to_row.get(cid)
+            if row is not None:
+                rows.append(row)
+        if not rows:
+            raise ValueError(
+                f"None of the provided chunk_ids {chunk_ids} were found "
+                "in the BM25 matrix."
+            )
+        return self._bm25_matrix[rows].toarray()
diff --git a/DashAI/back/models/RAG/retrievers/sparse/sparse_retriever.py b/DashAI/back/models/RAG/retrievers/sparse/sparse_retriever.py
new file mode 100644
index 000000000..2761374eb
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/sparse/sparse_retriever.py
@@ -0,0 +1,14 @@
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.RAG.retrievers.unit_retriever import UnitRetriever
+
+
+class SparseRetriever(UnitRetriever):
+    FLAGS: list[str] = ["abstract"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="Keyword Retriever",
+        es="Recuperador por Palabras Clave",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Sparse retriever using term-frequency based representations.",
+        es="Recuperador disperso que usa representaciones basadas en frecuencia de términos.",
+    )
diff --git a/DashAI/back/models/RAG/retrievers/sparse/tfidf_retriever.py b/DashAI/back/models/RAG/retrievers/sparse/tfidf_retriever.py
new file mode 100644
index 000000000..8af4749c9
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/sparse/tfidf_retriever.py
@@ -0,0 +1,334 @@
+import logging
+import os
+import pickle
+from typing import Dict, List, Tuple
+
+import numpy as np
+from sklearn.feature_extraction.text import TfidfVectorizer
+from sklearn.metrics.pairwise import pairwise_distances
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    bool_field,
+    component_field,
+    enum_field,
+    float_field,
+    int_field,
+    list_field,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.base_model import BaseModel
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.sparse.sparse_retriever import SparseRetriever
+
+log = logging.getLogger(__name__)
+
+
+class TFIDFVectorizerSchema(BaseSchema):
+    strip_accents: schema_field(
+        enum_field(enum=["ascii", "unicode", "None"]),
+        placeholder="None",
+        description=MultilingualString(
+            en="Whether to strip accents from the text.",
+            es="Si se deben eliminar los acentos del texto.",
+        ),
+    )  # type: ignore
+
+    lowercase: schema_field(
+        bool_field(),
+        placeholder=True,
+        description=MultilingualString(
+            en="Whether to convert all characters to lowercase.",
+            es="Si se deben convertir todos los caracteres a minúsculas.",
+        ),
+    )  # type: ignore
+
+    analyzer: schema_field(
+        enum_field(enum=["word", "char", "char_wb"]),
+        placeholder="word",
+        description=MultilingualString(
+            en="Whether the feature should be made of word or character n-grams.",
+            es="Si las características deben ser n-gramas de palabras o caracteres.",
+        ),
+    )  # type: ignore
+
+    stop_words: schema_field(
+        list_field(string_field(), min_items=0),
+        placeholder=[],
+        description=MultilingualString(
+            en="List of stop words. Leave empty to use none.",
+            es="Lista de palabras vacías. Dejar vacío para no usar ninguna.",
+        ),
+    )  # type: ignore
+
+    ngram_range: schema_field(
+        list_field(int_field(), min_items=2, max_items=2),
+        placeholder=[1, 1],
+        description=MultilingualString(
+            en="Lower and upper boundary of the n-gram range.",
+            es="Límite inferior y superior del rango de n-gramas.",
+        ),
+    )  # type: ignore
+
+    max_df: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=1.0,
+        description=MultilingualString(
+            en="Ignore terms with document frequency above this threshold.",
+            es="Ignorar términos con frecuencia de documento superior a este umbral.",
+        ),
+    )  # type: ignore
+
+    min_df: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.0,
+        description=MultilingualString(
+            en="Ignore terms with document frequency below this threshold.",
+            es="Ignorar términos con frecuencia de documento inferior a este umbral.",
+        ),
+    )  # type: ignore
+
+    max_features: schema_field(
+        int_field(ge=0),
+        placeholder=1000,
+        description=MultilingualString(
+            en="Maximum number of features. 0 means no limit.",
+            es="Número máximo de características. 0 significa sin límite.",
+        ),
+    )  # type: ignore
+
+    norm: schema_field(
+        enum_field(enum=["l1", "l2", "None"]),
+        placeholder="l2",
+        description=MultilingualString(
+            en="Norm used to normalize term vectors.",
+            es="Norma utilizada para normalizar los vectores de términos.",
+        ),
+    )  # type: ignore
+
+    use_idf: schema_field(
+        bool_field(),
+        placeholder=True,
+        description=MultilingualString(
+            en="Enable inverse-document-frequency reweighting.",
+            es="Activar reponderación por frecuencia inversa de documento.",
+        ),
+    )  # type: ignore
+
+    smooth_idf: schema_field(
+        bool_field(),
+        placeholder=True,
+        description=MultilingualString(
+            en="Smooth IDF weights to prevent zero divisions.",
+            es="Suavizar pesos IDF para prevenir divisiones por cero.",
+        ),
+    )  # type: ignore
+
+    sublinear_tf: schema_field(
+        bool_field(),
+        placeholder=False,
+        description=MultilingualString(
+            en="Apply sublinear TF scaling (1 + log(tf)).",
+            es="Aplicar escalado sublineal de TF (1 + log(tf)).",
+        ),
+    )  # type: ignore
+
+
+class TFIDFVectorizerModel(BaseModel):
+    SCHEMA = TFIDFVectorizerSchema
+    DISPLAY_NAME: str = MultilingualString(
+        en="TF-IDF Vectorizer Model",
+        es="Modelo de Vectorización TF-IDF",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Model component that encapsulates a TF-IDF vectorizer.",
+        es="Componente del modelo que encapsula un vectorizador TF-IDF.",
+    )
+
+    def __init__(self, **kwargs) -> None:
+        validated = self.SCHEMA.model_validate(kwargs)
+        self.params = dict(validated)
+        if self.params.get("strip_accents") == "None":
+            self.params["strip_accents"] = None
+        stop_words = self.params.get("stop_words") or None
+        ngram_range = tuple(self.params.get("ngram_range"))
+        self.model = TfidfVectorizer(
+            strip_accents=self.params.get("strip_accents"),
+            lowercase=self.params.get("lowercase"),
+            analyzer=self.params.get("analyzer"),
+            stop_words=stop_words,
+            ngram_range=ngram_range,
+            max_df=self.params.get("max_df"),
+            min_df=self.params.get("min_df"),
+            max_features=self.params.get("max_features"),
+            norm=self.params.get("norm"),
+            use_idf=self.params.get("use_idf"),
+            smooth_idf=self.params.get("smooth_idf"),
+            sublinear_tf=self.params.get("sublinear_tf"),
+        )
+
+    def save(self, filename: str = "") -> None:
+        pass
+
+    def load(self, filename: str = "") -> None:
+        pass
+
+    def train(self, **kwargs):
+        return
+
+
+class TFIDFRetrieverSchema(BaseSchema):
+    TFIDFVectorizer: schema_field(
+        component_field(parent="TFIDFVectorizerModel"),
+        placeholder={"component": "TFIDFVectorizerModel", "params": {}},
+        description=MultilingualString(
+            en="TF-IDF Vectorizer parameters.",
+            es="Parámetros del vectorizador TF-IDF.",
+        ),
+    )  # type: ignore
+
+    similarity_function: schema_field(
+        enum_field(["cityblock", "cosine", "euclidean", "l1", "l2", "manhattan"]),
+        placeholder="cosine",
+        description=MultilingualString(
+            en="Distance metric for comparing TF-IDF vectors.",
+            es="Métrica de distancia para comparar vectores TF-IDF.",
+        ),
+    )  # type: ignore
+
+    top_k: schema_field(
+        int_field(ge=1),
+        placeholder=5,
+        description=MultilingualString(
+            en="Number of chunks to select.",
+            es="Número de fragmentos a seleccionar.",
+        ),
+    )  # type: ignore
+
+
+class TFIDFRetriever(SparseRetriever):
+    FLAGS: list[str] = ["keyword", "sparse"]
+    DISPLAY_NAME: str = MultilingualString(
+        en="TF-IDF Retriever",
+        es="Recuperador TF-IDF",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="Sparse retriever using TF-IDF vectorization for document retrieval.",
+        es="Recuperador disperso que usa vectorización TF-IDF para recuperar documentos.",
+    )
+
+    SCHEMA = TFIDFRetrieverSchema
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+        vectorizer_model = self.params.pop("TFIDFVectorizer")
+        self._vectorizer = vectorizer_model.model
+        self.similarity_function_name = self.params.pop("similarity_function")
+        self._top_k = self.params.pop("top_k")
+
+    def init_model(self) -> None:
+        if not self.load():
+            self._fit()
+
+    def load(self) -> bool:
+        if self._persistence.model_dir is None:
+            return False
+        model_dir = self._persistence.model_dir
+        try:
+            with open(os.path.join(model_dir, "tfidf_vectorizer.pkl"), "rb") as f:
+                self._vectorizer = pickle.load(f)
+            with open(os.path.join(model_dir, "tf_idf_matrix.pkl"), "rb") as f:
+                self._tf_idf_matrix = pickle.load(f)
+            with open(
+                os.path.join(model_dir, "matrix_row_to_chunk_map.pkl"), "rb"
+            ) as f:
+                self.matrix_row_to_chunk_map = pickle.load(f)
+            return True
+        except Exception as e:
+            log.info("Error loading TFIDF state: %s", e)
+            return False
+
+    def save(self) -> None:
+        model_dir = self._persistence.model_dir
+        if model_dir is None:
+            raise ValueError(
+                "Cannot save TFIDFRetriever: persistence.model_dir is None."
+            )
+        os.makedirs(model_dir, exist_ok=True)
+        to_save = {
+            "tfidf_vectorizer.pkl": self._vectorizer,
+            "tf_idf_matrix.pkl": self._tf_idf_matrix,
+            "matrix_row_to_chunk_map.pkl": self.matrix_row_to_chunk_map,
+        }
+        for fname, obj in to_save.items():
+            with open(os.path.join(model_dir, fname), "wb") as f:
+                pickle.dump(obj, f)
+
+    def _fit(self):
+        chunk_texts = []
+        current_idx = 0
+        self.matrix_row_to_chunk_map: Dict[int, Chunk] = {}
+
+        for doc_chunks in self.chunks.values():
+            for chunk in doc_chunks.values():
+                chunk_texts.append(chunk.text)
+                self.matrix_row_to_chunk_map[current_idx] = chunk
+                current_idx += 1
+
+        self._tf_idf_matrix = self._vectorizer.fit_transform(chunk_texts)
+
+    @property
+    def retrieval_top_k(self) -> int:
+        return self._top_k
+
+
+
+    def retrieve(self, query: str, top_k: int | None = None) -> List[Chunk]:
+        self._check_infra()
+        k = top_k if top_k is not None else self._top_k
+        query_vector = self._vectorizer.transform([query])
+        distances = pairwise_distances(
+            query_vector,
+            self._tf_idf_matrix,
+            metric=self.similarity_function_name,
+        ).flatten()
+        top_indices = np.argsort(distances)[:k]
+        return [self.matrix_row_to_chunk_map[idx] for idx in top_indices]
+
+    def score_chunks(self, chunk_ids: List[int], query: str) -> List[Tuple[int, float]]:
+        self._check_infra()
+        query_vector = self._vectorizer.transform([query])
+        chunk_id_to_row = {c.id: r for r, c in self.matrix_row_to_chunk_map.items()}
+        rows, valid_ids = [], []
+        for cid in chunk_ids:
+            row = chunk_id_to_row.get(cid)
+            if row is not None:
+                rows.append(row)
+                valid_ids.append(cid)
+        if not rows:
+            return []
+        distances = pairwise_distances(
+            query_vector,
+            self._tf_idf_matrix[rows],
+            metric=self.similarity_function_name,
+        ).flatten()
+        scored = list(zip(valid_ids, distances.tolist()))
+        scored.sort(key=lambda x: x[1])
+        return scored
+
+    def get_chunk_vectors(self, chunk_ids: List[int]) -> np.ndarray:
+        chunk_id_to_row = {c.id: r for r, c in self.matrix_row_to_chunk_map.items()}
+        rows = []
+        for cid in chunk_ids:
+            row = chunk_id_to_row.get(cid)
+            if row is not None:
+                rows.append(row)
+        if not rows:
+            raise ValueError(
+                f"None of the provided chunk_ids {chunk_ids} were found "
+                "in the TF-IDF matrix."
+            )
+        return self._tf_idf_matrix[rows].toarray()
diff --git a/DashAI/back/models/RAG/retrievers/unit_retriever.py b/DashAI/back/models/RAG/retrievers/unit_retriever.py
new file mode 100644
index 000000000..c55bf8c84
--- /dev/null
+++ b/DashAI/back/models/RAG/retrievers/unit_retriever.py
@@ -0,0 +1,53 @@
+from abc import ABC
+from typing import Any, Dict, Final, List
+
+from DashAI.back.models.RAG.documents import Chunk
+from DashAI.back.models.RAG.retrievers.persistence import (
+    DensePersistence,
+    SparsePersistence,
+)
+from DashAI.back.models.RAG.retrievers.retriever_model import RetrieverModel
+
+
+class UnitRetriever(RetrieverModel, ABC):
+    """Leaf: abstract base for unit retrievers."""
+
+    TYPE: Final[str] = "RetrieverModel"
+    FLAGS: list[str] = ["abstract"]
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+    def inject_infra(
+        self,
+        env_rag_path: str,
+        chunks: Dict[int, Dict[int, Chunk]],
+        persistence: Any,
+    ) -> None:
+        if not isinstance(persistence, (SparsePersistence, DensePersistence)):
+            raise TypeError(
+                f"Expected SparsePersistence or DensePersistence, "
+                f"got {type(persistence).__name__}"
+            )
+        super().inject_infra(env_rag_path, chunks, persistence)
+
+    def _check_infra(self) -> None:
+        """Raise if infrastructure has not been injected."""
+        if self._persistence is None:
+            raise RuntimeError(
+                f"{self.__class__.__name__}: infrastructure not injected. "
+                "Call inject_infra() before retrieve()."
+            )
+
+    def add(self, child: RetrieverModel) -> None:
+        raise TypeError(
+            f"{self.__class__.__name__} is a unit retriever and cannot contain children."
+        )
+
+    def remove(self, child: RetrieverModel) -> None:
+        raise TypeError(
+            f"{self.__class__.__name__} is a unit retriever and cannot contain children."
+        )
+
+    def get_children(self) -> List[RetrieverModel]:
+        return []
diff --git a/DashAI/back/models/RAG/utils.py b/DashAI/back/models/RAG/utils.py
new file mode 100644
index 000000000..300162d3d
--- /dev/null
+++ b/DashAI/back/models/RAG/utils.py
@@ -0,0 +1,18 @@
+import hashlib
+
+
+def hash_function(content: str | bytes) -> str:
+    """Generate a SHA-256 hash for the given content bytes.
+
+    Args:
+        content: The content to hash (str or bytes).
+
+    Returns:
+        A hex-encoded SHA-256 hash string.
+
+    Raises:
+        UnicodeEncodeError: If a str content cannot be encoded as UTF-8.
+    """
+    if isinstance(content, str):
+        content = content.encode("utf-8")
+    return hashlib.sha256(content).hexdigest()
diff --git a/DashAI/back/models/__init__.py b/DashAI/back/models/__init__.py
index 9c0fa90a1..4f745c4db 100644
--- a/DashAI/back/models/__init__.py
+++ b/DashAI/back/models/__init__.py
@@ -1 +1,52 @@
 # flake8: noqa
+from DashAI.back.models.base_generative_model import BaseGenerativeModel
+from DashAI.back.models.base_model import BaseModel
+from DashAI.back.models.hugging_face.distilbert_transformer import DistilBertTransformer
+from DashAI.back.models.hugging_face.opus_mt_en_es_transformer import (
+    OpusMtEnESTransformer,
+)
+from DashAI.back.models.hugging_face.qwen_model import QwenModel
+from DashAI.back.models.hugging_face.stable_diffusion_v1_depth_controlnet import (
+    StableDiffusionXLV1ControlNet,
+)
+from DashAI.back.models.hugging_face.stable_diffusion_v2_model import (
+    StableDiffusionV2Model,
+)
+from DashAI.back.models.hugging_face.stable_diffusion_v3_model import (
+    StableDiffusionV3Model,
+)
+from DashAI.back.models.scikit_learn.bow_text_classification_model import (
+    BagOfWordsTextClassificationModel,
+)
+from DashAI.back.models.scikit_learn.decision_tree_classifier import (
+    DecisionTreeClassifier,
+)
+from DashAI.back.models.scikit_learn.dummy_classifier import DummyClassifier
+from DashAI.back.models.scikit_learn.gradient_boosting_regression import (
+    GradientBoostingR,
+)
+from DashAI.back.models.scikit_learn.hist_gradient_boosting_classifier import (
+    HistGradientBoostingClassifier,
+)
+from DashAI.back.models.scikit_learn.k_neighbors_classifier import KNeighborsClassifier
+from DashAI.back.models.scikit_learn.linear_regression import LinearRegression
+from DashAI.back.models.scikit_learn.linearSVR import LinearSVR
+from DashAI.back.models.scikit_learn.logistic_regression import LogisticRegression
+from DashAI.back.models.scikit_learn.mlp_regression import MLPRegression
+from DashAI.back.models.scikit_learn.random_forest_classifier import (
+    RandomForestClassifier,
+)
+from DashAI.back.models.scikit_learn.random_forest_regression import (
+    RandomForestRegression,
+)
+from DashAI.back.models.scikit_learn.ridge_regression import RidgeRegression
+from DashAI.back.models.scikit_learn.sklearn_like_classifier import (
+    SklearnLikeClassifier,
+)
+from DashAI.back.models.scikit_learn.sklearn_like_model import SklearnLikeModel
+from DashAI.back.models.scikit_learn.sklearn_like_regressor import SklearnLikeRegressor
+from DashAI.back.models.scikit_learn.svc import SVC
+
+from DashAI.back.models.RAG import RAGPipeline
+from DashAI.back.models.remote_models import OpenAITextToTextGenerationModel
+from DashAI.back.models.hugging_face.phi_4_mini_instruct_model import Phi4MiniInstructModel
diff --git a/DashAI/back/models/base_model.py b/DashAI/back/models/base_model.py
index 6cebb7818..71bf97bbc 100644
--- a/DashAI/back/models/base_model.py
+++ b/DashAI/back/models/base_model.py
@@ -26,6 +26,7 @@ class BaseModel(ConfigObject, metaclass=ABCMeta):
     """
 
     TYPE: Final[str] = "Model"
+    FLAGS: list[str] = []
     DISPLAY_NAME: str = ""
     DESCRIPTION: str = ""
     COLOR: str = "#795548"
diff --git a/DashAI/back/models/hugging_face/__init__.py b/DashAI/back/models/hugging_face/__init__.py
index e69de29bb..73da59ea1 100644
--- a/DashAI/back/models/hugging_face/__init__.py
+++ b/DashAI/back/models/hugging_face/__init__.py
@@ -0,0 +1,3 @@
+from DashAI.back.models.hugging_face.deep_seek_model import DeepSeekModel
+from DashAI.back.models.hugging_face.qwen_model import QwenModel
+from DashAI.back.models.hugging_face.phi_4_mini_instruct_model import Phi4MiniInstructModel
\ No newline at end of file
diff --git a/DashAI/back/models/hugging_face/deep_seek_model.py b/DashAI/back/models/hugging_face/deep_seek_model.py
new file mode 100644
index 000000000..ae8a45dad
--- /dev/null
+++ b/DashAI/back/models/hugging_face/deep_seek_model.py
@@ -0,0 +1,93 @@
+from typing import List
+
+from llama_cpp import Llama
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    float_field,
+    int_field,
+    schema_field,
+)
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+
+
+class DeepSeekSchema(BaseSchema):
+    """Schema for DeepSeek model."""
+
+    max_tokens: schema_field(
+        int_field(ge=1),
+        placeholder=100,
+        description="Maximum number of tokens to generate.",
+    )  # type: ignore
+
+    temperature: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.7,
+        description=(
+            "Sampling temperature. Higher values make the output more random, "
+            "while lower values make it more focused and deterministic."
+        ),
+    )  # type: ignore
+
+    frequency_penalty: schema_field(
+        float_field(ge=0.0, le=2.0),
+        placeholder=0.1,
+        description=(
+            "Penalty for repeated tokens in the output. Higher values reduce the "
+            "likelihood of repetition, encouraging more diverse text generation."
+        ),
+    )  # type: ignore
+
+    n_ctx: schema_field(
+        int_field(ge=1),
+        placeholder=4096,
+        description=(
+            "Maximum number of tokens the model can process in a single forward pass "
+            "(context window size)."
+        ),
+    )  # type: ignore
+
+
+class DeepSeekModel(TextToTextGenerationTaskModel):
+    """DeepSeek model for text generation using llama.cpp library."""
+
+    SCHEMA = DeepSeekSchema
+
+    def __init__(self, **kwargs):
+        kwargs = self.validate_and_transform(kwargs)
+        self.max_tokens = kwargs.pop("max_tokens", 100)
+        self.temperature = kwargs.pop("temperature", 0.7)
+        self.frequency_penalty = kwargs.pop("frequency_penalty", 0.1)
+        self.n_ctx = kwargs.pop("n_ctx", 512)
+
+        self.model_id = "TheBloke/deepseek-llm-7B-base-GGUF"
+        self.filename = "*Q3_K_S.gguf"
+
+        self.model = Llama.from_pretrained(
+            repo_id=self.model_id,
+            filename=self.filename,
+            verbose=True,
+            n_ctx=self.n_ctx,
+            n_gpu_layers=-1,
+        )
+
+    def generate(self, prompt: str) -> List[str]:
+        if len(prompt) > self.model.n_ctx():
+            prompt = prompt[-self.model.n_ctx() :]
+
+        """Generate text based on prompts."""
+
+        output = self.model(
+            prompt,
+            max_tokens=self.max_tokens,
+            temperature=self.temperature,
+            frequency_penalty=self.frequency_penalty,
+            stop=["Q:"],
+            echo=False,
+        )
+
+        generated_text = output["choices"][0]["text"]
+        clean_text = generated_text.replace(prompt, "").strip()
+        return [clean_text]
diff --git a/DashAI/back/models/hugging_face/llama_model.py b/DashAI/back/models/hugging_face/llama_model.py
index 18f36239b..425560dca 100644
--- a/DashAI/back/models/hugging_face/llama_model.py
+++ b/DashAI/back/models/hugging_face/llama_model.py
@@ -21,6 +21,8 @@
     "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF": "*Q4_K_M.gguf",
     "bartowski/Llama-3.2-1B-Instruct-GGUF": "*Q4_K_M.gguf",
     "bartowski/Llama-3.2-3B-Instruct-GGUF": "*Q4_K_M.gguf",
+    "TheBloke/Llama-2-7B-Chat-GGUF": "*Q4_K_M.gguf",
+    "TheBloke/Llama-2-13B-Chat-GGUF": "*Q4_K_M.gguf",
 }
 
 
@@ -39,6 +41,8 @@ class LlamaSchema(BaseSchema):
                 "bartowski/Meta-Llama-3.1-8B-Instruct-GGUF",
                 "bartowski/Llama-3.2-1B-Instruct-GGUF",
                 "bartowski/Llama-3.2-3B-Instruct-GGUF",
+                "TheBloke/Llama-2-7B-Chat-GGUF",
+                "TheBloke/Llama-2-13B-Chat-GGUF",
             ]
         ),
         placeholder="bartowski/Llama-3.2-3B-Instruct-GGUF",
@@ -50,7 +54,9 @@ class LlamaSchema(BaseSchema):
                 "'Llama-3.2-3B' (~3B parameters) offers a good speed/quality "
                 "trade-off. "
                 "'Meta-Llama-3.1-8B' (~8B parameters) delivers the highest quality "
-                "at the cost of more RAM and slower inference."
+                "at the cost of more RAM and slower inference. "
+                "Llama-2-7B-Chat and Llama-2-13B-Chat are also available via "
+                "TheBloke's community quantizations."
             ),
             es=(
                 "El checkpoint Meta Llama 3.x Instruct a cargar en formato GGUF "
@@ -59,7 +65,9 @@ class LlamaSchema(BaseSchema):
                 "ideal para sistemas solo con CPU. "
                 "'Llama-3.2-3B' (~3B parámetros) ofrece un buen equilibrio entre "
                 "velocidad y calidad. 'Meta-Llama-3.1-8B' (~8B parámetros) entrega "
-                "la mayor calidad a costa de más RAM e inferencia más lenta."
+                "la mayor calidad a costa de más RAM e inferencia más lenta. "
+                "Llama-2-7B-Chat y Llama-2-13B-Chat también están disponibles "
+                "mediante las cuantizaciones comunitarias de TheBloke."
             ),
             pt=(
                 "O checkpoint Meta Llama 3.x Instruct para carregar em formato GGUF "
@@ -69,7 +77,9 @@ class LlamaSchema(BaseSchema):
                 "'Llama-3.2-3B' (~3B parâmetros) oferece um bom equilíbrio entre "
                 "velocidade e qualidade. 'Meta-Llama-3.1-8B' (~8B parâmetros) "
                 "entrega a maior qualidade ao custo de mais RAM e "
-                "inferência mais lenta."
+                "inferência mais lenta. "
+                "Llama-2-7B-Chat e Llama-2-13B-Chat também estão disponíveis "
+                "via quantizações comunitárias de TheBloke."
             ),
             de=(
                 "Der im GGUF-Format zu ladende Meta Llama 3.x Instruct-Checkpoint "
@@ -97,6 +107,44 @@ class LlamaSchema(BaseSchema):
         ),
     )  # type: ignore
 
+    quantization: schema_field(
+        enum_field(
+            enum=[
+                "Q2_K", "Q2_K_L",
+                "Q3_K_S", "Q3_K_M", "Q3_K_L", "Q3_K_XL",
+                "Q4_K_S", "Q4_K_0", "Q4_K_M",
+                "Q5_K_S", "Q5_K_M",
+                "Q6_K", "Q6_K_L",
+                "Q8_0",
+                "F32"
+            ],),
+        placeholder="Q4_K_M",
+        description=MultilingualString(
+            en=(
+                "Quantization format for the GGUF checkpoint. 'Q4_K_M' is the "
+                "default and recommended for most users, offering a good balance "
+                "between speed and quality. Other formats may be faster or smaller "
+                "but can reduce output quality."
+            ),
+            es=(
+                "Formato de cuantización para el checkpoint GGUF. 'Q4_K_M' es el "
+                "predeterminado y recomendado para la mayoría de los usuarios, "
+                "ofreciendo un buen equilibrio entre velocidad y calidad. Otros "
+                "formatos pueden ser más rápidos o más pequeños pero pueden reducir "
+                "la calidad de salida."
+            ),
+            pt=(
+                "Formato de quantização para o checkpoint GGUF. 'Q4_K_M' é o "
+                "padrão e recomendado para a maioria dos usuários, oferecendo um "
+                "bom equilíbrio entre velocidade e qualidade. Outros formatos podem "
+                "ser mais rápidos ou menores, mas podem reduzir a qualidade da saída."
+            ),
+        ),
+        alias=MultilingualString(
+            en="Quantization", es="Cuantización", pt="Quantização"
+        ),
+    )  # type: ignore
+
     max_tokens: schema_field(
         int_field(ge=1),
         placeholder=100,
@@ -448,7 +496,10 @@ def __init__(self, **kwargs):
         self.frequency_penalty = kwargs.pop("frequency_penalty", 0.1)
         self.n_ctx = kwargs.pop("context_window", 512)
 
-        self.filename = LLAMA_FILENAME_MAP.get(self.model_name, "*Q4_K_M.gguf")
+        #self.filename = LLAMA_FILENAME_MAP.get(self.model_name, "*Q4_K_M.gguf")
+        self.quantization = kwargs.get("quantization", "Q4_K_M")
+        self.filename = f"*{self.quantization}.gguf"
+
         use_gpu = LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) >= 0
 
         self.model = Llama.from_pretrained(
diff --git a/DashAI/back/models/hugging_face/phi_4_mini_instruct_model.py b/DashAI/back/models/hugging_face/phi_4_mini_instruct_model.py
new file mode 100644
index 000000000..09a2747a8
--- /dev/null
+++ b/DashAI/back/models/hugging_face/phi_4_mini_instruct_model.py
@@ -0,0 +1,244 @@
+from typing import List
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    float_field,
+    int_field,
+    schema_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+from DashAI.back.models.utils import (
+    LLAMA_DEVICE_ENUM,
+    LLAMA_DEVICE_PLACEHOLDER,
+    LLAMA_DEVICE_TO_IDX,
+)
+
+
+class Phi4MiniInstructSchema(BaseSchema):
+    """Schema for Phi 4 Mini Instruct model hyperparameters"""
+
+    model_name: schema_field(
+        enum_field(
+            enum=[
+                "unsloth/Phi-4-mini-instruct-GGUF",
+            ]
+        ),
+        placeholder="unsloth/Phi-4-mini-instruct-GGUF",
+        description=MultilingualString(
+            en=("Phi-4-mini-instruct is a lightweight open model built upon synthetic "
+               "data and filtered publicly available websites - with a focus on high-quality, "
+               "reasoning dense data. The model belongs to the Phi-4 model family and "
+               "supports 128K token context length. The model underwent an enhancement "
+               "process, incorporating both supervised fine-tuning and direct preference "
+               "optimization to support precise instruction adherence and robust safety "
+               "measures."
+            ),
+            es=("Phi-4-mini-instruct es un modelo abierto ligero construido sobre datos "
+               "sintéticos y sitios web disponibles públicamente filtrados, con un enfoque"
+               "en datos de alta calidad y densos en razonamiento. El modelo pertenece a la "
+               "familia de modelos Phi-4 y soporta un contexto de longitud de 128K tokens. "
+               "El modelo se sometió a un proceso de mejora, incorporando tanto ajuste fino "
+               "(fine-tuning) supervisado como optimización directa de preferencias para "
+               "soportar una adherencia precisa a las instrucciones y medidas de seguridad."
+            ),
+        ),
+        alias=MultilingualString(en="Model Name", es="Nombre del Modelo"),
+    )  # type: ignore
+
+    quantization: schema_field(
+        enum_field(
+            enum=[
+                #"unsloth/Phi-4-mini-instruct-GGUF",
+                "Phi-4-mini-instruct-Q2_K.gguf",
+                "Phi-4-mini-instruct-Q2_K_L.gguf",
+                "Phi-4-mini-instruct-Q3_K_M.gguf",
+                "Phi-4-mini-instruct-Q4_K_M.gguf",
+                "Phi-4-mini-instruct-Q5_K_M.gguf",
+                "Phi-4-mini-instruct-Q6_K.gguf",
+                "Phi-4-mini-instruct.BF16.gguf",
+                "Phi-4-mini-instruct.Q8_0.gguf",
+            ]
+        ),
+        placeholder="Phi-4-mini-instruct.Q8_0.gguf",
+        description=MultilingualString(
+            en=(
+                "The specific Phi 4 Mini Instruct model quantization to use. Options "
+                "include various quantization sizes and the BF16 format. The choice of "
+                "quantization can affect the model's performance and resource usage, "
+                "with smaller quantizations typically requiring less memory but "
+                "potentially sacrificing some accuracy."
+            ),
+            es=(
+                "La cuantización específica del modelo Phi 4 Mini Instruct a utilizar. "
+                "Las opciones incluyen varios tamaños de cuantización y el formato BF16. "
+                "La elección de la cuantización puede afectar el rendimiento y el uso de "
+                "recursos del modelo, generalmente con cuantizaciones más pequeñas "
+                "requieren menos memoria pero potencialmente sacrifican algo de precisión."
+            ),
+        ),
+        alias=MultilingualString(en="Quantization", es="Cuantización"),
+    )  # type: ignore
+
+    max_tokens: schema_field(
+        int_field(ge=1),
+        placeholder=100,
+        description=MultilingualString(
+            en=(
+                "Maximum number of new tokens the model will generate per response. "
+                "Roughly 1 token ≈ 0.75 English words. Set to 100-200 for short "
+                "answers, 500-1000 for detailed explanations or code."
+            ),
+            es=(
+                "Número máximo de tokens nuevos que el modelo generará por respuesta. "
+                "Aproximadamente 1 token ≈ 0.75 palabras en español. Use 100-200 "
+                "para respuestas cortas, 500-1000 para explicaciones detalladas "
+                "o código."
+            ),
+        ),
+        alias=MultilingualString(en="Max tokens", es="Tokens máximos"),
+    )  # type: ignore
+
+    temperature: schema_field(
+        float_field(ge=0.0, le=1.0),
+        placeholder=0.7,
+        description=MultilingualString(
+            en=(
+                "Sampling temperature controlling output randomness (range 0.0-1.0). "
+                "At 0.0 the model picks the most likely token (deterministic). "
+                "Around 0.7 balances quality and creativity. At 1.0 outputs are "
+                "maximally varied."
+            ),
+            es=(
+                "Temperatura de muestreo que controla la aleatoriedad (rango 0.0-1.0). "
+                "En 0.0 el modelo elige el token más probable (determinista). "
+                "Alrededor de 0.7 equilibra calidad y creatividad. En 1.0 las salidas "
+                "son máximamente variadas."
+            ),
+        ),
+        alias=MultilingualString(en="Temperature", es="Temperatura"),
+    )  # type: ignore
+
+    frequency_penalty: schema_field(
+        float_field(ge=0.0, le=2.0),
+        placeholder=0.1,
+        description=MultilingualString(
+            en=(
+                "Penalizes tokens that have already appeared in the output based on "
+                "frequency (range 0.0-2.0). Higher values discourage repetition."
+            ),
+            es=(
+                "Penaliza los tokens que ya aparecieron en la salida según su "
+                "frecuencia (rango 0.0-2.0). Valores más altos desincentivan "
+                "la repetición."
+            ),
+        ),
+        alias=MultilingualString(
+            en="Frequency penalty", es="Penalización de frecuencia"
+        ),
+    )  # type: ignore
+
+    context_window: schema_field(
+        int_field(ge=1, le=131072),
+        placeholder=512,
+        description=MultilingualString(
+            en=(
+                "Total token budget for a single forward pass, including prompt and "
+                "response. Mistral-7B supports up to 32K tokens; Mistral-Nemo "
+                "supports up to 128K tokens."
+            ),
+            es=(
+                "Presupuesto total de tokens por pasada, incluyendo prompt y "
+                "respuesta. Mistral-7B soporta hasta 32K tokens; Mistral-Nemo "
+                "soporta hasta 128K tokens."
+            ),
+        ),
+        alias=MultilingualString(en="Context window", es="Ventana de contexto"),
+    )  # type: ignore
+
+    device: schema_field(
+        enum_field(enum=LLAMA_DEVICE_ENUM),
+        placeholder=LLAMA_DEVICE_PLACEHOLDER,
+        description=MultilingualString(
+            en=(
+                "Hardware device for llama.cpp inference. 'CPU' runs the model "
+                "fully in RAM. A GPU option offloads all layers for faster inference."
+            ),
+            es=(
+                "Dispositivo de hardware para inferencia con llama.cpp. 'CPU' ejecuta "
+                "el modelo en RAM. Una opción de GPU descarga todas las capas para "
+                "inferencia más rápida."
+            ),
+        ),
+        alias=MultilingualString(en="Device", es="Dispositivo"),
+    )  # type: ignore
+
+
+class Phi4MiniInstructModel(TextToTextGenerationTaskModel):
+    """Phi 4 Mini Instruct model for text generation using llama.cpp library."""
+
+    SCHEMA = Phi4MiniInstructSchema
+    COLOR: str = "#FF5733"
+    DISPLAY_NAME: str = MultilingualString(
+        en="Phi 4 Mini Instruct",
+        es="Phi 4 Mini Instruct",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en=(
+            "Phi-4-mini-instruct is a lightweight open model built upon synthetic "
+            "data and filtered publicly available websites - with a focus on high-quality, "
+            "reasoning dense data. The model belongs to the Phi-4 model family and "
+            "supports 128K token context length. The model underwent an enhancement "
+            "process, incorporating both supervised fine-tuning and direct preference "
+            "optimization to support precise instruction adherence and robust safety "
+            "measures."
+        ),
+        es=(
+            "Phi-4-mini-instruct es un modelo abierto ligero construido sobre datos "
+            "sintéticos y sitios web disponibles públicamente filtrados, con un enfoque"
+            "en datos de alta calidad y densos en razonamiento. El modelo pertenece a la "
+            "familia de modelos Phi-4 y soporta un contexto de longitud de 128K tokens. "
+            "El modelo se sometió a un proceso de mejora, incorporando tanto ajuste fino "
+            "(fine-tuning) supervisado como optimización directa de preferencias para "
+            "soportar una adherencia precisa a las instrucciones y medidas de seguridad."
+        ),
+    )
+
+    def __init__(self, **kwargs):
+        try:
+            from llama_cpp import Llama
+        except ImportError as e:
+            raise RuntimeError(
+                "llama-cpp-python is not installed. Please install it to use QwenModel."
+            ) from e
+
+        kwargs = self.validate_and_transform(kwargs)
+        self.model_name = kwargs.pop("model_name")
+        self.quantization = kwargs.pop("quantization", "Phi-4-mini-instruct.Q8_0.gguf")
+        self.max_tokens = kwargs.pop("max_tokens", 100)
+        self.temperature = kwargs.pop("temperature", 0.7)
+        self.frequency_penalty = kwargs.pop("frequency_penalty", 0.1)
+        self.n_ctx = kwargs.pop("context_window", 512)
+
+        use_gpu = LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) >= 0
+
+        self.model = Llama.from_pretrained(
+            repo_id=self.model_name,
+            filename=self.quantization,
+            verbose=True,
+            n_ctx=self.n_ctx,
+            n_gpu_layers=-1 if use_gpu else 0,
+            main_gpu=LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) if use_gpu else 0,
+        )
+
+    def generate(self, prompt: list[dict[str, str]]) -> List[str]:
+        output = self.model.create_chat_completion(
+            messages=prompt,
+            max_tokens=self.max_tokens,
+            temperature=self.temperature,
+            frequency_penalty=self.frequency_penalty,
+        )
+        return [output["choices"][0]["message"]["content"]]
diff --git a/DashAI/back/models/hugging_face/qwen_model.py b/DashAI/back/models/hugging_face/qwen_model.py
index 475c536bc..4224f238e 100644
--- a/DashAI/back/models/hugging_face/qwen_model.py
+++ b/DashAI/back/models/hugging_face/qwen_model.py
@@ -32,6 +32,10 @@ class QwenSchema(BaseSchema):
             enum=[
                 "Qwen/Qwen2.5-0.5B-Instruct-GGUF",
                 "Qwen/Qwen2.5-1.5B-Instruct-GGUF",
+                # "Qwen/Qwen3-4B-GGUF", This one is not working on llama-cpp 0.3.4
+                "Qwen/Qwen2.5-3B-Instruct-GGUF",
+                "Qwen/Qwen2.5-7B-Instruct-GGUF",
+                "Qwen/Qwen2.5-14B-Instruct-GGUF"
             ]
         ),
         placeholder="Qwen/Qwen2.5-1.5B-Instruct-GGUF",
@@ -431,17 +435,30 @@ def __init__(self, **kwargs):
         self.frequency_penalty = kwargs.pop("frequency_penalty", 0.1)
         self.n_ctx = kwargs.pop("context_window", 512)
 
-        self.filename = "*8_0.gguf"
+        self.filename = "*q8_0.gguf"
         use_gpu = LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) >= 0
 
-        self.model = Llama.from_pretrained(
-            repo_id=self.model_name,
-            filename=self.filename,
-            verbose=True,
-            n_ctx=self.n_ctx,
-            n_gpu_layers=-1 if use_gpu else 0,
-            main_gpu=LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) if use_gpu else 0,
-        )
+        try:
+            self.model = Llama.from_pretrained(
+                repo_id=self.model_name,
+                filename=self.filename,
+                verbose=True,
+                n_ctx=self.n_ctx,
+                n_gpu_layers=-1 if use_gpu else 0,
+                main_gpu=LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) if use_gpu else 0,
+            )
+        except ValueError:
+            # Fall back to Q2_K single-file quantization when Q8_0 is
+            # not available (e.g. large models whose Q8_0 is multi-part).
+            self.filename = "*q2_k.gguf"
+            self.model = Llama.from_pretrained(
+                repo_id=self.model_name,
+                filename=self.filename,
+                verbose=True,
+                n_ctx=self.n_ctx,
+                n_gpu_layers=-1 if use_gpu else 0,
+                main_gpu=LLAMA_DEVICE_TO_IDX.get(kwargs.get("device")) if use_gpu else 0,
+            )
 
     def generate(self, prompt: list[dict[str, str]]) -> List[str]:
         """Generate a reply for the given chat prompt.
diff --git a/DashAI/back/models/remote_models/__init__.py b/DashAI/back/models/remote_models/__init__.py
new file mode 100644
index 000000000..19f1f3f6f
--- /dev/null
+++ b/DashAI/back/models/remote_models/__init__.py
@@ -0,0 +1,4 @@
+from DashAI.back.models.remote_models.deepseek_text_to_text_generation_model import (
+    DeepSeekTextToTextGenerationModel,
+)
+from DashAI.back.models.remote_models.openai_text_to_text_generation_model import OpenAITextToTextGenerationModel
\ No newline at end of file
diff --git a/DashAI/back/models/remote_models/deepseek_text_to_text_generation_model.py b/DashAI/back/models/remote_models/deepseek_text_to_text_generation_model.py
new file mode 100644
index 000000000..c92eaa357
--- /dev/null
+++ b/DashAI/back/models/remote_models/deepseek_text_to_text_generation_model.py
@@ -0,0 +1,143 @@
+from typing import List
+
+from openai import OpenAI
+from pydantic import field_validator
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    float_field,
+    int_field,
+    none_type,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+
+deepseek_available_models = [
+    "deepseek-chat",
+    "deepseek-reasoner",
+]
+
+
+class DeepSeekTextToTextGenerationModelSchema(BaseSchema):
+    API_key: schema_field(
+        string_field(),
+        placeholder="",
+        description="API key for DeepSeek access.",
+    )  # type: ignore
+
+    model_name: schema_field(
+        enum_field(enum=deepseek_available_models),
+        placeholder="deepseek-chat",
+        description="The specific DeepSeek model version to use.",
+    )  # type: ignore
+
+    @field_validator(
+        "frequency_penalty",
+        "max_completions_tokens",
+        "presence_penalty",
+        "temperature",
+        "top_p",
+        mode="before",
+    )
+    def validate_optional_fields(cls, v):  # noqa: N805
+        if v == "":
+            return None
+        return v
+
+    frequency_penalty: schema_field(
+        none_type(float_field(ge=-2.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "Number between -2.0 and 2.0. Positive values penalize new tokens "
+            "based on their existing frequency in the text so far, decreasing the "
+            "model's likelihood to repeat the same line verbatim."
+        ),
+    )  # type: ignore
+
+    max_completions_tokens: schema_field(
+        none_type(int_field(ge=1)),
+        placeholder=None,
+        description=(
+            "An upper bound for the number of tokens that can be generated for a "
+            "completion, including visible output tokens and reasoning tokens."
+        ),
+    )  # type: ignore
+
+    presence_penalty: schema_field(
+        none_type(float_field(ge=-2.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "Number between -2.0 and 2.0. Positive values penalize new tokens "
+            "based on whether they appear in the text so far, increasing the "
+            "model's likelihood to talk about new topics."
+        ),
+    )  # type: ignore
+
+    temperature: schema_field(
+        none_type(float_field(ge=0.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "temperature: What sampling temperature to use, between 0 and 2. "
+            "Higher values like 0.8 will make the output more random, while lower "
+            "values like 0.2 will make it more focused and deterministic. "
+            "We generally recommend altering this or `top_p` but not both."
+        ),
+    )  # type: ignore
+
+    top_p: schema_field(
+        none_type(float_field(ge=0.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "An alternative to sampling with temperature, called nucleus sampling, "
+            "where the model considers the results of the tokens with top_p "
+            "probability mass. So 0.1 means only the tokens comprising the top 10% "
+            "probability mass are considered. "
+            "We generally recommend altering this or `temperature` but not both."
+        ),
+    )  # type: ignore
+
+
+class DeepSeekTextToTextGenerationModel(TextToTextGenerationTaskModel):
+    """Wrapper around DeepSeek's text-to-text generation models via API."""
+
+    SCHEMA = DeepSeekTextToTextGenerationModelSchema
+    DISPLAY_NAME: str = MultilingualString(
+        en="DeepSeek API",
+        es="API de DeepSeek",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en=(
+            "API for DeepSeek's language models, allowing you to select and configure the "
+            "text-to-text models supported by the DeepSeek API (https://deepseek.com/). "
+            "Note that it requires a private API key with an associated cost, that the "
+            "language model runs on DeepSeek's servers, and that your data will be used "
+            "according to that company's policies."
+        ),
+        es=(
+            "API para los modelos de lenguaje de DeepSeek, permite seleccionar y configurar los "
+            "modelos de texto a texto soportados por la API de DeepSeek (https://deepseek.com/). "
+            "Considere que requiere una API key (clave de API) privada con un costo asociado, "
+            "que el modelo de lenguaje se ejecuta en los servidores de DeepSeek y que sus datos "
+            "serán utilizados según las políticas de esa empresa."
+        ),
+    )
+
+    def __init__(self, **kwargs):
+        kwargs = self.validate_and_transform(kwargs)
+        self.client = OpenAI(
+            api_key=kwargs.get("API_key"),
+            base_url="https://api.deepseek.com",
+        )
+        self.model_name = kwargs.get("model_name")
+
+    def generate(self, prompt: list[dict[str, str]]) -> List[str]:
+        output = self.client.chat.completions.create(
+            model=self.model_name,
+            messages=prompt,
+        )
+        return [output.choices[0].message.content]
diff --git a/DashAI/back/models/remote_models/openai_text_to_text_generation_model.py b/DashAI/back/models/remote_models/openai_text_to_text_generation_model.py
new file mode 100644
index 000000000..ac6b9f544
--- /dev/null
+++ b/DashAI/back/models/remote_models/openai_text_to_text_generation_model.py
@@ -0,0 +1,176 @@
+from typing import List
+
+from openai import OpenAI
+from pydantic import field_validator
+
+from DashAI.back.core.schema_fields import (
+    BaseSchema,
+    enum_field,
+    float_field,
+    int_field,
+    none_type,
+    schema_field,
+    string_field,
+)
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.models.text_to_text_generation_model import (
+    TextToTextGenerationTaskModel,
+)
+
+# Hardcoded since the list of available models is only accessible via API call with
+# api key :c
+gpt_available_models = [
+    "gpt-4-0613",
+    "gpt-4",
+    "gpt-3.5-turbo",
+    "gpt-4-0314",
+    "gpt-3.5-turbo-instruct",
+    "gpt-3.5-turbo-instruct-0914",
+    "gpt-4-1106-preview",
+    "gpt-3.5-turbo-1106",
+    "gpt-4-0125-preview",
+    "gpt-4-turbo-preview",
+    "gpt-3.5-turbo-0125",
+    "gpt-4-turbo",
+    "gpt-4-turbo-2024-04-09",
+    "gpt-4o",
+    "gpt-4o-2024-05-13",
+    "gpt-4o-mini-2024-07-18",
+    "gpt-4o-mini",
+    "gpt-4o-2024-08-06",
+    "gpt-4o-2024-11-20",
+    "gpt-4o-mini-tts",
+    "gpt-4.1-2025-04-14",
+    "gpt-4.1",
+    "gpt-4.1-mini-2025-04-14",
+    "gpt-4.1-mini",
+    "gpt-4.1-nano-2025-04-14",
+    "gpt-4.1-nano",
+    "gpt-5-chat-latest",
+    "gpt-5-2025-08-07",
+    "gpt-5",
+    "gpt-5-mini-2025-08-07",
+    "gpt-5-mini",
+    "gpt-5-nano-2025-08-07",
+    "gpt-5-nano",
+    "gpt-5-pro-2025-10-06",
+    "gpt-5-pro",
+    "gpt-3.5-turbo-16k",
+]
+
+
+class OpenAITextToTextGenerationModelSchema(BaseSchema):
+    API_key: schema_field(
+        string_field(),
+        placeholder="",
+        description="API key for OpenAI access.",
+    )  # type: ignore
+
+    model_name: schema_field(
+        enum_field(enum=gpt_available_models),
+        placeholder="gpt-3.5-turbo",
+        description="The specific OpenAI model version to use.",
+    )  # type: ignore
+
+    @field_validator(
+        "frequency_penalty",
+        "max_completions_tokens",
+        "presence_penalty",
+        "temperature",
+        "top_p",
+        mode="before",
+    )
+    def validate_optional_fields(cls, v):  # noqa: N805
+        if v == "":
+            return None
+        return v
+
+    frequency_penalty: schema_field(
+        none_type(float_field(ge=-2.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "Number between -2.0 and 2.0. Positive values penalize new tokens "
+            "based on their existing frequency in the text so far, decreasing the "
+            "model's likelihood to repeat the same line verbatim."
+        ),
+    )  # type: ignore
+
+    max_completions_tokens: schema_field(
+        none_type(int_field(ge=1)),
+        placeholder=None,
+        description=(
+            "An upper bound for the number of tokens that can be generated for a "
+            "completion, including visible output tokens and reasoning tokens."
+        ),
+    )  # type: ignore
+
+    presence_penalty: schema_field(
+        none_type(float_field(ge=-2.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "Number between -2.0 and 2.0. Positive values penalize new tokens "
+            "based on whether they appear in the text so far, increasing the "
+            "model's likelihood to talk about new topics."
+        ),
+    )  # type: ignore
+
+    temperature: schema_field(
+        none_type(float_field(ge=0.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "temperature: What sampling temperature to use, between 0 and 2. "
+            "Higher values like 0.8 will make the output more random, while lower "
+            "values like 0.2 will make it more focused and deterministic. "
+            "We generally recommend altering this or `top_p` but not both."
+        ),
+    )  # type: ignore
+
+    top_p: schema_field(
+        none_type(float_field(ge=0.0, le=2.0)),
+        placeholder=None,
+        description=(
+            "An alternative to sampling with temperature, called nucleus sampling, "
+            "where the model considers the results of the tokens with top_p "
+            "probability mass. So 0.1 means only the tokens comprising the top 10% "
+            "probability mass are considered. "
+            "We generally recommend altering this or `temperature` but not both."
+        ),
+    )  # type: ignore
+
+
+class OpenAITextToTextGenerationModel(TextToTextGenerationTaskModel):
+    """Wrapper around OpenAI's text-to-text generation models."""
+
+    SCHEMA = OpenAITextToTextGenerationModelSchema
+    DISPLAY_NAME:str = MultilingualString(
+        en="OpenAI API",
+        es="API de OpenAI"
+    )
+    DESCRIPTION:str = MultilingualString(
+        en = (
+            "API for OpenAI's language models, allowing you to select and configure the "
+            "text-to-text models supported by the OpenAI API (https://openai.com/api/). "
+            "Note that it requires a private API key with an associated cost, that the "
+            "language model runs on OpenAI's servers, and that your data will be used "
+            "according to that company's policies."
+        ),
+        es = (
+            "API para los modelos de lenguaje de OpenAI, permite seleccionar y configurar los "
+            "modelos de texto a texto soportados por la API de OpenAI (https://openai.com/api/). "
+            "Considere que requiere una API key (clave de API) privada con un costo asociado, "
+            "que el modelo de lenguaje se ejecuta en los servidores de OpenAI y que sus datos "
+            "serán utilizados según las políticas de esa empresa."
+        )
+    )
+
+    def __init__(self, **kwargs):
+        kwargs = self.validate_and_transform(kwargs)
+        self.client = OpenAI(api_key=kwargs.get("API_key"))
+        self.model_name = kwargs.get("model_name")
+
+    def generate(self, prompt: list[dict[str, str]]) -> List[str]:
+        output = self.client.chat.completions.create(
+            model=self.model_name,
+            messages=prompt,
+        )
+        return [output.choices[0].message.content]
diff --git a/DashAI/back/models/text_to_text_generation_model.py b/DashAI/back/models/text_to_text_generation_model.py
index a546b9569..4ebf9b04e 100644
--- a/DashAI/back/models/text_to_text_generation_model.py
+++ b/DashAI/back/models/text_to_text_generation_model.py
@@ -8,4 +8,6 @@ class TextToTextGenerationTaskModel(BaseGenerativeModel):
     ``generate``. Compatible with ``TextToTextGenerationTask``.
     """
 
-    COMPATIBLE_COMPONENTS = ["TextToTextGenerationTask"]
+    COMPATIBLE_COMPONENTS = ["TextToTextGenerationTask", "RAGTask"]
+
+    REQUIRED_EXTRA_KWARGS = []
diff --git a/DashAI/back/tasks/RAG_task.py b/DashAI/back/tasks/RAG_task.py
new file mode 100644
index 000000000..bbb281f19
--- /dev/null
+++ b/DashAI/back/tasks/RAG_task.py
@@ -0,0 +1,164 @@
+import json
+from typing import Any, List, Optional, Tuple
+
+from DashAI.back.core.utils import MultilingualString
+from DashAI.back.dependencies.database.models import ProcessData
+from DashAI.back.models.RAG.RAG_pipeline import RAGGenerationOutput
+from DashAI.back.tasks.base_generative_task import BaseGenerativeTask
+
+
+class RAGTask(BaseGenerativeTask):
+    """Class for RAG Task.
+
+    Here you can change the methods provided by class Task.
+    """
+
+    metadata: dict = {
+        "inputs": {"str": {"min": 1, "max": 1}},
+        "outputs": {"str": {"min": 1, "max": 1}, "Dict": {"min": 1, "max": 1}},
+    }
+
+    DISPLAY_NAME: str = MultilingualString(
+        en="RAG Task",
+        es="Tarea RAG",
+    )
+    DESCRIPTION: str = MultilingualString(
+        en="""
+        This task generates a text response with an LLM model based on documents provided.
+        """,
+        es="""
+        Esta tarea genera una respuesta de texto con un modelo LLM basado en los documentos proporcionados.
+        """,
+    )
+
+    USE_HISTORY: bool = True
+
+    def prepare_for_task(
+        self,
+        input: List[ProcessData],
+        history: Optional[List[Tuple[str, str]]] = None,
+    ) -> list[dict[str, str]]:
+        """Prepare the input by including the history.
+
+        Parameters
+        ----------
+        input : str
+            The current input to be processed.
+            E.g.:
+            ["Tell me a joke."]
+
+        history : Optional[List[Tuple[str, str]]], optional
+            The history of previous inputs and outputs, by default None. E.g.:
+            [("Hello!", "Hello! How can I assist you today?")]
+
+        Returns
+        -------
+        str
+            The input prepared with the history to be used by the model.
+            E.g.:
+                [{"role": "user", "content": "Hello!"},
+                 {"role": "assistant", "content": "Hello! How can I assist you today?"},
+                 {"role": "user", "content": "Tell me a joke."}]
+        """
+        input = str(input[0].data)
+
+        prepared_input = [{"role": "user", "content": input}]
+
+        if not history:
+            return prepared_input
+
+        context = [
+            (
+                {"role": "user", "content": h_input},
+                {"role": "assistant", "content": h_output},
+            )
+            for (h_input, h_output) in history
+        ]
+        context = [entry for input_output in context for entry in input_output]
+        prepared_input = context + prepared_input
+        return prepared_input
+
+    def prepare_input_for_database(
+        self,
+        input: List[str],
+        **kwargs: Any,
+    ) -> List[Tuple[str, str]]:
+        """Prepare the input for the database.
+
+        Parameters
+        ----------
+        input : str
+            The input to be prepared.
+
+        Returns
+        -------
+        List[Tuple[str, str]]
+            Input with the new types as a list of tuples containing the data
+            and its type
+
+        """
+        return [(input[0], "str")]
+
+    def process_output(
+        self,
+        output: RAGGenerationOutput,
+        **kwargs: Any,
+    ) -> List[Tuple[str, str]]:
+        """Process the output of a generative model.
+
+        Parameters
+        ----------
+        output : RAGGenerationOutput
+            The typed output from RAGPipeline.generate().
+        """
+        message = output.message
+        chunks = output.chunks
+        return [
+            (str(message), "str"),
+            (
+                json.dumps(
+                    {k: v.to_dict() for k, v in chunks.items()},
+                    ensure_ascii=False,
+                ),
+                "Dict",
+            ),
+        ]
+
+    def process_output_from_database(
+        self,
+        output: List[ProcessData],
+        **kwargs: Any,
+    ) -> List[ProcessData]:
+        """Process the output from the database.
+
+        Parameters
+        ----------
+        output : list[str]
+            The output data to be processed.
+
+        Returns
+        -------
+        list[str]
+            The processed output data.
+        """
+
+        return output
+
+    def process_input_from_database(
+        self,
+        input: List[ProcessData],
+        **kwargs: Any,
+    ) -> List[ProcessData]:
+        """Process the input from the database.
+
+        Parameters
+        ----------
+        input : list[str]
+            The input data to be processed.
+
+        Returns
+        -------
+        list[str]
+            The processed input data.
+        """
+        return input
diff --git a/DashAI/back/tasks/__init__.py b/DashAI/back/tasks/__init__.py
index 9c0fa90a1..999f64b17 100644
--- a/DashAI/back/tasks/__init__.py
+++ b/DashAI/back/tasks/__init__.py
@@ -1 +1,12 @@
 # flake8: noqa
+from DashAI.back.tasks.base_generative_task import BaseGenerativeTask
+from DashAI.back.tasks.base_task import BaseTask
+from DashAI.back.tasks.classification_task import ClassificationTask
+from DashAI.back.tasks.controlnet_task import ControlNetTask
+from DashAI.back.tasks.RAG_task import RAGTask
+from DashAI.back.tasks.regression_task import RegressionTask
+from DashAI.back.tasks.tabular_classification_task import TabularClassificationTask
+from DashAI.back.tasks.text_classification_task import TextClassificationTask
+from DashAI.back.tasks.text_to_image_generation_task import TextToImageGenerationTask
+from DashAI.back.tasks.text_to_text_generation_task import TextToTextGenerationTask
+from DashAI.back.tasks.translation_task import TranslationTask
\ No newline at end of file
diff --git a/DashAI/front/package.json b/DashAI/front/package.json
index 915f700fa..04a097a00 100644
--- a/DashAI/front/package.json
+++ b/DashAI/front/package.json
@@ -11,6 +11,7 @@
     "@mui/material": "^7",
     "@mui/system": "^7",
     "@mui/utils": "^7",
+    "@mui/x-data-grid": "^8.27.2",
     "@mui/x-date-pickers": "^8.27.2",
     "@tanstack/react-table": "^8.21.3",
     "axios": "^1.3.4",
diff --git a/DashAI/front/src/App.jsx b/DashAI/front/src/App.jsx
index 903e514dc..2f6be13a8 100644
--- a/DashAI/front/src/App.jsx
+++ b/DashAI/front/src/App.jsx
@@ -1,6 +1,6 @@
 import React from "react";
 
-import { BrowserRouter, Outlet, Route, Routes } from "react-router-dom";
+import { BrowserRouter, Navigate, Outlet, Route, Routes } from "react-router-dom";
 import { TourRegistryProvider } from "./contexts/TourRegistryContext";
 import ModuleThemeWrapper from "./components/ModuleThemeWrapper";
 
@@ -13,10 +13,15 @@ import PluginsPage from "./pages/plugins/Plugins";
 import PipelinesPage from "./pages/pipelines/Pipelines";
 import PluginsDetails from "./pages/plugins/components/PluginsDetails";
 import Generative from "./pages/generative/Generative";
+import { GenerativeProvider } from "./components/generative/GenerativeContext";
 import NewPipelineWrapper from "./pages/pipelines/newPipelineWrapper";
 import HubContent from "./pages/hub/HubContent";
 import HubImportPage from "./pages/hub/HubImportPage";
 import JobQueueWidget from "./components/jobs/JobQueueWidget";
+import RAGDocumentsPage from "./pages/generative/RAG/RAGDocumentsPage";
+import RAGPromptsPage from "./pages/generative/RAG/RAGPromptsPage";
+import RAGSessionPage from "./pages/generative/RAGSession/RAGSessionPage";
+import SessionRouter from "./pages/generative/SessionRouter";
 import { DatasetsAndNotebooksProvider } from "./components/custom/contexts/DatasetsAndNotebooksContext";
 
 function DataSectionLayout() {
@@ -62,12 +67,32 @@ function App() {
             element={<ModelsPage />}
           />
           <Route path="/app/generative" element={<Generative />} />
+          <Route
+            path="/app/generative/RAG"
+            element={<Navigate to="/app/generative/sessions/new" replace />}
+          />
+          <Route
+            path="/app/generative/RAG/documents"
+            element={
+              <GenerativeProvider>
+                <RAGDocumentsPage />
+              </GenerativeProvider>
+            }
+          />
+          <Route
+            path="/app/generative/RAG/prompts"
+            element={
+              <GenerativeProvider>
+                <RAGPromptsPage />
+              </GenerativeProvider>
+            }
+          />
           <Route path="/app/generative/sessions/new" element={<Generative />} />
           <Route
             path="/app/generative/sessions/new/:modelName"
             element={<Generative />}
           />
-          <Route path="/app/generative/sessions/:id" element={<Generative />} />
+          <Route path="/app/generative/sessions/:id" element={<SessionRouter />} />
           <Route path="/app/pipelines" element={<PipelinesPage />} />
           <Route path="/app/pipelines/new" element={<NewPipelineWrapper />} />
           <Route
diff --git a/DashAI/front/src/api/component.ts b/DashAI/front/src/api/component.ts
index cc66da278..14dce2e67 100644
--- a/DashAI/front/src/api/component.ts
+++ b/DashAI/front/src/api/component.ts
@@ -1,6 +1,7 @@
 import api from "./api";
 import type { IComponent } from "../types/component";
 
+
 interface componentQuery {
   model?: string;
   selectTypes?: string[];
@@ -50,6 +51,12 @@ export const getComponents = async ({
   return response.data;
 };
 
+export const getChildComponents = async (componentName: string, recursive: boolean): Promise<IComponent[]> => {
+  const response = await api.get<IComponent[]>(`/v1/component/${componentName}/children`, {
+    params: { recursive },
+  });
+  return response.data;
+};
 export const getComponentById = async (id: string): Promise<IComponent> => {
   const response = await api.get<IComponent>(`/v1/component/${id}/`);
   return response.data;
diff --git a/DashAI/front/src/api/generativeTask.ts b/DashAI/front/src/api/generativeTask.ts
index d64c2c18f..2dac403c7 100644
--- a/DashAI/front/src/api/generativeTask.ts
+++ b/DashAI/front/src/api/generativeTask.ts
@@ -12,15 +12,22 @@ export const getGenerativeTask = async (): Promise<IGenerativeTask[]> => {
 export const getRelatedComponents = async (
   relatedComponent: string,
 ): Promise<IGenerativeTask[]> => {
-  const response = await api.get<IGenerativeTask[]>(
-    `/v1/component/?related_component=${encodeURIComponent(relatedComponent)}`,
-  );
-  return response.data;
+  try{
+    const response = await api.get<IGenerativeTask[]>(
+      `/v1/component/?related_component=${encodeURIComponent(relatedComponent)}`,
+    );
+    console.log("Related components:", response.data);
+    return response.data;
+  } catch (error) {
+    return [];
+  }
 };
 
 export const createGenerativeSession = async (
   sessionData: Omit<ISession, "id" | "created" | "last_modified">,
 ): Promise<ISession> => {
+
+  console.log("Creating new generative session with data:", sessionData);
   const response = await api.post<ISession>(
     "/v1/generative-session/",
     sessionData,
diff --git a/DashAI/front/src/api/job.ts b/DashAI/front/src/api/job.ts
index 684fd434d..a75e47a6e 100644
--- a/DashAI/front/src/api/job.ts
+++ b/DashAI/front/src/api/job.ts
@@ -176,6 +176,24 @@ export const enqueueGenerativeProcessJob = async (
   return response.data;
 };
 
+export const enqueueRAGProcessJob = async (
+  processId: number,
+): Promise<object> => {
+  const data = {
+    job_type: "RAGJob",
+    kwargs: { rag_process_id: processId },
+  };
+  const formData = new FormData();
+  formData.append("job_type", data.job_type);
+  formData.append("kwargs", JSON.stringify(data.kwargs));
+  const response = await api.post<object>("/v1/job/", formData, {
+    headers: {
+      "Content-Type": "multipart/form-data",
+    },
+  });
+  return response.data;
+}
+
 export const enqueueConverterJob = async (
   converterId: number,
 ): Promise<object> => {
diff --git a/DashAI/front/src/api/rag.ts b/DashAI/front/src/api/rag.ts
new file mode 100644
index 000000000..45b945c42
--- /dev/null
+++ b/DashAI/front/src/api/rag.ts
@@ -0,0 +1,295 @@
+import api from "./api";
+import { ISession } from "../types/session";
+import { IGenerativeTask } from "../types/generativeTask";
+import { IDocumentResponse } from "../types/documentResponse";
+import { IComponent } from "../types/component";
+import { IRAGPrompt } from "../types/ragPrompt";
+import { getChildComponents } from "./component";
+
+export const createRAGPrompt = async (prompt: {
+  class_name: string;
+  name: string;
+  parameters?: Record<string, any>;
+}): Promise<{ id: number }> => {
+  const response = await api.post("/v1/prompt/", prompt);
+  if (response.status !== 201) {
+    throw new Error(`Failed to create RAG prompt: ${response.statusText}`);
+  }
+  return response.data;
+};
+
+export const getRAGSessions = async (): Promise<ISession[]> => {
+  const response = await api.get<ISession[]>("/v1/generative-session/");
+  if (response.status !== 200) {
+    throw new Error(`Failed to fetch RAG sessions: ${response.statusText}`);
+  }
+
+  const ragSessions = response.data.filter(
+    (session) => session.task_name === "RAGTask",
+  );
+  return ragSessions;
+};
+
+export const getRAGSession = async (sessionId: number): Promise<ISession> => {
+  const response = await api.get<ISession>(
+    `/v1/generative-session/${sessionId}`,
+  );
+  if (response.status !== 200) {
+    throw new Error(`Failed to fetch RAG session: ${response.statusText}`);
+  }
+
+  return response.data;
+};
+
+export const createRAGSession = async (
+  sessionData: Omit<ISession, "id" | "created" | "last_modified">,
+): Promise<ISession> => {
+  console.log("Creating RAG session with data:", sessionData);
+  const params = sessionData.parameters as {
+    documents: number[];
+    chunking_model: {
+      component: string;
+      params: Record<string, any>;
+    };
+    retriever_model: {
+      component: string;
+      params: Record<string, any>;
+    };
+    generation_model: {
+      component: string;
+      params: Record<string, any>;
+    };
+    prompt: {
+      component: string;
+      params: Record<string, any>;
+    };
+  };
+
+  const transformedSession: Omit<ISession, "id" | "created" | "last_modified"> =
+    {
+      name: sessionData.name,
+      description: sessionData.description,
+      task_name: "RAGTask",
+      model_name: "RAGPipeline",
+      display_name: "",
+      parameters: {
+        documents: params.documents,
+        chunking_model: {
+          component: params.chunking_model.component,
+          params: params.chunking_model.params,
+        },
+        retriever_model: {
+          component: params.retriever_model.component,
+          params: params.retriever_model.params,
+        },
+        generation_model: {
+          component: params.generation_model.component,
+          params: params.generation_model.params,
+        },
+        prompt: params.prompt,
+      },
+    };
+
+  const response = await api.post<ISession>(
+    "/v1/generative-session/",
+    transformedSession,
+  );
+
+  if (response.status !== 201) {
+    throw new Error(`Failed to create RAG session: ${response.statusText}`);
+  }
+
+  return response.data;
+};
+
+export const updateRAGSession = async (
+  sessionId: number,
+  sessionData: Partial<ISession>,
+): Promise<ISession> => {
+  const response = await api.put<ISession>(
+    `/v1/generative-session/${sessionId}`,
+    sessionData,
+  );
+  if (response.status !== 200) {
+    throw new Error(`Failed to update RAG session: ${response.statusText}`);
+  }
+
+  return response.data;
+};
+
+export const deleteRAGSession = async (sessionId: number): Promise<void> => {
+  const response = await api.delete(`/v1/generative-session/${sessionId}`);
+  if (response.status !== 204) {
+    throw new Error(`Failed to delete RAG session: ${response.statusText}`);
+  }
+};
+
+export const updateGenerativeSessionParams = async (
+  sessionId: number,
+  newParams: Record<string, any>,
+): Promise<ISession> => {
+  const response = await api.put<ISession>(
+    `/v1/generative-session/${sessionId}/parameters`,
+    newParams,
+  );
+  if (response.status !== 200) {
+    throw new Error(
+      `Failed to update RAG session parameters: ${response.statusText}`,
+    );
+  }
+  return response.data;
+};
+
+export const getRetrievalParadigm = async (): Promise<IComponent[]> => {
+  const response = await getChildComponents("RetrieverModel", true);
+  if (!response) {
+    throw new Error(`Failed to fetch retrieval options`);
+  }
+  return response;
+};
+
+export const getRetrieverComponents = async (
+  retrievalParadigm: string,
+): Promise<IComponent[]> => {
+  const response = await getChildComponents(retrievalParadigm, true);
+
+  if (!response) {
+    throw new Error(`Failed to fetch retriever components`);
+  }
+
+  return response;
+};
+
+export const getGeneratorComponents = async (): Promise<IGenerativeTask[]> => {
+  const response = await api.get(
+    `/v1/component/?related_component=TextToTextGenerationTask`,
+  );
+
+  if (response.status !== 200) {
+    throw new Error(
+      `Failed to fetch generator components: ${response.statusText}`,
+    );
+  }
+
+  return response.data;
+};
+
+export const getChunkingComponents = async (): Promise<IComponent[]> => {
+  const response = await getChildComponents("BaseChunkingModel", false);
+  if (!response) {
+    throw new Error(`Failed to fetch chunking components`);
+  }
+  return response;
+};
+
+export const loadDocuments = async (): Promise<IDocumentResponse[]> => {
+  const response = await api.get<IDocumentResponse[]>("/v1/document/");
+  if (response.status !== 200) {
+    throw new Error(`Failed to load documents: ${response.statusText}`);
+  }
+  return response.data;
+};
+
+export const getSessionDocuments = async (
+  sessionId: number,
+): Promise<IDocumentResponse[]> => {
+  const response = await api.get<IDocumentResponse[]>(
+    `/v1/document/session/${sessionId}`,
+  );
+  if (response.status !== 200) {
+    throw new Error(`Failed to load session documents: ${response.statusText}`);
+  }
+  return response.data;
+};
+
+export const deleteDocument = async (documentId: number): Promise<void> => {
+  const response = await api.delete(`/v1/document/${documentId}`);
+  if (response.status !== 204) {
+    throw new Error(`Failed to delete document: ${response.statusText}`);
+  }
+};
+
+export const addDocument = async ({
+  file,
+  optional_metadata,
+}: {
+  file: File;
+  optional_metadata?: Record<string, any>;
+}): Promise<IDocumentResponse> => {
+  if (optional_metadata) {
+    optional_metadata.last_modified = file.lastModified;
+  }
+  const metadata = {
+    file_name: file.name,
+    last_modified: file.lastModified,
+    optional_metadata,
+  };
+
+  const formData = new FormData();
+  formData.append("file", file);
+  formData.append("metadata", JSON.stringify(metadata));
+
+  const response = await api.post<IDocumentResponse>(
+    "/v1/document/",
+    formData,
+    {
+      headers: { "Content-Type": "multipart/form-data" },
+    },
+  );
+
+  if (response.status !== 201) {
+    throw new Error(`Failed to upload document: ${response.statusText}`);
+  }
+
+  return response.data;
+};
+
+/** Class name prefix that identifies non-generation prompt types. */
+const AUGMENTATION_PROMPT_CLASS_PREFIX = "Augmentation";
+
+/**
+ * Check whether a prompt's class_name corresponds to a generation prompt
+ * (i.e. NOT an augmentation prompt).
+ */
+export function isGenerationPromptClass(className: string): boolean {
+  return !className.includes(AUGMENTATION_PROMPT_CLASS_PREFIX);
+}
+
+export const getDefaultPrompts = async (): Promise<IComponent[]> => {
+  return getChildComponents("RAGGenerationPrompt", false);
+};
+
+export const getRAGPrompts = async (): Promise<IRAGPrompt[]> => {
+  const response = await api.get<IRAGPrompt[]>("/v1/prompt/");
+  if (response.status !== 200) {
+    throw new Error(`Failed to fetch RAG prompts: ${response.statusText}`);
+  }
+  return response.data;
+};
+
+export const getCustomPrompts = async (
+  types: string[] = ["RAGGenerationPrompt", "AugmentationPrompt"],
+): Promise<IComponent[]> => {
+  let allChildren: IComponent[] = [];
+  for (const type of types) {
+    const response = await api.get<IComponent[]>(
+      `/v1/component/${type}/children`,
+      { params: { recursive: false } },
+    );
+    if (response.status !== 200) {
+      throw new Error(
+        `Failed to fetch ${type} children: ${response.statusText}`,
+      );
+    }
+    const filtered = response.data.filter(
+      (child) =>
+        !(
+          child.name &&
+          typeof child.name === "string" &&
+          child.name.includes("Default")
+        ),
+    );
+    allChildren.push(...filtered);
+  }
+  return allChildren;
+};
diff --git a/DashAI/front/src/api/session.ts b/DashAI/front/src/api/session.ts
index a08e4d534..2d727a819 100644
--- a/DashAI/front/src/api/session.ts
+++ b/DashAI/front/src/api/session.ts
@@ -31,7 +31,7 @@ export const updateGenerativeSession = async ({
   formData,
 }: {
   id: string;
-  formData: { name?: string; task_name?: string };
+  formData: { name?: string; task_name?: string; description?: string };
 }): Promise<object> => {
   const response = await api.patch(`/v1/generative-session/${id}`, null, {
     params: formData,
diff --git a/DashAI/front/src/components/configurableObject/Inputs/ArrayInput.jsx b/DashAI/front/src/components/configurableObject/Inputs/ArrayInput.jsx
index 46be6e56e..10b2ecba5 100644
--- a/DashAI/front/src/components/configurableObject/Inputs/ArrayInput.jsx
+++ b/DashAI/front/src/components/configurableObject/Inputs/ArrayInput.jsx
@@ -14,7 +14,15 @@ function ArrayInput({
   itemType,
   ...props
 }) {
-  const [inputValue, setInputValue] = useState(value.join(","));
+  // Ensure value is an array before using join
+  const safeValue = Array.isArray(value) ? value : [];
+  const [inputValue, setInputValue] = useState(safeValue.join(","));
+
+  // Update inputValue when value prop changes
+  useEffect(() => {
+    const newSafeValue = Array.isArray(value) ? value : [];
+    setInputValue(newSafeValue.join(","));
+  }, [value]);
 
   useEffect(() => {
     if (Array.isArray(value)) {
diff --git a/DashAI/front/src/components/configurableObject/Inputs/ClassInput.jsx b/DashAI/front/src/components/configurableObject/Inputs/ClassInput.jsx
index 5750881f8..d2c6a43f5 100644
--- a/DashAI/front/src/components/configurableObject/Inputs/ClassInput.jsx
+++ b/DashAI/front/src/components/configurableObject/Inputs/ClassInput.jsx
@@ -164,6 +164,18 @@ function ClassInput({
         open={open}
         onClose={handleClose}
         sx={{ display: modal ? "show" : "none" }}
+        maxWidth="sm"
+        fullWidth
+        PaperProps={{
+          sx: {
+            zIndex: 1400, // Ensure it appears above parent Dialog (which has z-index 1300)
+          }
+        }}
+        BackdropProps={{
+          sx: {
+            zIndex: 1399, // Backdrop should be below the Dialog but above parent
+          }
+        }}
       >
         <DialogTitle>{`${selectedOption} parameters`}</DialogTitle>
         <DialogContent key={selectedOption}>
diff --git a/DashAI/front/src/components/configurableObject/Inputs/TextInput.jsx b/DashAI/front/src/components/configurableObject/Inputs/TextInput.jsx
index fbabeb814..49447ba95 100644
--- a/DashAI/front/src/components/configurableObject/Inputs/TextInput.jsx
+++ b/DashAI/front/src/components/configurableObject/Inputs/TextInput.jsx
@@ -20,6 +20,9 @@ function TextInput({
   description,
   ...props
 }) {
+  const isEmpty = value === undefined || value === "";
+  const showError = error || (isEmpty ? `${name} is a required field` : "");
+
   return (
     <FormInputWrapper name={name} description={description}>
       <InputWithDebounce
@@ -30,8 +33,9 @@ function TextInput({
         value={value === null ? "none" : value}
         onChange={onChange}
         autoComplete="off"
-        error={error !== undefined}
-        helperText={error}
+        error={!!showError}
+        helperText={showError || " "}
+        margin="dense"
       />
     </FormInputWrapper>
   );
diff --git a/DashAI/front/src/components/custom/TemplateModal.jsx b/DashAI/front/src/components/custom/TemplateModal.jsx
new file mode 100644
index 000000000..789127a51
--- /dev/null
+++ b/DashAI/front/src/components/custom/TemplateModal.jsx
@@ -0,0 +1,51 @@
+import React from "react";
+import PropTypes from "prop-types";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogContentText,
+  DialogActions,
+  Button,
+  Typography,
+} from "@mui/material";
+
+export default function TemplateModal({ 
+  open, 
+  handleClose, 
+  template,
+  title = "Prompt",
+  formatText = true,
+}) {
+  return (
+    <Dialog open={open} onClose={handleClose} maxWidth="md" fullWidth>
+      <DialogTitle>{title}</DialogTitle>
+      <DialogContent>
+        <DialogContentText component="div">
+          <Typography
+            variant="body1"
+            sx={{
+              whiteSpace: formatText ? "pre-wrap" : "normal",
+              wordBreak: "break-word",
+            }}
+          >
+            {template}
+          </Typography>
+        </DialogContentText>
+      </DialogContent>
+      <DialogActions>
+        <Button onClick={handleClose} autoFocus>
+          Close
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
+TemplateModal.propTypes = {
+  open: PropTypes.bool.isRequired,
+  handleClose: PropTypes.func.isRequired,
+  template: PropTypes.string,
+  title: PropTypes.string,
+  formatText: PropTypes.bool,
+};
diff --git a/DashAI/front/src/components/datasets/ConfigureAndUploadDataset.jsx b/DashAI/front/src/components/datasets/ConfigureAndUploadDataset.jsx
index 9a4a27d3e..b6c3a3247 100644
--- a/DashAI/front/src/components/datasets/ConfigureAndUploadDataset.jsx
+++ b/DashAI/front/src/components/datasets/ConfigureAndUploadDataset.jsx
@@ -1,7 +1,7 @@
 import React, { useState, useEffect } from "react";
-import { Grid, Paper } from "@mui/material";
+import { Grid, Paper, Typography } from "@mui/material";
 import PropTypes from "prop-types";
-import Upload from "./Upload";
+import Upload from "../shared/Upload";
 import { getComponents as getComponentsRequest } from "../../api/component";
 import { useSnackbar } from "notistack";
 import DataloaderConfiguration from "./DataloaderConfiguration";
diff --git a/DashAI/front/src/components/documents/DocumentTable.jsx b/DashAI/front/src/components/documents/DocumentTable.jsx
new file mode 100644
index 000000000..1ecd4471e
--- /dev/null
+++ b/DashAI/front/src/components/documents/DocumentTable.jsx
@@ -0,0 +1,205 @@
+import React, { useEffect, useState, useCallback } from "react";
+import { DataGrid, GridToolbar } from "@mui/x-data-grid";
+import { useSnackbar } from "notistack";
+import { Button, Grid, Paper, Typography, LinearProgress, Box } from "@mui/material";
+import {
+  get_documents_json,
+  delete_document as deleteDocumentRequest,
+} from "../../api/documents";
+import {
+  AddCircleOutline as AddIcon,
+  Update as UpdateIcon,
+} from "@mui/icons-material";
+import DeleteItemModal from "../custom/DeleteItemModal";
+import DocumentSummaryModal from "./DocumentSummaryModal";
+
+function DocumentTable({
+  handleNewDocument,
+  updateTableFlag,
+  setUpdateTableFlag,
+}) {
+  const { enqueueSnackbar } = useSnackbar();
+  const [loading, setLoading] = useState(true);
+  const [documents, setDocuments] = useState([]);
+  const [selectionModel, setSelectionModel] = useState([]);
+
+  const getDocuments = useCallback(async () => {
+    setLoading(true);
+    try {
+      const docs = await get_documents_json();
+      setDocuments(docs);
+    } catch (error) {
+      enqueueSnackbar("Error when trying to get the documents", {
+        variant: "error",
+      });
+      console.error("Error fetching documents:", error);
+    } finally {
+      setLoading(false);
+    }
+  }, [enqueueSnackbar]);
+
+  const deleteDocument = async (doc_name) => {
+    try {
+      await deleteDocumentRequest(doc_name);
+      setUpdateTableFlag(true);
+      enqueueSnackbar("Document successfully deleted.", {
+        variant: "success",
+      });
+    } catch (error) {
+      enqueueSnackbar("Error when trying to delete the document", {
+        variant: "error",
+      });
+      console.error("Error deleting document:", error);
+    }
+  };
+
+  const createDeleteHandler = useCallback(
+    (doc_name) => () => {
+      deleteDocument(doc_name);
+    },
+    [deleteDocument],
+  );
+
+  useEffect(() => {
+    getDocuments();
+  }, [getDocuments]);
+
+  useEffect(() => {
+    if (updateTableFlag) {
+      getDocuments();
+      setUpdateTableFlag(false);
+    }
+  }, [updateTableFlag, setUpdateTableFlag, getDocuments]);
+
+  const columns = React.useMemo(
+    () => [
+      {
+        field: "id",
+        headerName: "ID",
+        minWidth: 30,
+        editable: false,
+      },
+      {
+        field: "doc_name",
+        headerName: "Document Name",
+        minWidth: 200,
+        editable: false,
+      },
+      {
+        field: "created_at",
+        headerName: "Created At",
+        minWidth: 200,
+        editable: false,
+      },
+      {
+        field: "actions",
+        type: "actions",
+        minWidth: 150,
+        getActions: (params) => [
+          <DeleteItemModal
+            key="delete-component"
+            deleteFromTable={createDeleteHandler(params.row.doc_name)}
+          />,
+          <DocumentSummaryModal
+            key="summary-component"
+            docName={params.row.doc_name}
+          />,
+        ],
+      },
+    ],
+    [createDeleteHandler],
+  );
+
+  return (
+    <Paper sx={{ py: 4, px: 6, height: "100%", minHeight: 400, display: "flex", flexDirection: "column" }}>
+      <Grid
+        container
+        direction="row"
+        justifyContent="space-between"
+        alignItems="center"
+        sx={{ mb: 4 }}
+      >
+        <Typography variant="h5" component="h2">
+          Current documents
+        </Typography>
+        <Grid item>
+          <Grid container spacing={2}>
+            <Grid item>
+              <Button
+                variant="contained"
+                onClick={handleNewDocument}
+                endIcon={<AddIcon />}
+              >
+                New Document
+              </Button>
+            </Grid>
+            <Grid item>
+              <Button
+                variant="contained"
+                onClick={() => setUpdateTableFlag(true)}
+                endIcon={<UpdateIcon />}
+                disabled={loading}
+              >
+                Update
+              </Button>
+            </Grid>
+          </Grid>
+        </Grid>
+      </Grid>
+
+      <Box sx={{ flex: 1, minHeight: 300, display: "flex", flexDirection: "column" }}>
+        {(!loading && documents.length === 0) ? (
+          <Box
+            sx={{
+              flex: 1,
+              minHeight: 300,
+              display: "flex",
+              alignItems: "center",
+              justifyContent: "center",
+              width: "100%",
+            }}
+          >
+            <Typography variant="h6" color="text.secondary" align="center" sx={{ width: "100%" }}>
+              No documents available
+            </Typography>
+          </Box>
+        ) : (
+          <DataGrid
+            rows={documents}
+            columns={columns}
+            initialState={{
+              pagination: {
+                paginationModel: {
+                  pageSize: 5,
+                },
+              },
+            }}
+            getRowId={(row) => String(row.id)}
+            pageSize={5}
+            sortModel={[{ field: "id", sort: "asc" }]}
+            pageSizeOptions={[5, 10]}
+            checkboxSelection
+            selectionModel={selectionModel}
+            onSelectionModelChange={setSelectionModel}
+            disableRowSelectionOnClick={false}
+            autoHeight={false}
+            loading={loading}
+            slots={{
+              toolbar: GridToolbar,
+              loadingOverlay: LinearProgress,
+            }}
+            sx={{
+              flex: 1,
+              minHeight: 300,
+              "& .MuiDataGrid-cell:focus": {
+                outline: "none",
+              },
+            }}
+          />
+        )}
+      </Box>
+    </Paper>
+  );
+}
+
+export default DocumentTable;
\ No newline at end of file
diff --git a/DashAI/front/src/components/generative/CreateSessionCenter.jsx b/DashAI/front/src/components/generative/CreateSessionCenter.jsx
index 2480df2b4..49701a6ca 100644
--- a/DashAI/front/src/components/generative/CreateSessionCenter.jsx
+++ b/DashAI/front/src/components/generative/CreateSessionCenter.jsx
@@ -7,11 +7,14 @@ import {
 } from "@mui/material";
 import { useCallback, useEffect } from "react";
 import { useTranslation } from "react-i18next";
+import { useNavigate } from "react-router-dom";
 import ComponentSelector from "../custom/ComponentSelector";
 import GenerativeBreadcrumbs from "./GenerativeBreadcrumbs";
 import { useCreateSession } from "./CreateSessionContext";
+import { useGenerative } from "./GenerativeContext";
 import StepperNavigationFooter from "../shared/StepperNavigationFooter";
 import { useTourContext } from "../tour/TourProvider";
+import RAGSessionSetup from "../../pages/generative/RAGSession/RAGSessionSetup";
 
 export default function CreateSessionCenter() {
   const { t } = useTranslation(["generative", "common"]);
@@ -38,6 +41,9 @@ export default function CreateSessionCenter() {
     handleCreate,
   } = useCreateSession();
 
+  const navigate = useNavigate();
+  const { sessions, setSessions } = useGenerative();
+
   const handleSelectModelWithTour = useCallback(
     (model) => {
       handleSelectModel(model);
@@ -56,6 +62,18 @@ export default function CreateSessionCenter() {
     handleCreate();
   }, [handleCreate, tourContext]);
 
+  const handleRagSessionCreated = useCallback(
+    (createdSession) => {
+      setSessions((prev) => [...prev, createdSession]);
+      navigate(`/app/generative/sessions/${createdSession.id}`);
+    },
+    [setSessions, navigate],
+  );
+
+  const handleRagClose = useCallback(() => {
+    handleBack();
+  }, [handleBack]);
+
   useEffect(() => {
     if (!tourContext?.run) return;
     const currentTarget = tourContext.steps?.[tourContext.stepIndex]?.target;
@@ -84,18 +102,20 @@ export default function CreateSessionCenter() {
       }}
     >
       <GenerativeBreadcrumbs />
-      <Box sx={{ mb: 4 }}>
-        <Typography variant="h5" component="h2">
-          {step === 0
-            ? t("generative:label.selectModel")
-            : t("generative:label.configureSession")}
-        </Typography>
-        <Typography variant="body2" color="text.secondary">
-          {step === 0
-            ? t("generative:label.pickAModelGroupedByTask")
-            : t("generative:label.nameAndDescribeYourSession")}
-        </Typography>
-      </Box>
+      {step === 1 && selectedModel?.task_name === "RAGTask" ? null : (
+        <Box sx={{ mb: 4 }}>
+          <Typography variant="h5" component="h2">
+            {step === 0
+              ? t("generative:label.selectModel")
+              : t("generative:label.configureSession")}
+          </Typography>
+          <Typography variant="body2" color="text.secondary">
+            {step === 0
+              ? t("generative:label.pickAModelGroupedByTask")
+              : t("generative:label.nameAndDescribeYourSession")}
+          </Typography>
+        </Box>
+      )}
 
       <Box
         data-tour={step === 1 ? "session-config" : undefined}
@@ -132,6 +152,12 @@ export default function CreateSessionCenter() {
               tourDataMatchFn={(c) => c.name.toLowerCase().includes("qwen")}
             />
           )
+        ) : selectedModel?.task_name === "RAGTask" ? (
+          <RAGSessionSetup
+            onClose={handleRagClose}
+            onSessionCreated={handleRagSessionCreated}
+            existingSessions={sessions}
+          />
         ) : (
           <Stack spacing={4} sx={{ maxWidth: "100%" }}>
             <TextField
@@ -158,24 +184,26 @@ export default function CreateSessionCenter() {
         )}
       </Box>
 
-      <StepperNavigationFooter
-        onBack={handleBack}
-        onNext={step === 0 ? handleNextWithTour : handleCreateWithTour}
-        backDisabled={submitting}
-        nextDisabled={step === 0 ? !canGoNext : !canCreate}
-        nextLabel={
-          step === 0 ? t("common:next") : t("generative:button.createSession")
-        }
-        loading={submitting}
-        variant={step === 0 ? "next" : "save"}
-        nextDataTour={
-          tourContext?.run
-            ? step === 0
-              ? "create-session-next"
-              : "create-session-button"
-            : null
-        }
-      />
+      {step === 1 && selectedModel?.task_name === "RAGTask" ? null : (
+        <StepperNavigationFooter
+          onBack={handleBack}
+          onNext={step === 0 ? handleNextWithTour : handleCreateWithTour}
+          backDisabled={submitting}
+          nextDisabled={step === 0 ? !canGoNext : !canCreate}
+          nextLabel={
+            step === 0 ? t("common:next") : t("generative:button.createSession")
+          }
+          loading={submitting}
+          variant={step === 0 ? "next" : "save"}
+          nextDataTour={
+            tourContext?.run
+              ? step === 0
+                ? "create-session-next"
+                : "create-session-button"
+              : null
+          }
+        />
+      )}
     </Box>
   );
 }
diff --git a/DashAI/front/src/components/generative/CreateSessionContext.jsx b/DashAI/front/src/components/generative/CreateSessionContext.jsx
index 43a639088..b895bc2fc 100644
--- a/DashAI/front/src/components/generative/CreateSessionContext.jsx
+++ b/DashAI/front/src/components/generative/CreateSessionContext.jsx
@@ -50,7 +50,8 @@ export function CreateSessionProvider({ children }) {
     setLoadingModels(true);
 
     Promise.all(
-      tasks.map((task) =>
+      tasks
+        .map((task) =>
         getRelatedComponents(task.name).then((components) =>
           components.map((c) => ({
             ...c,
diff --git a/DashAI/front/src/components/generative/CreateSessionLanding.jsx b/DashAI/front/src/components/generative/CreateSessionLanding.jsx
index 015d418e6..b47c51d25 100644
--- a/DashAI/front/src/components/generative/CreateSessionLanding.jsx
+++ b/DashAI/front/src/components/generative/CreateSessionLanding.jsx
@@ -9,7 +9,7 @@ export default function CreateSessionLanding() {
   const { t } = useTranslation(["generative", "common"]);
   const tourContext = useTourContext();
 
-  const handleCreateSession = () => {
+  const handleOption = () => {
     if (tourContext?.run) {
       tourContext.nextStep();
     }
@@ -18,7 +18,7 @@ export default function CreateSessionLanding() {
 
   return (
     <SelectOptionMenu
-      goToNextStep={handleCreateSession}
+      goToNextStep={handleOption}
       title={t("generative:label.generativeModule")}
       subtitle={t("generative:label.createNewSessionDescription")}
       options={[
diff --git a/DashAI/front/src/components/generative/CreateSessionRight.jsx b/DashAI/front/src/components/generative/CreateSessionRight.jsx
index 82c86d12e..4b4c34ff0 100644
--- a/DashAI/front/src/components/generative/CreateSessionRight.jsx
+++ b/DashAI/front/src/components/generative/CreateSessionRight.jsx
@@ -1,10 +1,11 @@
-import { Box, Divider, Typography } from "@mui/material";
+import { Box, Typography } from "@mui/material";
 import { useTranslation } from "react-i18next";
 import SideBar from "../threeSectionLayout/panelContainers/SideBar";
 import ComponentDetailsPanel from "../custom/ComponentDetailsPanel";
 import FormSchemaRenderFields from "../shared/FormSchemaRenderFields";
 import { useTheme } from "@mui/material/styles";
 import { useCreateSession } from "./CreateSessionContext";
+import RAGInfoBar from "./RAG/RAGInfoBar";
 
 export default function CreateSessionRight() {
   const { t } = useTranslation(["generative", "common"]);
@@ -26,6 +27,14 @@ export default function CreateSessionRight() {
     );
   }
 
+  if (selectedModel?.task_name === "RAGTask") {
+    return (
+      <Box sx={{ height: "100%", overflow: "auto", p: 2 }}>
+        <RAGInfoBar />
+      </Box>
+    );
+  }
+
   return (
     <SideBar data-tour="model-parameters">
       {/* Title */}
diff --git a/DashAI/front/src/components/generative/DocumentReferencesModal.jsx b/DashAI/front/src/components/generative/DocumentReferencesModal.jsx
new file mode 100644
index 000000000..8f3300da6
--- /dev/null
+++ b/DashAI/front/src/components/generative/DocumentReferencesModal.jsx
@@ -0,0 +1,165 @@
+import React, { useState } from 'react';
+import { useTranslation } from 'react-i18next';
+import {
+  Box,
+  Typography,
+  IconButton,
+  List,
+  ListItem,
+  Divider,
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+} from '@mui/material';
+import {
+  Close as CloseIcon,
+  Article as ArticleIcon,
+  ContentCopy as CopyIcon,
+} from '@mui/icons-material';
+
+const DocumentReferencesModal = ({ open, onClose, document, chunks, onOpenReference }) => {
+  const { t } = useTranslation('generative');
+  if (!document || !chunks) return null;
+
+  const getDocumentTitle = (docId, chunks) => {
+    const firstChunk = chunks[0];
+    if (firstChunk.document_title) return firstChunk.document_title;
+    if (firstChunk.document_name) return firstChunk.document_name;
+    if (firstChunk.title) return firstChunk.title;
+    if (firstChunk.name) return firstChunk.name;
+    return t('documentReferences.fallbackTitle', { id: docId, defaultValue: `Document ${docId}` });
+  };
+
+  const handleCopyChunk = async (chunkText) => {
+    try {
+      const cleanText = chunkText.replace(/\\n/g, '\n');
+      await navigator.clipboard.writeText(cleanText);
+      // You could add a toast notification here if you have a notification system
+    } catch (err) {
+      console.error('Failed to copy text: ', err);
+    }
+  };
+
+  return (
+    <Dialog 
+      open={open} 
+      onClose={onClose}
+      maxWidth="md"
+      fullWidth
+      PaperProps={{
+        sx: {
+          borderRadius: 2,
+          maxHeight: '80vh'
+        }
+      }}
+    >
+      <DialogTitle sx={{ 
+        display: 'flex', 
+        alignItems: 'center', 
+        justifyContent: 'space-between',
+        pb: 1
+      }}>
+        <Box sx={{ display: 'flex', alignItems: 'center', gap: 1 }}>
+          <ArticleIcon color="primary" />
+          <Typography variant="h6">
+            {getDocumentTitle(document.id, chunks)}
+          </Typography>
+        </Box>
+        <IconButton onClick={onClose} size="small">
+          <CloseIcon />
+        </IconButton>
+      </DialogTitle>
+
+      <DialogContent dividers sx={{ p: 0 }}>
+        <Typography variant="body2" sx={{ p: 2, pb: 1, color: 'text.secondary' }}>
+          {t('documentReferences.chunksCount', { count: chunks.length })}
+        </Typography>
+        
+        <List dense disablePadding>
+          {chunks.map((chunk, index) => (
+            <React.Fragment key={chunk.key}>
+              <ListItem disablePadding>
+                <Box sx={{ width: '100%', p: 2 }}>
+                  <Box sx={{ 
+                    display: 'flex', 
+                    alignItems: 'center', 
+                    justifyContent: 'space-between',
+                    mb: 1 
+                  }}>
+                    <Typography 
+                      variant="subtitle2" 
+                      sx={{ 
+                        fontWeight: 'bold',
+                        color: 'text.primary',
+                        display: 'flex',
+                        alignItems: 'center',
+                        gap: 1
+                      }}
+                    >
+                      <Box
+                        sx={{
+                          width: 6,
+                          height: 6,
+                          borderRadius: '50%',
+                          backgroundColor: 'primary.main',
+                          flexShrink: 0,
+                        }}
+                      />
+                      {chunk.document_position 
+                        ? t('documentReferences.chunkLabel', { position: chunk.document_position })
+                        : t('documentReferences.chunkLabel', { position: index + 1 })
+                      }
+                    </Typography>
+                    
+                    <IconButton 
+                      size="small" 
+                      onClick={() => handleCopyChunk(chunk.text)}
+                      sx={{ 
+                        color: 'text.secondary',
+                        '&:hover': {
+                          color: 'primary.main',
+                          backgroundColor: 'action.hover'
+                        }
+                      }}
+                    >
+                      <CopyIcon fontSize="small" />
+                    </IconButton>
+                  </Box>
+                  
+                  <Typography 
+                    variant="body2" 
+                    sx={{ 
+                      color: 'text.primary',
+                      lineHeight: 1.6,
+                      whiteSpace: 'pre-wrap',
+                      backgroundColor: 'background.default',
+                      p: 1.5,
+                      borderRadius: 1,
+                      border: 1,
+                      borderColor: 'divider'
+                    }}
+                  >
+                    {chunk.text.replace(/\\n/g, '\n')}
+                  </Typography>
+                </Box>
+              </ListItem>
+              {index < chunks.length - 1 && (
+                <Divider variant="middle" />
+              )}
+            </React.Fragment>
+          ))}
+        </List>
+      </DialogContent>
+
+      <DialogActions sx={{ p: 2 }}>
+        <Button onClick={onClose} variant="contained">
+          {t('documentReferences.close')}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+};
+
+export default DocumentReferencesModal;
\ No newline at end of file
diff --git a/DashAI/front/src/components/generative/GenerativeChat.jsx b/DashAI/front/src/components/generative/GenerativeChat.jsx
index 5f12cd2d3..27b9846fc 100644
--- a/DashAI/front/src/components/generative/GenerativeChat.jsx
+++ b/DashAI/front/src/components/generative/GenerativeChat.jsx
@@ -1,5 +1,4 @@
 import { Box, Divider, IconButton, Typography } from "@mui/material";
-import { useTheme } from "@mui/material/styles";
 import InfoIcon from "@mui/icons-material/Info";
 import ArrowRightAltIcon from "@mui/icons-material/ArrowRightAlt";
 import KeyboardArrowDownIcon from "@mui/icons-material/KeyboardArrowDown";
@@ -17,9 +16,15 @@ import { getHistoryBySessionId, getSessionById } from "../../api/session";
 import InfoSessionModal from "./InfoSessionModal";
 import { useSnackbar } from "notistack";
 import { MediaInput } from "./MediaInput";
+import JobQueueWidget from "../jobs/JobQueueWidget";
+import { getRunStatus } from "../../utils/runStatus";
+import TemplateModal from "../custom/TemplateModal";
+import SourcesDisplay from "./SourcesDisplay";
+import RAGBreadcrumbs from "./RAG/RAGBreadcrumbs";
 import { Trans, useTranslation } from "react-i18next";
 import { useGenerative } from "./GenerativeContext";
 import { useTourContext } from "../tour/TourProvider";
+import { useTheme } from "@mui/material/styles";
 
 export default function GenerativeChat() {
   const theme = useTheme();
@@ -45,9 +50,13 @@ export default function GenerativeChat() {
   const [showScrollButton, setShowScrollButton] = useState(false);
   const [sessionInfo, setSessionInfo] = useState(null);
   const [sessionInfoVisible, setSessionInfoVisible] = useState(false);
+  const [referenceModalOpen, setReferenceModalOpen] = useState(false);
+  const [selectedReferenceText, setSelectedReferenceText] = useState("");
+  const [referenceModalTitle, setReferenceModalTitle] = useState("");
   const { enqueueSnackbar } = useSnackbar();
   const { t } = useTranslation(["generative"]);
   const tourContext = useTourContext();
+  const [shouldAutoScroll, setShouldAutoScroll] = useState(true);
 
   const scrollToBottom = (force = false) => {
     const el = chatContainerRef.current;
@@ -61,6 +70,20 @@ export default function GenerativeChat() {
     }
   };
 
+  const isAtBottom = () => {
+    if (!chatContainerRef.current) return true;
+    const { scrollTop, scrollHeight, clientHeight } = chatContainerRef.current;
+    return Math.abs(scrollHeight - clientHeight - scrollTop) < 5; // 5px threshold
+  };
+
+  const handleOpenReference = (ref, key) => {
+    const title = `Document ${ref.document_id}${ref.document_name ? ` (${ref.document_name})` : ''}${ref.document_position ? ` - Chunk ${ref.document_position}` : ''}`;
+    setReferenceModalTitle(title);
+    // Convert escaped newlines to actual newlines
+    setSelectedReferenceText(ref.text.replace(/\\n/g, '\n'));
+    setReferenceModalOpen(true);
+  };
+
   const handleScroll = () => {
     const el = chatContainerRef.current;
     if (!el) return;
@@ -79,6 +102,7 @@ export default function GenerativeChat() {
 
   const getMessages = () => {
     getProcessesBySessionId(sessionId).then((response) => {
+      console.log('Fetched messages:', response); // Add here
       setIsLoadingMessage(false);
       setMessages(response);
     });
@@ -92,6 +116,7 @@ export default function GenerativeChat() {
 
   const handleSendMessage = (input) => {
     setIsLoadingMessage(true);
+    setShouldAutoScroll(true); // Enable auto-scroll when sending new message
 
     postProcess(sessionId, input).then((response) => {
       // Add the new message to the chat
@@ -186,13 +211,78 @@ export default function GenerativeChat() {
   }, [messages]);
 
   useEffect(() => {
+    console.log('Combining messages and history for display'); // Add here
+    console.log('Messages:', messages);
+    console.log('TASK NAME:', taskName);
+    console.log('session name:', sessionInfo?.name);
+    console.log('session description:', sessionInfo?.description);
     let messagesObject = messages.map((process) => {
+      // Check if there's reference data in the output (only for RAGTask)
+      let referenceOutput = null;
+      let mainOutput = process.output;
+      
+      if (taskName === "RAGTask" && process.output && process.output.length > 1) {
+        // Look for Dict type output that contains reference information
+        const referenceItem = process.output.find(item => item.data_type === "Dict");
+        if (referenceItem) {
+          console.log('Raw reference data:', referenceItem.data);
+          try {
+            // The data might be a Python dict string, try to parse it as JSON
+            let dataStr = referenceItem.data;
+            
+            // If it starts with { but isn't valid JSON, it might be a Python dict
+            // Try to convert Python dict format to JSON format
+            if (dataStr.startsWith('{') && !dataStr.startsWith('{"')) {
+              // Replace Python dict format with JSON format
+              dataStr = dataStr
+                .replace(/'/g, '"')  // Replace single quotes with double quotes
+                .replace(/True/g, 'true')  // Replace Python True with JSON true
+                .replace(/False/g, 'false')  // Replace Python False with JSON false
+                .replace(/None/g, 'null');  // Replace Python None with JSON null
+            }
+            
+            console.log('Processed data string:', dataStr);
+            const parsedData = JSON.parse(dataStr);
+            referenceOutput = parsedData;
+            // Keep only non-Dict outputs as main output
+            mainOutput = process.output.filter(item => item.data_type !== "Dict");
+          } catch (e) {
+            console.log('Could not parse reference data:', e);
+            console.log('Original data:', referenceItem.data);
+            // If parsing fails, try to extract info using regex as fallback for multiple references
+            try {
+              // Updated regex to capture all fields: document_id, document_name, document_position, text
+              const matches = [...referenceItem.data.matchAll(/(\d+):\s*\{\s*['"]?document_id['"]?\s*:\s*(\d+).*?['"]?document_name['"]?\s*:\s*['"]([^'"]*)['"]\s*.*?['"]?document_position['"]?\s*:\s*(\d+).*?['"]?text['"]?\s*:\s*['"]([^'"]*)['"]/gs)];
+              console.log('Regex matches for references:', matches);
+              
+              if (matches.length > 0) {
+                referenceOutput = {};
+                matches.forEach((match) => {
+                  const refId = match[1];
+                  referenceOutput[refId] = {
+                    document_id: parseInt(match[2]),
+                    document_name: match[3],
+                    document_position: parseInt(match[4]),
+                    text: match[5]
+                  };
+                });
+                mainOutput = process.output.filter(item => item.data_type !== "Dict");
+                console.log('Fallback parsing successful:', referenceOutput);
+              }
+            } catch (fallbackError) {
+              console.log('Fallback parsing also failed:', fallbackError);
+            }
+          }
+        }
+      }
+
       return {
         type: "message",
         timestamp: process.created,
         id: process.id,
         input: process.input,
-        output: process.output,
+        output: mainOutput,
+        referenceOutput: referenceOutput,
         status: process.status,
         end_time: process.end_time,
       };
@@ -212,8 +302,8 @@ export default function GenerativeChat() {
               whiteSpace: "pre-wrap",
             }}
           >
-            {change.parameter}: {change.oldValue}{" "}
-            <ArrowRightAltIcon fontSize="small" /> {change.newValue}{" "}
+              {change.parameter}: {formatHistoryValue(change.oldValue)}{" "}
+              <ArrowRightAltIcon fontSize="small" /> {formatHistoryValue(change.newValue)}{" "}
           </span>
         )),
       };
@@ -226,6 +316,34 @@ export default function GenerativeChat() {
     setMessagesWithHistory(combinedMessages);
   }, [messages, history]);
 
+  const formatHistoryValue = (value) => {
+    if (value === null || value === undefined) {
+      return "";
+    }
+
+    if (typeof value === "string") {
+      return value;
+    }
+
+    if (typeof value === "number" || typeof value === "boolean") {
+      return String(value);
+    }
+
+    if (typeof value === "object") {
+      if (value.component) {
+        return value.component;
+      }
+
+      try {
+        return JSON.stringify(value);
+      } catch (error) {
+        return String(value);
+      }
+    }
+
+    return String(value);
+  };
+
   return (
     <Box
       display="flex"
@@ -236,6 +354,13 @@ export default function GenerativeChat() {
       height={"100%"}
       sx={{ overflow: "hidden", minHeight: 0 }}
     >
+      {/* RAG Breadcrumbs - only show for RAG tasks */}
+      {taskName === "RAGTask" && (
+        <Box sx={{ width: "100%", px: 2, pt: 2 }}>
+          <RAGBreadcrumbs sessionName={sessionInfo?.name} />
+        </Box>
+      )}
+
       {/* Model display */}
       <Box
         sx={{
@@ -285,13 +410,13 @@ export default function GenerativeChat() {
           flexDirection="column"
           justifyContent="flex-start"
           alignItems="flex-start"
-          gap={4}
+          gap={1}
           width={"100%"}
           flex={1}
           minHeight={0}
           overflow={"auto"}
-          mt={4}
-          p={8}
+          mt={1}
+          p={2}
           ref={chatContainerRef}
           onScroll={handleScroll}
         >
@@ -303,10 +428,10 @@ export default function GenerativeChat() {
                 flexDirection="column"
                 justifyContent="flex-start"
                 flexGrow={0}
-                gap={4}
+                gap={1}
                 width={"100%"}
                 //height={"100%"}
-                mt={4}
+                mt={1}
               >
                 {message.type === "history" ? (
                   <Typography variant="body1" sx={{ opacity: 0.8 }}>
@@ -319,19 +444,53 @@ export default function GenerativeChat() {
                     <ChatBubble
                       messages={message.input}
                       sender={"User"}
-                      timestamp={new Date(
-                        message.timestamp,
-                      ).toLocaleTimeString()}
+                      timestamp={new Date(message.timestamp).toLocaleTimeString()}
                       isUser={true}
                     />
                     {message.status === 3 ? (
-                      <ChatBubble
-                        messages={message.output}
-                        sender={"Model"}
-                        timestamp={new Date(
-                          message.end_time,
-                        ).toLocaleTimeString()}
-                      />
+                      <>
+                        {taskName === "RAGTask" && message.referenceOutput ? (
+                          // For RAG messages, we need custom layout to insert sources before timestamp
+                          <>
+                            <ChatBubble
+                              messages={message.output}
+                              sender={"Model"}
+                              timestamp={null} // We'll handle timestamp separately
+                            />
+                            <SourcesDisplay 
+                              references={message.referenceOutput}
+                              onOpenReference={handleOpenReference}
+                              isUser={false}
+                            />
+                            {/* Add timestamp after sources with proper alignment */}
+                            <Box sx={{ 
+                              ml: '40px', // Same alignment as sources and message content
+                              mt: 1,
+                              display: 'flex',
+                              justifyContent: 'flex-start'
+                            }}>
+                              <Box
+                                sx={{
+                                  fontSize: "0.75rem",
+                                  color: "text.secondary",
+                                  opacity: 0.7,
+                                }}
+                              >
+                                {new Date(message.end_time).toLocaleTimeString()}
+                              </Box>
+                            </Box>
+                          </>
+                        ) : (
+                          // For non-RAG messages, use normal ChatBubble with timestamp
+                          <ChatBubble
+                            messages={message.output}
+                            sender={"Model"}
+                            timestamp={new Date(
+                              message.end_time,
+                            ).toLocaleTimeString()}
+                          />
+                        )}
+                      </>
                     ) : (
                       <ChatBubble isWaiting={true} sender="Model" />
                     )}
diff --git a/DashAI/front/src/components/generative/InfoSessionModal.jsx b/DashAI/front/src/components/generative/InfoSessionModal.jsx
index c8b9b7451..600830ab5 100644
--- a/DashAI/front/src/components/generative/InfoSessionModal.jsx
+++ b/DashAI/front/src/components/generative/InfoSessionModal.jsx
@@ -96,18 +96,30 @@ export default function InfoSessionModal({ sessionData, open, onClose }) {
             <Table size="small">
               <TableBody>
                 {Object.entries(sessionData.parameters || {}).map(
-                  ([key, value]) => (
-                    <TableRow key={key}>
-                      <TableCell
-                        component="th"
-                        scope="row"
-                        sx={{ color: "text.secondary" }}
-                      >
-                        {key.replace(/_/g, " ")}
-                      </TableCell>
-                      <TableCell align="right">{value}</TableCell>
-                    </TableRow>
-                  ),
+                  ([key, value]) => {
+                    let displayValue = value;
+                    if (
+                      typeof value === "object" &&
+                      value !== null &&
+                      !Array.isArray(value)
+                    ) {
+                      displayValue = JSON.stringify(value, null, 2);
+                    } else if (Array.isArray(value)) {
+                      displayValue = JSON.stringify(value);
+                    }
+                    return (
+                      <TableRow key={key}>
+                        <TableCell
+                          component="th"
+                          scope="row"
+                          sx={{ color: "text.secondary" }}
+                        >
+                          {key.replace(/_/g, " ")}
+                        </TableCell>
+                        <TableCell align="right">{displayValue}</TableCell>
+                      </TableRow>
+                    );
+                  },
                 )}
               </TableBody>
             </Table>
diff --git a/DashAI/front/src/components/generative/MainGenerativeBox.jsx b/DashAI/front/src/components/generative/MainGenerativeBox.jsx
new file mode 100644
index 000000000..2cc0250ed
--- /dev/null
+++ b/DashAI/front/src/components/generative/MainGenerativeBox.jsx
@@ -0,0 +1,20 @@
+import React from "react";
+import { Box } from "@mui/material";
+
+export default function MainGenerativeBox({ children }) {
+  return (
+    <Box
+      width={"100%"}
+      height={"100%"}
+      sx={{
+        border: "0.2px solid",
+        borderColor: "grey.600",
+      }}
+      borderRadius={2}
+      //bgcolor={"background.box"}
+      p={2}
+    >
+      {children}
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/MediaInput.jsx b/DashAI/front/src/components/generative/MediaInput.jsx
index fd7f311e9..7bfe3ac68 100644
--- a/DashAI/front/src/components/generative/MediaInput.jsx
+++ b/DashAI/front/src/components/generative/MediaInput.jsx
@@ -149,7 +149,6 @@ export function MediaInput({
               />
             );
           })}
-
           <Button
             variant="contained"
             color="primary"
diff --git a/DashAI/front/src/components/generative/RAG/DocumentList.jsx b/DashAI/front/src/components/generative/RAG/DocumentList.jsx
new file mode 100644
index 000000000..e9abdc48b
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentList.jsx
@@ -0,0 +1,62 @@
+import React, { useState } from "react";
+import { Box } from "@mui/material";
+import DocumentListItem from "./DocumentListItem";
+import DocumentPreviewModal from "./DocumentPreviewModal";
+import { normalizeUrl } from "../../../utils/urlUtils";
+
+export default function DocumentList({ documents }) {
+  const [previewOpen, setPreviewOpen] = useState(false);
+  const [previewDoc, setPreviewDoc] = useState(null);
+  const [txtContent, setTxtContent] = useState("");
+
+  const handleOpenPreview = async (doc) => {
+    setPreviewDoc(doc);
+    if (doc.file_type === "txt" && doc.preview) {
+      try {
+        const res = await fetch(normalizeUrl(doc.preview));
+        const text = await res.text();
+        setTxtContent(text);
+      } catch (e) {
+        console.error("Error loading TXT:", e);
+        setTxtContent("Error loading TXT file");
+      }
+    }
+    setPreviewOpen(true);
+  };
+
+  const handleClosePreview = () => {
+    setPreviewOpen(false);
+    setPreviewDoc(null);
+    setTxtContent("");
+  };
+
+  return (
+    <>
+      <Box
+        sx={{
+          display: "flex",
+          flexDirection: "column",
+          gap: 1.5,
+          width: "100%",
+          minWidth: 0,
+          maxWidth: "100%",
+        }}
+      >
+        {documents.map((document) => (
+          <DocumentListItem
+            key={document.id}
+            document={document}
+            disabled={false}
+            onClick={() => handleOpenPreview(document)}
+          />
+        ))}
+      </Box>
+      <DocumentPreviewModal
+        open={previewOpen}
+        onClose={handleClosePreview}
+        document={previewDoc}
+        txtContent={txtContent}
+      />
+    </>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/DocumentListItem.jsx b/DashAI/front/src/components/generative/RAG/DocumentListItem.jsx
new file mode 100644
index 000000000..445195015
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentListItem.jsx
@@ -0,0 +1,126 @@
+import React, { useState } from "react";
+import { Box, Typography, useTheme } from "@mui/material";
+import {
+  Description,
+  PictureAsPdf,
+  InsertDriveFile,
+  Article,
+} from "@mui/icons-material";
+import { useTranslation } from "react-i18next";
+
+const getDocumentIcon = (fileType) => {
+  const iconMap = {
+    pdf: PictureAsPdf,
+    txt: Article,
+  };
+
+  const IconComponent = iconMap[fileType?.toLowerCase()] || InsertDriveFile;
+  return IconComponent;
+};
+
+export default function DocumentListItem({
+  document,
+  disabled = false,
+  onClick,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [isHovered, setIsHovered] = useState(false);
+  const theme = useTheme();
+
+  const DocumentIcon = getDocumentIcon(document.type);
+  const colorMap = { pdf: "primary.main", txt: "primary.main" };
+  const documentColor =
+    colorMap[document.type?.toLowerCase()] || theme.palette.text.disabled;
+
+  const disabledOverlay = `repeating-linear-gradient(45deg, transparent, transparent 10px, ${theme.palette.action.disabled} 10px, ${theme.palette.action.disabled} 20px)`;
+
+  return (
+    <Box
+      onMouseEnter={() => setIsHovered(true)}
+      onMouseLeave={() => setIsHovered(false)}
+      onClick={disabled ? null : onClick}
+      sx={{
+        display: "flex",
+        alignItems: "center",
+        gap: 1.5,
+        p: 1.5,
+        width: "100%",
+        minWidth: 0,
+        maxWidth: "100%",
+        bgcolor: disabled ? "action.disabledBackground" : "background.paper",
+        border: 1,
+        borderColor: "divider",
+        borderRadius: 1,
+        cursor: disabled ? "not-allowed" : "pointer",
+        transition: "all 0.2s",
+        opacity: disabled ? 0.5 : 1,
+        filter: disabled ? "grayscale(0.6)" : "none",
+        position: "relative",
+        "&:hover": {
+          bgcolor: disabled ? "action.disabledBackground" : "action.hover",
+          borderColor: disabled ? "divider" : documentColor,
+          transform: disabled ? "none" : "translateX(4px)",
+        },
+        "&::after": disabled
+          ? {
+              content: '""',
+              position: "absolute",
+              inset: 0,
+              borderRadius: 1,
+              pointerEvents: "none",
+              background: disabledOverlay,
+            }
+          : {},
+      }}
+    >
+      <Box
+        sx={{
+          display: "flex",
+          alignItems: "center",
+          justifyContent: "center",
+          width: 36,
+          height: 36,
+          borderRadius: 1,
+          bgcolor: "action.selected",
+          color: disabled ? "text.disabled" : "text.primary",
+          flexShrink: 0,
+          transition: "all 0.2s",
+        }}
+      >
+        <DocumentIcon sx={{ fontSize: 20 }} />
+      </Box>
+
+      <Box sx={{ flex: 1, minWidth: 0 }}>
+        <Box
+          sx={{
+            display: "flex",
+            alignItems: "center",
+            gap: 1,
+            mb: 0.5,
+          }}
+        >
+          <Typography
+            variant="body2"
+            sx={{
+              color: disabled ? "text.disabled" : "text.primary",
+              overflow: "hidden",
+              textOverflow: "ellipsis",
+              whiteSpace: "nowrap",
+            }}
+          >
+            {document.name}
+          </Typography>
+        </Box>
+        <Typography
+          variant="caption"
+          sx={{
+            color: disabled ? "text.disabled" : "text.secondary",
+            textTransform: "uppercase",
+          }}
+        >
+          {document.type || t("generative:rag.documents.table.unknownType")}
+        </Typography>
+      </Box>
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/DocumentPreviewModal.jsx b/DashAI/front/src/components/generative/RAG/DocumentPreviewModal.jsx
new file mode 100644
index 000000000..42be717bc
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentPreviewModal.jsx
@@ -0,0 +1,48 @@
+import { Dialog, DialogTitle, DialogContent, Box } from "@mui/material";
+import { createPortal } from "react-dom";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { normalizeUrl } from "../../../utils/urlUtils";
+
+export default function DocumentPreviewModal({
+  open,
+  onClose,
+  document,
+  txtContent,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const fileType = document?.file_type || document?.type;
+  const preview = normalizeUrl(document?.preview);
+
+  const dialogContent = (
+    <Dialog open={open} onClose={onClose} maxWidth="md" fullWidth>
+      <DialogTitle>{t("generative:rag.documentPreview.title")}</DialogTitle>
+      <DialogContent>
+        {document && fileType === "pdf" && preview && (
+          <iframe
+            src={preview}
+            title={t("generative:rag.documentPreview.pdfPreview")}
+            width="100%"
+            height="600px"
+            style={{ border: 0 }}
+          />
+        )}
+        {document && fileType === "txt" && (
+          <pre style={{ maxHeight: 600, overflow: "auto" }}>{txtContent}</pre>
+        )}
+        {document && !["pdf", "txt"].includes(fileType) && (
+          <Box>{t("generative:rag.documentPreview.noPreview")}</Box>
+        )}
+      </DialogContent>
+    </Dialog>
+  );
+
+  return createPortal(dialogContent, globalThis.document.body);
+}
+
+DocumentPreviewModal.propTypes = {
+  open: PropTypes.bool.isRequired,
+  onClose: PropTypes.func.isRequired,
+  document: PropTypes.object,
+  txtContent: PropTypes.string,
+};
diff --git a/DashAI/front/src/components/generative/RAG/DocumentSelector.jsx b/DashAI/front/src/components/generative/RAG/DocumentSelector.jsx
new file mode 100644
index 000000000..e6b07b59d
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentSelector.jsx
@@ -0,0 +1,415 @@
+import { Box, Button, Dialog, IconButton, Tooltip, Typography } from "@mui/material";
+import { useEffect, useState, useCallback, useRef, useMemo } from "react";
+import PropTypes from "prop-types";
+import AddIcon from "@mui/icons-material/AddCircleOutline";
+import { Visibility, Delete } from "@mui/icons-material";
+import { useTranslation } from "react-i18next";
+import { useTheme } from "@mui/material/styles";
+import {
+  MaterialReactTable,
+  useMaterialReactTable,
+  MRT_GlobalFilterTextField,
+} from "material-react-table";
+import { MRT_Localization_ES } from "material-react-table/locales/es";
+import { MRT_Localization_EN } from "material-react-table/locales/en";
+import Upload from "../../shared/Upload";
+import { loadDocuments, addDocument, deleteDocument } from "../../../api/rag";
+import { formatDate } from "../../../utils";
+import { normalizeUrl } from "../../../utils/urlUtils";
+import DocumentPreviewModal from "./DocumentPreviewModal";
+import RAGSectionColumn from "../../../pages/generative/RAGSession/components/RAGSectionColumn";
+
+export default function DocumentSelector({
+  selectedIds: initialSelectedIds = [],
+  onSelect,
+  sx,
+}) {
+  const { t, i18n } = useTranslation(["generative"]);
+  const theme = useTheme();
+  const [documents, setDocuments] = useState([]);
+  const [selectedIds, setSelectedIds] = useState(initialSelectedIds);
+  const [isLoading, setIsLoading] = useState(true);
+  const [uploadOpen, setUploadOpen] = useState(false);
+
+  const [previewOpen, setPreviewOpen] = useState(false);
+  const [previewDoc, setPreviewDoc] = useState(null);
+  const [txtContent, setTxtContent] = useState("");
+
+  const previousSelectedIdsRef = useRef(JSON.stringify([...initialSelectedIds].map(String).sort()));
+
+  const localization = i18n.language.startsWith("es")
+    ? MRT_Localization_ES
+    : MRT_Localization_EN;
+
+  const tableData = useMemo(
+    () =>
+      documents.map((doc) => ({
+        ...doc,
+        preview: doc.file_url,
+        file_type: doc.file_name.split(".").pop().toLowerCase(),
+      })),
+    [documents],
+  );
+
+  const getNormalizedIdsKey = (ids) =>
+    JSON.stringify([...ids].map(String).sort());
+
+  useEffect(() => {
+    const fetchDocuments = async () => {
+      setIsLoading(true);
+      try {
+        const docs = await loadDocuments();
+        const sortedDocs = docs.sort((a, b) => {
+          const dateA = new Date(a.created);
+          const dateB = new Date(b.created);
+          return dateB - dateA;
+        });
+        setDocuments(sortedDocs);
+      } catch (error) {
+        console.error("Failed to load documents:", error);
+      } finally {
+        setIsLoading(false);
+      }
+    };
+
+    fetchDocuments();
+  }, []);
+
+  useEffect(() => {
+    if (getNormalizedIdsKey(selectedIds) !== getNormalizedIdsKey(initialSelectedIds)) {
+      setSelectedIds([...initialSelectedIds]);
+    }
+  }, [initialSelectedIds]);
+
+  useEffect(() => {
+    const currentKey = getNormalizedIdsKey(selectedIds);
+    if (currentKey !== previousSelectedIdsRef.current) {
+      const selectedIdSet = new Set(selectedIds.map(String));
+      const selectedDocs = documents.filter((doc) =>
+        selectedIdSet.has(String(doc.id)),
+      );
+      onSelect?.(selectedDocs);
+      previousSelectedIdsRef.current = currentKey;
+    }
+  }, [selectedIds, documents, onSelect]);
+
+  const handleToggleSelection = useCallback((id) => {
+    setSelectedIds((prev) => {
+      const newSelected = prev.includes(id)
+        ? prev.filter((x) => x !== id)
+        : [...prev, id];
+      return newSelected;
+    });
+  }, []);
+
+  const handleOpenPreview = async (doc) => {
+    setPreviewDoc(doc);
+    if (doc.file_type === "txt" && doc.preview) {
+      try {
+        const res = await fetch(normalizeUrl(doc.preview));
+        const text = await res.text();
+        setTxtContent(text);
+      } catch (e) {
+        setTxtContent("Error loading TXT file");
+      }
+    }
+    setPreviewOpen(true);
+  };
+
+  const handleClosePreview = () => {
+    setPreviewOpen(false);
+    setPreviewDoc(null);
+    setTxtContent("");
+  };
+
+  const handleSelectAll = useCallback(() => {
+    setSelectedIds(documents.map((doc) => doc.id));
+  }, [documents]);
+
+  const handleDeselectAll = useCallback(() => {
+    setSelectedIds([]);
+  }, []);
+
+  const handleAddDocument = useCallback(async (newDoc) => {
+    try {
+      const savedDoc = await addDocument(newDoc);
+      return savedDoc;
+    } catch (error) {
+      console.error("Failed to add document:", error);
+      return null;
+    }
+  }, []);
+
+  const handleRemoveDocument = useCallback(async (id) => {
+    try {
+      await deleteDocument(id);
+      setDocuments((prev) => prev.filter((doc) => doc.id !== id));
+      setSelectedIds((prev) => prev.filter((x) => x !== id));
+    } catch (error) {
+      console.error("Failed to delete document:", error);
+    }
+  }, []);
+
+  const handleFileUpload = useCallback(
+    async (files, url) => {
+      if (!files) return;
+
+      console.log("Files to upload at handleFileUpload:", files);
+
+      const fileList = Array.isArray(files) ? files : [files];
+      const uploadedDocuments = [];
+
+      for (const file of fileList) {
+        const docToAdd = {
+          file,
+          optional_metadata: {
+            name: file.name,
+            source: url || "local_upload",
+          },
+        };
+        const savedDoc = await handleAddDocument(docToAdd);
+        if (savedDoc) {
+          uploadedDocuments.push(savedDoc);
+        }
+      }
+
+      if (uploadedDocuments.length > 0) {
+        setDocuments((prev) => {
+          const nextDocs = [...uploadedDocuments, ...prev];
+          return nextDocs.filter(
+            (doc, index, array) =>
+              index === array.findIndex((candidate) => candidate.id === doc.id),
+          );
+        });
+        setSelectedIds((prev) => {
+          const nextSelected = new Set(prev.map(String));
+          uploadedDocuments.forEach((doc) => nextSelected.add(String(doc.id)));
+          return Array.from(nextSelected);
+        });
+      }
+
+      setUploadOpen(false);
+    },
+    [handleAddDocument],
+  );
+
+  const selectedIdSet = useMemo(
+    () => new Set(selectedIds.map((id) => String(id))),
+    [selectedIds],
+  );
+
+  const columns = useMemo(
+    () => [
+      {
+        accessorKey: "file_name",
+        header: t("generative:rag.documents.table.name"),
+        size: 250,
+        Cell: ({ row }) => row.original.file_name,
+      },
+      {
+        accessorKey: "file_type",
+        header: t("generative:rag.documents.table.type"),
+        size: 80,
+        Cell: ({ row }) => row.original.file_type?.toUpperCase() || "-",
+      },
+      {
+        accessorKey: "created",
+        header: t("generative:rag.documents.table.created"),
+        size: 150,
+        Cell: ({ row }) => formatDate(row.original.created) || "-",
+      },
+      {
+        id: "actions",
+        header: t("generative:rag.documents.table.actions"),
+        size: 100,
+        enableSorting: false,
+        enableColumnFilter: false,
+        Cell: ({ row }) => (
+          <Box sx={{ display: "flex", gap: 0.5, justifyContent: "center" }}>
+            <Tooltip title={t("generative:rag.documents.table.preview")}>
+              <IconButton
+                size="small"
+                onClick={() => handleOpenPreview(row.original)}
+              >
+                <Visibility fontSize="small" />
+              </IconButton>
+            </Tooltip>
+            <Tooltip title={t("generative:rag.documents.table.delete")}>
+              <IconButton
+                size="small"
+                onClick={() => handleRemoveDocument(row.original.id)}
+                color="error"
+              >
+                <Delete fontSize="small" />
+              </IconButton>
+            </Tooltip>
+          </Box>
+        ),
+      },
+    ],
+    [handleRemoveDocument, t],
+  );
+
+  const rowSelection = useMemo(
+    () =>
+      tableData.reduce((acc, doc) => {
+        acc[String(doc.id)] = selectedIdSet.has(String(doc.id));
+        return acc;
+      }, {}),
+    [tableData, selectedIdSet],
+  );
+
+  const table = useMaterialReactTable({
+    columns,
+    data: tableData,
+    enableSelectAll: true,
+    enableRowSelection: true,
+    selectAllMode: "all",
+    enableColumnOrdering: false,
+    enableColumnActions: false,
+    enableColumnHiding: true,
+    enableDensityToggle: false,
+    enableFullScreenToggle: false,
+    enablePagination: true,
+    enableBottomToolbar: true,
+    enableTopToolbar: true,
+    enableGlobalFilter: true,
+    initialState: {
+      columnVisibility: {
+        file_type: false,
+        created: false,
+      },
+      pagination: {
+        pageIndex: 0,
+        pageSize: 5,
+      },
+    },
+    muiPaginationProps: {
+      rowsPerPageOptions: [5, 10, 25, 50],
+      showFirstButton: false,
+      showLastButton: false,
+    },
+    muiTablePaperProps: {
+      sx: {
+        boxShadow: "none",
+        borderRadius: 1,
+        display: "flex",
+        flexDirection: "column",
+      },
+    },
+    muiTableContainerProps: {
+      sx: {
+        maxHeight: "none",
+        flex: 1,
+      },
+    },
+    muiTableHeadCellProps: {
+      sx: {
+        backgroundColor: theme.palette.action.hover,
+      },
+    },
+    muiTableBodyCellProps: {
+      sx: {
+        padding: "8px 16px",
+      },
+    },
+    muiTableBodyRowProps: ({ row }) => ({
+      sx: {
+        backgroundColor: selectedIdSet.has(String(row.original.id))
+          ? theme.palette.action.selected
+          : "inherit",
+        "&:hover": {
+          backgroundColor: theme.palette.action.hover,
+        },
+      },
+    }),
+    rowNumberDisplayMode: "hidden",
+    renderTopToolbarCustomActions: () => (
+      <Box sx={{ display: "flex", gap: 1, alignItems: "center" }}>
+        <MRT_GlobalFilterTextField table={table} />
+      </Box>
+    ),
+    state: {
+      rowSelection,
+      isLoading,
+    },
+    onRowSelectionChange: (updater) => {
+      const nextRowSelection =
+        typeof updater === "function" ? updater(rowSelection) : updater;
+
+      const nextSelectedIdSet = new Set(
+        Object.entries(nextRowSelection)
+          .filter(([, isSelected]) => Boolean(isSelected))
+          .map(([rowId]) => String(rowId)),
+      );
+
+      tableData.forEach((doc) => {
+        const idKey = String(doc.id);
+        const wasSelected = selectedIdSet.has(idKey);
+        const isSelectedNow = nextSelectedIdSet.has(idKey);
+        if (wasSelected !== isSelectedNow) {
+          handleToggleSelection(doc.id);
+        }
+      });
+    },
+    getRowId: (row) => String(row.id),
+    localization,
+  });
+
+  return (
+    <RAGSectionColumn sx={sx}>
+      <Typography variant="body2" color="textSecondary">
+        {t("generative:rag.setup.selectDocumentsDescription")}
+      </Typography>
+      
+      <MaterialReactTable table={table} />
+
+      <Button
+        variant="contained"
+        color="primary"
+        size="small"
+        startIcon={<AddIcon />}
+        onClick={() => setUploadOpen(true)}
+        sx={{ alignSelf: "flex-start", width: "fit-content" }}
+      >
+        {t("generative:rag.documents.uploadButton")}
+      </Button>
+
+      <Dialog
+        open={uploadOpen}
+        onClose={() => setUploadOpen(false)}
+        maxWidth="sm"
+        fullWidth
+        paperProps={{
+          sx: {
+            maxHeight: "80vh",
+            minHeight: "300px",
+            display: "flex",
+            flexDirection: "column",
+          }
+        }}
+      >
+        <Upload
+          onFileUpload={handleFileUpload}
+          multiple={true}
+          emptyUploadText={t("generative:rag.documents.emptyUploadText")}
+        />
+
+      </Dialog>
+      <DocumentPreviewModal
+        open={previewOpen}
+        onClose={handleClosePreview}
+        document={previewDoc}
+        txtContent={txtContent}
+      />
+    </RAGSectionColumn>
+  );
+}
+
+DocumentSelector.propTypes = {
+  selectedIds: PropTypes.arrayOf(
+    PropTypes.oneOfType([PropTypes.string, PropTypes.number]),
+  ),
+  onSelect: PropTypes.func,
+  sx: PropTypes.oneOfType([PropTypes.array, PropTypes.object, PropTypes.func]),
+};
diff --git a/DashAI/front/src/components/generative/RAG/DocumentTable.jsx b/DashAI/front/src/components/generative/RAG/DocumentTable.jsx
new file mode 100644
index 000000000..d7d3964a9
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentTable.jsx
@@ -0,0 +1,234 @@
+import React, { useState } from "react";
+import Dialog from "@mui/material/Dialog";
+import Upload from "../../shared/Upload";
+import { addDocument } from "../../../api/rag";
+import AddIcon from "@mui/icons-material/AddCircleOutline";
+import {
+  Paper,
+  Typography,
+  IconButton,
+  Tooltip,
+  LinearProgress,
+  Button,
+  Grid,
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { DataGrid } from "@mui/x-data-grid";
+import VisibilityIcon from "@mui/icons-material/Visibility";
+import { formatDate } from "../../../utils";
+import DeleteItemModal from "../../custom/DeleteItemModal";
+import DocumentPreviewModal from "./DocumentPreviewModal";
+import { normalizeUrl } from "../../../utils/urlUtils";
+import { useTranslation } from "react-i18next";
+
+export default function DocumentTable({
+  documents,
+  onRemove,
+  onAddDocument,
+  isLoading = false,
+  tableTitle = null,
+  showTableTitle = false,
+}) {
+  const { t } = useTranslation(["generative", "common"]);
+  const [previewOpen, setPreviewOpen] = useState(false);
+  const [previewDoc, setPreviewDoc] = useState(null);
+  const [txtContent, setTxtContent] = useState("");
+  const [uploadOpen, setUploadOpen] = useState(false);
+
+  const handleOpenPreview = async (doc) => {
+    setPreviewDoc(doc);
+    if (doc.file_type === "txt" && doc.preview) {
+      try {
+        const res = await fetch(normalizeUrl(doc.preview));
+        const text = await res.text();
+        setTxtContent(text);
+      } catch (e) {
+        setTxtContent("Error loading TXT file");
+      }
+    }
+    setPreviewOpen(true);
+  };
+
+  const handleRemoveDocument = (id) => {
+    if (onRemove) onRemove(id);
+  };
+
+  const handleFileUpload = async (files, url) => {
+    if (!files) return;
+    const fileList = Array.isArray(files) ? files : [files];
+    for (const file of fileList) {
+      const docToAdd = {
+        file,
+        optional_metadata: {
+          name: file.name,
+          source: url || "local_upload",
+        },
+      };
+      try {
+        const savedDoc = await addDocument(docToAdd);
+        if (onAddDocument) onAddDocument(savedDoc);
+      } catch (error) {}
+    }
+    setUploadOpen(false);
+  };
+
+  const handleClosePreview = () => {
+    setPreviewOpen(false);
+    setPreviewDoc(null);
+    setTxtContent("");
+  };
+
+  const columns = [
+    { field: "id", headerName: t("generative:rag.documents.table.id"), flex: 0.1, editable: false },
+    {
+      field: "file_name",
+      headerName: t("generative:rag.documents.table.name"),
+      flex: 0.6,
+      editable: false,
+    },
+    {
+      field: "created",
+      headerName: t("generative:rag.documents.table.created"),
+      flex: 0.4,
+      editable: false,
+      valueGetter: (value) => formatDate(value),
+    },
+    {
+      field: "last_modified",
+      headerName: t("generative:rag.documents.table.lastModified"),
+      flex: 0.4,
+      editable: false,
+      valueGetter: (value, row) => {
+        return row?.optional_metadata?.last_modified
+          ? formatDate(row.optional_metadata.last_modified)
+          : t("common:na");
+      },
+    },
+    {
+      field: "actions",
+      type: "actions",
+      headerName: t("generative:rag.documents.table.actions"),
+      flex: 0.3,
+      getActions: (params) => [
+        <Tooltip title={t("generative:rag.documents.table.preview")} key="preview">
+          <IconButton
+            size="small"
+            onClick={() => handleOpenPreview(params.row)}
+          >
+            <VisibilityIcon fontSize="small" />
+          </IconButton>
+        </Tooltip>,
+        <DeleteItemModal
+          key="delete-button"
+          deleteFromTable={() => handleRemoveDocument(params.row.id)}
+          item="document"
+        />,
+      ],
+    },
+  ];
+
+  return (
+    <Paper sx={{ py: 4, px: 4 }}>
+      {showTableTitle ? (
+        <Grid
+          container
+          justifyContent="space-between"
+          alignItems="center"
+          sx={{ mb: 4 }}
+        >
+          <Typography variant="h5" component="h2">
+            {tableTitle || t("generative:rag.documents.table.currentDocuments")}
+          </Typography>
+          <Button
+            variant="contained"
+            color="primary"
+            startIcon={<AddIcon />}
+            onClick={() => setUploadOpen(true)}
+          >
+            {t("generative:rag.documents.table.addDocument")}
+          </Button>
+        </Grid>
+      ) : (
+        <Grid
+          container
+          justifyContent="flex-end"
+          alignItems="center"
+          sx={{ mb: 4 }}
+        >
+          <Button
+            variant="contained"
+            color="primary"
+            startIcon={<AddIcon />}
+            onClick={() => setUploadOpen(true)}
+          >
+            {t("generative:rag.documents.table.addDocument")}
+          </Button>
+        </Grid>
+      )}
+      {documents.length === 0 && !isLoading ? (
+        <Typography
+          variant="body1"
+          color="warning.main"
+          textAlign="center"
+          mt={16}
+          mx={"auto "}
+        >
+          {t("generative:rag.documents.table.noDocumentsAvailable")}
+        </Typography>
+      ) : (
+        <DataGrid
+          rows={documents}
+          columns={columns}
+          initialState={{
+            pagination: { paginationModel: { pageSize: 5 } },
+          }}
+          pageSizeOptions={[5, 10]}
+          disableRowSelectionOnClick
+          autoHeight
+          loading={isLoading}
+          slots={{
+            loadingOverlay: LinearProgress,
+          }}
+          getRowId={(row) => row.id}
+          sx={{
+            "& .MuiDataGrid-cell:focus": { outline: "none" },
+            minHeight: 300,
+          }}
+        />
+      )}
+      <DocumentPreviewModal
+        open={previewOpen}
+        onClose={handleClosePreview}
+        document={previewDoc}
+        txtContent={txtContent}
+      />
+      <Dialog
+        open={uploadOpen}
+        onClose={() => setUploadOpen(false)}
+        maxWidth="sm"
+        fullWidth
+      >
+        <Upload
+          onFileUpload={handleFileUpload}
+          multiple={true}
+          emptyUploadText={t("generative:rag.documents.table.emptyUploadText")}
+        />
+      </Dialog>
+    </Paper>
+  );
+}
+
+DocumentTable.propTypes = {
+  documents: PropTypes.arrayOf(
+    PropTypes.shape({
+      id: PropTypes.string.isRequired,
+      name: PropTypes.string.isRequired,
+      createdAt: PropTypes.string.isRequired,
+      preview: PropTypes.string,
+    }),
+  ).isRequired,
+  onRemove: PropTypes.func.isRequired,
+  isLoading: PropTypes.bool,
+  tableTitle: PropTypes.string,
+  showTableTitle: PropTypes.bool,
+};
diff --git a/DashAI/front/src/components/generative/RAG/DocumentsBar.jsx b/DashAI/front/src/components/generative/RAG/DocumentsBar.jsx
new file mode 100644
index 000000000..c503422ba
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/DocumentsBar.jsx
@@ -0,0 +1,260 @@
+import { useEffect, useState } from "react";
+import { useTranslation } from "react-i18next";
+import { Box, Typography, Button, Dialog, IconButton, Tooltip } from "@mui/material";
+import AddIcon from "@mui/icons-material/AddCircleOutline";
+import ViewListIcon from "@mui/icons-material/ViewList";
+import { useNavigate } from "react-router-dom";
+import SearchBar from "../../threeSectionLayout/SearchBar";
+import DocumentList from "./DocumentList";
+import Upload from "../../shared/Upload";
+import { useSnackbar } from "notistack";
+import { getSessionDocuments, addDocument, loadDocuments } from "../../../api/rag";
+
+export default function DocumentsBar({ selectedSessionId, taskName, onDocumentChange, showSearch = true }) {
+  const { t } = useTranslation("generative");
+  const [searchQuery, setSearchQuery] = useState("");
+  const [documents, setDocuments] = useState([]);
+  const [filteredDocuments, setFilteredDocuments] = useState([]);
+  const [uploadOpen, setUploadOpen] = useState(false);
+  const { enqueueSnackbar } = useSnackbar();
+  const navigate = useNavigate();
+
+  const goToDocumentsDetail = () => {
+    navigate("/app/generative/RAG/documents");
+  };
+
+  useEffect(() => {
+    const fetchDocuments = async () => {
+      try {
+        let data;
+        if (selectedSessionId) {
+          data = await getSessionDocuments(selectedSessionId);
+        } else {
+          data = await loadDocuments();
+        }
+
+        const transformedDocuments = data.map((doc) => ({
+          id: doc.id,
+          name: doc.file_name,
+          type: doc.file_type,
+          uploadedAt: doc.created,
+          file_name: doc.file_name,
+          file_type: doc.file_type,
+          preview: doc.file_url,
+          created: doc.created,
+          optional_metadata: doc.optional_metadata,
+        }));
+
+        setDocuments(transformedDocuments);
+        setFilteredDocuments(transformedDocuments);
+      } catch (error) {
+        enqueueSnackbar(t("documentsBar.failedFetch"), {
+          variant: "error",
+        });
+        console.error("Failed to fetch documents:", error);
+      }
+    };
+
+    fetchDocuments();
+  }, [selectedSessionId, enqueueSnackbar, t]);
+
+  useEffect(() => {
+    if (!searchQuery.trim()) {
+      setFilteredDocuments(documents);
+      return;
+    }
+
+    const filtered = documents.filter((doc) =>
+      doc.name.toLowerCase().includes(searchQuery.toLowerCase()),
+    );
+    setFilteredDocuments(filtered);
+  }, [searchQuery, documents]);
+
+  const handleFileUpload = async (files, url) => {
+    if (!files) return;
+    
+    const fileList = Array.isArray(files) ? files : [files];
+    
+    try {
+      for (const file of fileList) {
+        const docToAdd = {
+          file,
+          optional_metadata: {
+            name: file.name,
+            source: url || "local_upload",
+          },
+        };
+        
+        const savedDoc = await addDocument(docToAdd);
+        
+        // Add to local state immediately for UI feedback
+        const transformedDoc = {
+          id: savedDoc.id,
+          name: savedDoc.file_name,
+          type: savedDoc.file_type,
+          uploadedAt: savedDoc.created,
+          file_name: savedDoc.file_name,
+          file_type: savedDoc.file_type,
+          preview: savedDoc.file_url,
+          created: savedDoc.created,
+          optional_metadata: savedDoc.optional_metadata,
+        };
+        
+        setDocuments(prevDocs => [transformedDoc, ...prevDocs]);
+      }
+      
+      enqueueSnackbar(
+        t("documentsBar.successUpload", { count: fileList.length }),
+        { variant: "success" }
+      );
+      
+      if (onDocumentChange) {
+        onDocumentChange();
+      }
+    } catch (error) {
+      enqueueSnackbar(t("documentsBar.failedUpload"), {
+        variant: "error",
+      });
+      console.error("Failed to upload document:", error);
+    } finally {
+      setUploadOpen(false);
+    }
+  };
+
+  const handleDetailedView = () => {
+    navigate("/app/generative/RAG/documents");
+  };
+
+  return (
+    <Box
+      sx={{
+        display: "flex",
+        flexDirection: "column",
+        overflow: "hidden",
+        height: "100%",
+        width: "100%",
+        minWidth: 0, // Prevent flex shrinking issues
+        maxWidth: "100%", // Ensure consistent width
+      }}
+    >
+      <Box sx={{ p: 2, flexShrink: 0 }}>
+        <Box sx={{ display: 'flex', alignItems: 'center', justifyContent: 'space-between' }}>
+          <Typography variant="h6">{t("documentsBar.title")}</Typography>
+          <Tooltip title={t("documentsBar.detailedView")}>
+            <IconButton size="small" onClick={goToDocumentsDetail}>
+              <ViewListIcon />
+            </IconButton>
+          </Tooltip>
+        </Box>
+        <Typography variant="caption" sx={{ color: "text.secondary", mt: 1 }}>
+          {t("documentsBar.documentCount", { count: filteredDocuments.length })}
+          {selectedSessionId ? t("documentsBar.inCurrentSession") : t("documentsBar.available")}
+        </Typography>
+      </Box>
+      {/* Add documents button - only show when no session is selected */}
+      {!selectedSessionId && (
+        <Box sx={{ flexShrink: 0, px: 2, pb: 2 }}>
+          <Button
+            variant="contained"
+            fullWidth
+            color="primary"
+            startIcon={<AddIcon />}
+            onClick={() => setUploadOpen(true)}
+          >
+            {t("documentsBar.addDocuments")}
+          </Button>
+        </Box>
+      )}
+      {showSearch && documents.length >= 1 && (
+        <Box sx={{ p: 2, borderBottom: 1, borderColor: "divider", flexShrink: 0 }}>
+          <Box sx={{ display: "flex", alignItems: "center", gap: 1 }}>
+            <Box sx={{ flex: 1 }}>
+              <SearchBar
+                value={searchQuery}
+                onChange={(e) => setSearchQuery(e.target.value)}
+                onClear={() => setSearchQuery("")}
+                placeholder={t("documentsBar.searchPlaceholder")}
+              />
+            </Box>
+            {!selectedSessionId && (
+              <Tooltip title={t("documentsBar.detailedView")} placement="top">
+                <IconButton
+                  size="medium"
+                  onClick={handleDetailedView}
+                  sx={{ 
+                    color: "text.secondary",
+                    "&:hover": { color: "primary.main" }
+                  }}
+                >
+                  <ViewListIcon fontSize="small" />
+                </IconButton>
+              </Tooltip>
+            )}
+          </Box>
+        </Box>
+      )}
+
+      <Box
+        sx={{
+          flex: "0 1 45vh", // Take up to 40% of viewport height, but can shrink
+          overflowY: "auto",
+          overflowX: "hidden",
+          p: 2,
+          width: "100%",
+          minWidth: 0,
+          maxWidth: "100%",
+          minHeight: 0, // Allow shrinking when needed
+        }}
+      >
+        {filteredDocuments.length > 0 ? (
+          <DocumentList documents={filteredDocuments} />
+        ) : (
+          <Box
+            sx={{
+              display: "flex",
+              alignItems: "center",
+              justifyContent: "center",
+              height: "100%",
+              p: 2,
+            }}
+          >
+            <Typography
+              variant="body2"
+              sx={{ color: "text.secondary", textAlign: "center" }}
+            >
+              {searchQuery
+                ? t("documentsBar.noDocumentsFound")
+                : selectedSessionId
+                ? t("documentsBar.noDocumentsInSession")
+                : t("documentsBar.noDocumentsAvailable")}
+            </Typography>
+          </Box>
+        )}
+      </Box>
+
+      <Dialog
+        open={uploadOpen}
+        onClose={() => setUploadOpen(false)}
+        maxWidth="sm"
+        fullWidth
+        PaperProps={{
+          sx: {
+            maxHeight: "80vh",
+            minHeight: "300px",
+            display: "flex",
+            flexDirection: "column",
+          },
+        }}
+      >
+        <Box sx={{ flex: 1, overflow: "auto", display: "flex", flexDirection: "column" }}>
+          <Upload
+            onFileUpload={handleFileUpload}
+            multiple={true}
+            emptyUploadText={t("documentsBar.uploadDocuments")}
+            sx={{ flex: 1 }}
+          />
+        </Box>
+      </Dialog>
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/GeneratorParamsCard.jsx b/DashAI/front/src/components/generative/RAG/GeneratorParamsCard.jsx
new file mode 100644
index 000000000..3e2b406ae
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/GeneratorParamsCard.jsx
@@ -0,0 +1,111 @@
+import { useState, useMemo } from "react";
+import {
+  Box,
+  Typography,
+  Card,
+  CardContent,
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import IconButton from "@mui/material/IconButton";
+import Tooltip from "@mui/material/Tooltip";
+import InfoIcon from "@mui/icons-material/Info";
+import ExpandMoreIcon from "@mui/icons-material/ExpandMore";
+import GeneratorBody from "../../../pages/generative/RAGSession/components/GeneratorBody";
+import RAGSectionColumn from "../../../pages/generative/RAGSession/components/RAGSectionColumn";
+
+export default function GeneratorParamsCard({
+  generatorModel,
+  setGeneratorModel,
+  chunkSize = 0,
+  topK = 0,
+  promptTokenCount = 0,
+  setIsValid,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [showDescription, setShowDescription] = useState(false);
+  const [isExpanded, setIsExpanded] = useState(false);
+  const [initialModelParams, setInitialModelParams] = useState(null);
+
+  const isAdvanced = useMemo(() => {
+    if (!generatorModel?.component || !generatorModel?.params || !initialModelParams) return false;
+    return Object.keys(generatorModel.params).some(key => {
+      return generatorModel.params[key] !== initialModelParams[key];
+    });
+  }, [generatorModel?.params, initialModelParams]);
+
+  return (
+    <Card sx={{ width: "100%", backgroundColor: "background.paper" }}>
+      <CardContent sx={{ p: 2 }}>
+        <Box sx={{ mb: 2 }}>
+          <Box sx={{ display: "flex", alignItems: "center", justifyContent: "space-between" }}>
+            <Box>
+              <Typography variant="subtitle2">
+                {t("generative:rag.generator.modelLabel")}
+              </Typography>
+              {isExpanded && isAdvanced && (
+                <Typography variant="caption" sx={{ color: "warning.main" }}>
+                  {t("generative:rag.generator.advancedApplied")}
+                </Typography>
+              )}
+            </Box>
+            <Box>
+              {isExpanded && (
+                <Tooltip title={t("generative:rag.generator.description")}>
+                  <IconButton
+                    size="small"
+                    onClick={() => setShowDescription((s) => !s)}
+                    aria-label="generator-info"
+                    sx={{ color: "text.secondary" }}
+                  >
+                    <InfoIcon fontSize="small" />
+                  </IconButton>
+                </Tooltip>
+              )}
+              <Tooltip title={isExpanded ? t("generative:rag.prompt.collapse") : t("generative:rag.prompt.expand")}>
+                <IconButton
+                  size="small"
+                  onClick={() => setIsExpanded((s) => !s)}
+                  aria-label="toggle-generator-card"
+                >
+                  <ExpandMoreIcon
+                    fontSize="small"
+                    sx={{ transform: isExpanded ? "rotate(180deg)" : "rotate(0deg)", transition: "transform 0.2s" }}
+                  />
+                </IconButton>
+              </Tooltip>
+            </Box>
+          </Box>
+          {isExpanded && showDescription && (
+            <Typography variant="body2" color="textSecondary" sx={{ mt: 1 }}>
+              {t("generative:rag.generator.description")}
+            </Typography>
+          )}
+        </Box>
+
+        <RAGSectionColumn>
+          <GeneratorBody
+            generatorModel={generatorModel}
+            setGeneratorModel={setGeneratorModel}
+            chunkSize={chunkSize}
+            topK={topK}
+            promptTokenCount={promptTokenCount}
+            setIsValid={setIsValid}
+            isAdvanced={isAdvanced}
+            setInitialModelParams={setInitialModelParams}
+            showDetails={isExpanded}
+          />
+        </RAGSectionColumn>
+      </CardContent>
+    </Card>
+  );
+}
+
+GeneratorParamsCard.propTypes = {
+  generatorModel: PropTypes.object,
+  setGeneratorModel: PropTypes.func.isRequired,
+  chunkSize: PropTypes.number,
+  topK: PropTypes.number,
+  promptTokenCount: PropTypes.number,
+  setIsValid: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/components/generative/RAG/HighlightedTextarea.jsx b/DashAI/front/src/components/generative/RAG/HighlightedTextarea.jsx
new file mode 100644
index 000000000..8e031446f
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/HighlightedTextarea.jsx
@@ -0,0 +1,136 @@
+import { forwardRef, useRef, useCallback, useImperativeHandle } from "react";
+import { Box, Typography, useTheme } from "@mui/material";
+import { renderTemplateWithHighlights } from "../../../pages/generative/RAGSession/components/sectionUtils";
+
+/**
+ * A multiline text field that highlights placeholders ({chunks}, {input}, etc.)
+ * inside the editable area using an overlay technique.
+ *
+ * Both the highlighted backdrop and the native textarea share the same CSS Grid
+ * cell and identical styling (padding, font, line-height, border), so the
+ * highlights align perfectly with the text. The backdrop scrolls in sync.
+ *
+ * The ref forwards to the underlying <textarea> DOM element so parents can
+ * access selectionStart / selectionEnd for cursor-based operations.
+ */
+const HighlightedTextarea = forwardRef(function HighlightedTextarea(
+  { value, onChange, label, placeholder, minRows = 4, sx, ...textareaProps },
+  ref,
+) {
+  const theme = useTheme();
+  const backdropRef = useRef(null);
+  const inputRef = useRef(null);
+
+  useImperativeHandle(ref, () => inputRef.current, []);
+
+  const bgColor =
+    theme.palette.placeholder?.bg || theme.palette.warning.light;
+  const textColor =
+    theme.palette.placeholder?.text || theme.palette.warning.dark;
+
+  const handleScroll = useCallback(() => {
+    const input = inputRef.current;
+    const backdrop = backdropRef.current;
+    if (input && backdrop) {
+      backdrop.scrollTop = input.scrollTop;
+      backdrop.scrollLeft = input.scrollLeft;
+    }
+  }, []);
+
+  /** Shared styles applied to both backdrop and textarea for perfect alignment */
+  const sharedInputSx = {
+    fontFamily: theme.typography.code.fontFamily,
+    fontSize: theme.typography.body2.fontSize,
+    lineHeight: 1.6,
+    p: "16.5px 14px",
+    border: 1,
+    borderColor: "divider",
+    borderRadius: 1,
+    // Stack both elements in the same grid cell so they overlap exactly
+    gridArea: "1 / 1",
+  };
+
+  return (
+    <Box sx={{ mt: 2, width: "100%", ...sx }}>
+      {label && (
+        <Typography
+          variant="body2"
+          sx={{ mb: 0.5, color: "text.secondary" }}
+        >
+          {label}
+        </Typography>
+      )}
+
+      <Box
+        sx={{
+          display: "grid",
+          gridTemplateColumns: "1fr",
+          gridTemplateRows: "auto",
+          width: "100%",
+        }}
+      >
+        <Box
+          ref={backdropRef}
+          aria-hidden="true"
+          sx={{
+            ...sharedInputSx,
+            pointerEvents: "none",
+            whiteSpace: "pre-wrap",
+            wordBreak: "break-word",
+            overflow: "hidden",
+            zIndex: 0,
+            borderColor: "transparent",
+            color: "text.primary",
+          }}
+        >
+          {value
+            ? renderTemplateWithHighlights(value, {
+                bg: bgColor,
+                text: textColor,
+              }, theme.typography.code.fontFamily)
+            : placeholder || ""}
+        </Box>
+
+        <Box
+          component="textarea"
+          ref={inputRef}
+          value={value}
+          onChange={onChange}
+          onScroll={handleScroll}
+          placeholder={placeholder}
+          rows={minRows}
+          sx={{
+            ...sharedInputSx,
+            position: "relative",
+            zIndex: 1,
+            display: "block",
+            width: "100%",
+            color: "transparent",
+            caretColor: (t) => t.palette.text.primary,
+            backgroundColor: "transparent",
+            resize: "vertical",
+            outline: "none",
+            "&::placeholder": {
+              color: (t) => t.palette.text.secondary,
+              opacity: 1,
+            },
+            "&::selection": {
+              backgroundColor: "primary.light",
+            },
+            "&:focus": {
+              borderColor: "primary.main",
+              borderWidth: 2,
+              p: "15.5px 13px",
+            },
+            "&:hover": {
+              borderColor: (t) => t.palette.text.primary,
+            },
+          }}
+          {...textareaProps}
+        />
+      </Box>
+    </Box>
+  );
+});
+
+export default HighlightedTextarea;
diff --git a/DashAI/front/src/components/generative/RAG/PlaceholdersList.jsx b/DashAI/front/src/components/generative/RAG/PlaceholdersList.jsx
new file mode 100644
index 000000000..2695a047d
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/PlaceholdersList.jsx
@@ -0,0 +1,104 @@
+import React from "react";
+import { Box, IconButton, Tooltip, Typography, useTheme } from "@mui/material";
+import AddCircleOutlineIcon from "@mui/icons-material/AddCircleOutline";
+import CheckCircleIcon from "@mui/icons-material/CheckCircle";
+import WarningAmberIcon from "@mui/icons-material/WarningAmber";
+import HelpOutlineIcon from "@mui/icons-material/HelpOutline";
+import { useTranslation } from "react-i18next";
+
+/**
+ * Displays a list of required placeholders with icons and a "+" button
+ * to insert each placeholder into the prompt template textarea.
+ */
+export default function PlaceholdersList({
+  required = [],
+  descriptions = {},
+  template = "",
+  onInsertPlaceholder,
+}) {
+  const theme = useTheme();
+  const { t } = useTranslation(["generative"]);
+  return (
+    <Box sx={{ mt: 2, mb: 2 }}>
+      <Typography variant="subtitle1" gutterBottom>
+        {t("generative:rag.placeholders.title")}
+      </Typography>
+      <Box
+        component="ul"
+        sx={{
+          listStyle: "none",
+          m: 0,
+          p: 0,
+          display: "flex",
+          flexDirection: "column",
+          gap: 0.5,
+        }}
+      >
+        {required.map((ph) => {
+          const isPresent = template.includes(ph);
+          return (
+            <Box
+              component="li"
+              key={ph}
+              sx={{
+                display: "flex",
+                alignItems: "center",
+                gap: 0.5,
+              }}
+            >
+              {isPresent ? (
+                <CheckCircleIcon
+                  fontSize="small"
+                  color="success"
+                  sx={{ flexShrink: 0 }}
+                />
+              ) : (
+                <WarningAmberIcon
+                  fontSize="small"
+                  color="warning"
+                  sx={{ flexShrink: 0 }}
+                />
+              )}
+
+              {onInsertPlaceholder && (
+                <Tooltip title={t("generative:rag.placeholders.insert", { placeholder: ph })} placement="top">
+                  <IconButton
+                    size="small"
+                    onClick={() => onInsertPlaceholder(ph)}
+                    sx={{
+                      p: 0.25,
+                      color: "primary.main",
+                      "&:hover": { backgroundColor: "primary.light" },
+                    }}
+                  >
+                    <AddCircleOutlineIcon fontSize="small" />
+                  </IconButton>
+                </Tooltip>
+              )}
+
+              <Typography
+                component="code"
+                variant="body2"
+                sx={{
+                  fontFamily: theme.typography.code.fontFamily,
+                }}
+              >
+                {ph}
+              </Typography>
+
+              {descriptions[ph] && (
+                <Tooltip title={descriptions[ph]} placement="right">
+                  <HelpOutlineIcon
+                    fontSize="small"
+                    color="action"
+                    sx={{ cursor: "pointer", flexShrink: 0 }}
+                  />
+                </Tooltip>
+              )}
+            </Box>
+          );
+        })}
+      </Box>
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/PromptParamsCard.jsx b/DashAI/front/src/components/generative/RAG/PromptParamsCard.jsx
new file mode 100644
index 000000000..2b0ac8bf4
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/PromptParamsCard.jsx
@@ -0,0 +1,456 @@
+import { useState, useEffect, useCallback, useMemo, useRef } from "react";
+import {
+  Box,
+  Typography,
+  Card,
+  CardContent,
+  Autocomplete,
+  TextField,
+  Button,
+  MenuItem,
+  useTheme,
+} from "@mui/material";
+import { useNavigate } from "react-router-dom";
+import {
+  ViewList as ViewListIcon,
+  Info as InfoIcon,
+  ExpandMore as ExpandMoreIcon,
+  AddCircleOutline as AddIcon,
+} from "@mui/icons-material";
+import IconButton from "@mui/material/IconButton";
+import Tooltip from "@mui/material/Tooltip";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import {
+  getRAGPrompts,
+  getDefaultPrompts,
+  isGenerationPromptClass,
+} from "../../../api/rag";
+import NewPromptModal from "../../../pages/generative/RAGSession/advanced/NewPromptModal";
+import RAGSectionColumn from "../../../pages/generative/RAGSession/components/RAGSectionColumn";
+import { getDescription, renderTemplateWithHighlights } from "../../../pages/generative/RAGSession/components/sectionUtils";
+
+import { LANGUAGE_CODES } from "../../../constants/languages";
+
+const CREATE_NEW_ID = "__create-new__";
+const DEFAULT_IDS = {
+  DefaultRAGGenerationPrompt: "default-generation",
+  DefaultQARAGGenerationPrompt: "default-QA",
+};
+
+function getDefaultDisplayName(option, t) {
+  // Use class_name (set explicitly by our code) rather than name
+  // (raw API field) to avoid potential encoding / serialization mismatches.
+  const cname = option.class_name || option.name || "";
+  if (cname.includes("DefaultQARAGGenerationPrompt")) {
+    return t("generative:rag.prompt.defaultQAGenerationPrompt");
+  }
+  if (cname.includes("DefaultRAGGenerationPrompt")) {
+    return t("generative:rag.prompt.defaultGenerationPrompt");
+  }
+  return option.name || cname;
+}
+
+function getOptionLabel(option, t) {
+  if (option._isCreateNew) return option.name;
+  if (option._isDefault) return getDefaultDisplayName(option, t);
+  return option.name;
+}
+
+export default function PromptParamsCard({
+  promptModel,
+  setPromptModel,
+  onTokenCountChange,
+}) {
+  const navigate = useNavigate();
+  const goToPromptsDetail = () => navigate("/app/generative/RAG/prompts");
+  const { t, i18n } = useTranslation(["generative"]);
+  const theme = useTheme();
+  const placeholderColors = useMemo(
+    () => ({
+      bg: theme.palette.placeholder?.bg || theme.palette.warning.light,
+      text: theme.palette.placeholder?.text || theme.palette.warning.dark,
+    }),
+    [theme],
+  );
+  const [showDescription, setShowDescription] = useState(false);
+  const [isExpanded, setIsExpanded] = useState(false);
+
+  const platformLang = useMemo(() => {
+    const lang = (i18n.language || "en").split("-")[0];
+    return ["en", "es", "pt"].includes(lang) ? lang : "en";
+  }, [i18n.language]);
+
+  const [customPrompts, setCustomPrompts] = useState([]);
+  const [defaultPrompts, setDefaultPrompts] = useState([]);
+  const [selectedPrompt, setSelectedPrompt] = useState(null);
+  const [newPromptModalOpen, setNewPromptModalOpen] = useState(false);
+  const [loading, setLoading] = useState(true);
+  const [selectedLanguage, setSelectedLanguage] = useState(platformLang);
+  const prevSelectedRef = useRef(null);
+  const isInitializedRef = useRef(false);
+
+  const loadPrompts = useCallback(async () => {
+    try {
+      const dbPrompts = await getRAGPrompts();
+      // Keep only user-created generation prompts:
+      // 1. Exclude system defaults (class_name starts with "Default")
+      // 2. Exclude augmentation prompts (wrong type for this selector)
+      setCustomPrompts(
+        (dbPrompts || []).filter(
+          (p) =>
+            !p.class_name.startsWith("Default") &&
+            isGenerationPromptClass(p.class_name),
+        ),
+      );
+    } catch (error) {
+      console.error("Error loading custom RAG prompts:", error);
+      setCustomPrompts([]);
+    }
+
+    try {
+      const defaultData = await getDefaultPrompts();
+      // Filter out CustomRAGGenerationPrompt — it's a base class for
+      // user-created prompts, not a selectable template. It has no
+      // templates in its metadata and inherits a generic description
+      // ("Base class for RAG prompts.") from Prompt.
+      setDefaultPrompts(
+        (defaultData || []).filter(
+          (dp) => dp.name !== "CustomRAGGenerationPrompt",
+        ),
+      );
+    } catch (error) {
+      console.error("Error loading default RAG prompts:", error);
+      setDefaultPrompts([]);
+    }
+  }, []);
+
+  useEffect(() => {
+    const load = async () => {
+      await loadPrompts();
+      setLoading(false);
+    };
+    load();
+  }, [loadPrompts]);
+
+  const mergedOptions = useMemo(() => {
+    const defaults = defaultPrompts.map((dp) => ({
+      ...dp,
+      id: DEFAULT_IDS[dp.name] || dp.name,
+      _isDefault: true,
+      class_name: dp.name,
+    }));
+    const customs = customPrompts.map((cp) => ({
+      ...cp,
+      _isDefault: false,
+    }));
+    return [
+      ...defaults,
+      ...customs,
+      {
+        id: CREATE_NEW_ID,
+        name: t("generative:rag.prompt.createNewPrompt"),
+        _isCreateNew: true,
+        _isDefault: false,
+        class_name: "",
+      },
+    ];
+  }, [defaultPrompts, customPrompts, t]);
+
+  const currentTemplate = useMemo(() => {
+    if (!selectedPrompt) return "";
+    if (selectedPrompt._isDefault) {
+      return selectedPrompt.metadata?.templates?.[selectedLanguage] || "";
+    }
+    if (selectedPrompt.parameters?.templates) {
+      return selectedPrompt.parameters.templates[selectedLanguage] || "";
+    }
+    return selectedPrompt.parameters?.template || "";
+  }, [selectedPrompt, selectedLanguage]);
+
+  const isDefault = selectedPrompt?._isDefault;
+
+  useEffect(() => {
+    if (!selectedPrompt) {
+      if (onTokenCountChange) onTokenCountChange(0);
+      return;
+    }
+
+    if (!isInitializedRef.current && promptModel?.component === (selectedPrompt.class_name || selectedPrompt.name)) {
+      isInitializedRef.current = true;
+      return;
+    }
+    isInitializedRef.current = true;
+
+    setPromptModel({
+      component: selectedPrompt.class_name || selectedPrompt.name,
+      params: {
+        template: currentTemplate,
+        language: selectedLanguage,
+        ...(selectedPrompt._isDefault || selectedPrompt.parameters?.templates
+          ? { templates: selectedPrompt.parameters?.templates }
+          : {}),
+      },
+    });
+    if (onTokenCountChange) {
+      const tokenCount = Math.ceil(currentTemplate.length / 4);
+      onTokenCountChange(tokenCount);
+    }
+  }, [selectedPrompt, selectedLanguage]);
+
+  useEffect(() => {
+    const selectable = mergedOptions.filter((o) => !o._isCreateNew);
+    if (!selectable.length) return;
+
+    if (promptModel?.component) {
+      const found = selectable.find((p) => {
+        if (p._isDefault) {
+          return p.class_name === promptModel.component;
+        }
+        return (
+          p.class_name === promptModel.component &&
+          p.parameters?.template === promptModel.params?.template
+        );
+      });
+      if (found?.id !== selectedPrompt?.id) {
+        setSelectedPrompt(found || null);
+        prevSelectedRef.current = found || null;
+        if (found?._isDefault) {
+          setSelectedLanguage(promptModel.params?.language || platformLang);
+        }
+        isInitializedRef.current = false;
+      }
+      return;
+    }
+
+    if (!selectedPrompt) {
+      const firstDefault =
+        selectable.find((p) => p._isDefault) || selectable[0];
+      if (firstDefault) {
+        setSelectedPrompt(firstDefault);
+        setSelectedLanguage(platformLang);
+        prevSelectedRef.current = firstDefault;
+      }
+    }
+  }, [mergedOptions, promptModel]);
+
+  const handlePromptChange = (_event, newValue) => {
+    if (newValue?._isCreateNew) {
+      setNewPromptModalOpen(true);
+      setSelectedPrompt(prevSelectedRef.current);
+      return;
+    }
+    prevSelectedRef.current = newValue;
+    setSelectedPrompt(newValue);
+    if (newValue?._isDefault) {
+      setSelectedLanguage(platformLang);
+    }
+  };
+
+  const handleLanguageChange = (event) => {
+    setSelectedLanguage(event.target.value);
+  };
+
+  const handlePromptCreated = useCallback(
+    async (newPromptId) => {
+      const updatedPrompts = await getRAGPrompts();
+      setCustomPrompts(
+        (updatedPrompts || []).filter(
+          (p) =>
+            !p.class_name.startsWith("Default") &&
+            isGenerationPromptClass(p.class_name),
+        ),
+      );
+
+      const newPrompt = (updatedPrompts || []).find(
+        (p) => p.id === newPromptId,
+      );
+      if (newPrompt) {
+        const wrapped = { ...newPrompt, _isDefault: false };
+        setSelectedPrompt(wrapped);
+        prevSelectedRef.current = wrapped;
+        // Sync language from the newly created prompt's saved language
+        // so local selectedLanguage stays aligned with the persisted value.
+        if (newPrompt.parameters?.language) {
+          setSelectedLanguage(newPrompt.parameters.language);
+        }
+        // Sync parent promptModel immediately to prevent infinite
+        // useEffect 1 <-> useEffect 2 ping-pong when the new prompt's
+        // class_name differs from the previously selected prompt.
+        setPromptModel({
+          component: newPrompt.class_name,
+          params: {
+            template: newPrompt.parameters?.template || "",
+            language: newPrompt.parameters?.language || "",
+          },
+        });
+        const template = newPrompt.parameters?.template || "";
+        const tokenCount = Math.ceil(template.length / 4);
+        onTokenCountChange?.(tokenCount);
+      }
+
+      setNewPromptModalOpen(false);
+    },
+    [onTokenCountChange, setPromptModel],
+  );
+
+  if (loading) {
+    return null;
+  }
+
+  return (
+    <Card sx={{ backgroundColor: "background.paper" }}>
+      <CardContent sx={{ p: 2 }}>
+        <Box sx={{ mb: 2 }}>
+          <Box sx={{ display: "flex", alignItems: "center", justifyContent: "space-between" }}>
+            <Typography variant="subtitle2">{t("generative:rag.prompt.promptLabel")}</Typography>
+            <Box>
+              {isExpanded && (
+                <Tooltip title={t("generative:rag.prompt.descriptionToggle")}>
+                  <IconButton
+                    size="small"
+                    onClick={() => setShowDescription((s) => !s)}
+                    aria-label="prompt-info"
+                    sx={{ color: "text.secondary" }}
+                  >
+                    <InfoIcon fontSize="small" />
+                  </IconButton>
+                </Tooltip>
+              )}
+              <Tooltip title={t("generative:rag.prompt.openPrompts")}>
+                <IconButton size="small" onClick={goToPromptsDetail} aria-label="open-prompt-library">
+                  <ViewListIcon fontSize="small" />
+                </IconButton>
+              </Tooltip>
+              <Tooltip title={isExpanded ? t("generative:rag.prompt.collapse") : t("generative:rag.prompt.expand")}>
+                <IconButton
+                  size="small"
+                  onClick={() => setIsExpanded((s) => !s)}
+                  aria-label="toggle-prompt-card"
+                >
+                  <ExpandMoreIcon
+                    fontSize="small"
+                    sx={{ transform: isExpanded ? "rotate(180deg)" : "rotate(0deg)", transition: "transform 0.2s" }}
+                  />
+                </IconButton>
+              </Tooltip>
+            </Box>
+          </Box>
+          {isExpanded && showDescription && (
+            <Typography variant="body2" color="textSecondary" sx={{ mt: 1 }}>
+              {t("generative:rag.prompt.description")}
+            </Typography>
+          )}
+        </Box>
+
+        <RAGSectionColumn>
+          <Box>
+            <Autocomplete
+              options={mergedOptions}
+              value={selectedPrompt}
+              onChange={handlePromptChange}
+              getOptionLabel={(option) => getOptionLabel(option, t)}
+              isOptionEqualToValue={(option, value) => option.id === value.id}
+              renderOption={(props, option) => (
+                <li {...props}>{getOptionLabel(option, t)}</li>
+              )}
+              renderInput={(params) => (
+                <TextField
+                  {...params}
+                  label={t("generative:rag.prompt.selectTemplate")}
+                  placeholder={t(
+                    "generative:rag.prompt.selectTemplatePlaceholder",
+                  )}
+                />
+              )}
+              sx={{}}
+            />
+          </Box>
+
+          {isExpanded && (isDefault || selectedPrompt?.parameters?.templates) && (
+            <TextField
+              select
+              fullWidth
+              label={t("generative:rag.prompt.language")}
+              value={selectedLanguage}
+              onChange={handleLanguageChange}
+              size="small"
+            >
+              {LANGUAGE_CODES.map((code) => (
+                <MenuItem key={code} value={code}>
+                  {t(`generative:rag.prompt.languages.${code}`)}
+                </MenuItem>
+              ))}
+            </TextField>
+          )}
+
+          {isExpanded && selectedPrompt && (
+            <Box>
+              <Typography variant="subtitle2" sx={{ mb: 2 }}>
+                {t("generative:rag.prompt.selectedTemplate")}
+              </Typography>
+
+              <Box
+                sx={{
+                  p: 2,
+                  backgroundColor: "background.paper",
+                  border: "1px solid",
+                  borderColor: "divider",
+                  borderRadius: 1,
+              fontFamily: theme.typography.code.fontFamily,
+                  lineHeight: 1.6,
+                  whiteSpace: "pre-wrap",
+                  wordBreak: "break-word",
+                }}
+              >
+                {renderTemplateWithHighlights(currentTemplate, placeholderColors, theme.typography.code.fontFamily)}
+              </Box>
+
+              {getDescription(selectedPrompt.description, i18n) && (
+                <Typography
+                  variant="caption"
+                  display="block"
+                  sx={{ color: "text.secondary" }}
+                >
+                  {getDescription(selectedPrompt.description, i18n)}
+                </Typography>
+              )}
+            </Box>
+          )}
+
+          {isExpanded && (
+            <Button
+              variant="contained"
+              fullWidth
+              color="primary"
+              size="large"
+              startIcon={<AddIcon />}
+              onClick={() => setNewPromptModalOpen(true)}
+            >
+              {t("generative:rag.prompt.newPromptButton")}
+            </Button>
+          )}
+        </RAGSectionColumn>
+      </CardContent>
+
+      <NewPromptModal
+        open={newPromptModalOpen}
+        handleClose={() => setNewPromptModalOpen(false)}
+        onPromptCreated={handlePromptCreated}
+        existingPrompts={customPrompts}
+      />
+    </Card>
+  );
+}
+
+PromptParamsCard.propTypes = {
+  promptModel: PropTypes.shape({
+    component: PropTypes.string,
+    params: PropTypes.shape({
+      template: PropTypes.string,
+      language: PropTypes.string,
+    }),
+  }),
+  setPromptModel: PropTypes.func.isRequired,
+  onTokenCountChange: PropTypes.func,
+};
diff --git a/DashAI/front/src/components/generative/RAG/PromptSelectionTable.jsx b/DashAI/front/src/components/generative/RAG/PromptSelectionTable.jsx
new file mode 100644
index 000000000..637b0ed4c
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/PromptSelectionTable.jsx
@@ -0,0 +1,268 @@
+import React, { useState } from "react";
+import PropTypes from "prop-types";
+import {
+  Box,
+  Paper,
+  Tooltip,
+  IconButton,
+  Button,
+  Grid,
+  Typography,
+} from "@mui/material";
+import { AddCircleOutline as AddIcon } from "@mui/icons-material";
+import VisibilityIcon from "@mui/icons-material/Visibility";
+import { DataGrid } from "@mui/x-data-grid";
+import { useTranslation } from "react-i18next";
+import { formatDate } from "../../../utils";
+import PromptViewModal from "./PromptViewModal";
+import NewPromptModal from "../../../pages/generative/RAGSession/advanced/NewPromptModal";
+import { getRAGPrompts } from "../../../api/rag";
+
+/**
+ * Expand default prompts (with `templates` dict) into one row per language.
+ * Custom prompts (single `template`) remain as a single row.
+ *
+ * Each expanded row includes a `_parentPrompt` reference to the original
+ * prompt object, used by the view modal to show all language versions.
+ *
+ * @param {Array} prompts - Raw prompt objects from the API.
+ * @returns {Array} Flattened rows ready for the DataGrid.
+ */
+function expandPromptRows(prompts) {
+  const rows = [];
+  for (const prompt of prompts) {
+    const templates = prompt.parameters?.templates;
+    if (templates && Object.keys(templates).length > 0) {
+      for (const lang of Object.keys(templates)) {
+        rows.push({
+          ...prompt,
+          id: `${prompt.id}-${lang}`,
+          _originalId: prompt.id,
+          language: lang,
+          _parentPrompt: prompt,
+        });
+      }
+    } else {
+      rows.push({
+        ...prompt,
+        language: prompt.parameters?.language || null,
+        _originalId: prompt.id,
+        _parentPrompt: prompt,
+      });
+    }
+  }
+  return rows;
+}
+
+export default function PromptSelectionTable({
+  prompts = [],
+  loading = false,
+  rowSelectionModel = [],
+  onRowSelectionModelChange,
+  showTableTitle = false,
+  setSessionData,
+}) {
+  const [modalOpen, setModalOpen] = useState(false);
+  const [selectedPrompt, setSelectedPrompt] = useState(null);
+  const [newPromptModalOpen, setNewPromptModalOpen] = useState(false);
+  const [promptRows, setPromptRows] = useState([]);
+  const [rawPrompts, setRawPrompts] = useState([]);
+  const { t } = useTranslation(["generative"]);
+
+  React.useEffect(() => {
+    async function fetchPrompts() {
+      const initialPrompts = await getRAGPrompts();
+      initialPrompts.sort((a, b) => new Date(b.created) - new Date(a.created));
+      setRawPrompts(initialPrompts);
+      setPromptRows(expandPromptRows(initialPrompts));
+    }
+    fetchPrompts();
+  }, []);
+
+  const handleViewPrompt = (row) => {
+    // Always open the full parent prompt so the modal shows all languages
+    setSelectedPrompt(row._parentPrompt || row);
+    setModalOpen(true);
+  };
+
+  const handleCloseModal = () => {
+    setModalOpen(false);
+    setSelectedPrompt(null);
+  };
+
+  const columns = React.useMemo(
+    () => [
+      {
+        field: "id",
+        headerName: t("generative:rag.promptView.table.id"),
+        minWidth: 50,
+        flex: 0.3,
+        editable: false,
+      },
+      {
+        field: "name",
+        headerName: t("generative:rag.promptView.table.name"),
+        minWidth: 140,
+        flex: 1,
+        editable: false,
+      },
+      {
+        field: "class_name",
+        headerName: t("generative:rag.promptView.table.type"),
+        minWidth: 140,
+        flex: 1,
+        editable: false,
+      },
+      {
+        field: "language",
+        headerName: t("generative:rag.promptView.table.language"),
+        minWidth: 100,
+        flex: 0.7,
+        editable: false,
+        valueGetter: (value, row) => {
+          if (row.language) {
+            return t(`generative:rag.prompt.languages.${row.language}`) || row.language;
+          }
+          return "-";
+        },
+      },
+      {
+        field: "created",
+        headerName: t("generative:rag.promptView.table.created"),
+        minWidth: 140,
+        flex: 1,
+        editable: false,
+        valueGetter: (value) => formatDate(value),
+      },
+      {
+        field: "last_modified",
+        headerName: t("generative:rag.promptView.table.edited"),
+        minWidth: 140,
+        flex: 1,
+        editable: false,
+        valueGetter: (value) => formatDate(value),
+      },
+      {
+        field: "actions",
+        type: "actions",
+        headerName: t("generative:rag.promptView.table.actions"),
+        minWidth: 80,
+        flex: 0.4,
+        getActions: (params) => [
+          <Tooltip title={t("generative:rag.promptView.table.viewPrompt")} key="preview">
+            <IconButton
+              size="small"
+              onClick={() => handleViewPrompt(params.row)}
+            >
+              <VisibilityIcon fontSize="small" />
+            </IconButton>
+          </Tooltip>,
+        ],
+      },
+    ],
+    [t],
+  );
+
+  const handlePromptCreated = async (newPromptId) => {
+    const updatedPrompts = await getRAGPrompts();
+    updatedPrompts.sort((a, b) => new Date(b.created) - new Date(a.created));
+    setRawPrompts(updatedPrompts);
+    setPromptRows(expandPromptRows(updatedPrompts));
+    if (onRowSelectionModelChange && newPromptId) {
+      onRowSelectionModelChange([newPromptId]);
+    }
+    if (setSessionData && newPromptId) {
+      setSessionData((prev) => ({
+        ...prev,
+        parameters: {
+          ...prev.parameters,
+          prompt_id: newPromptId,
+        },
+      }));
+    }
+    setNewPromptModalOpen(false);
+  };
+
+  return (
+    <Paper sx={{ py: 4, px: 4 }}>
+      {showTableTitle && (
+        <Grid
+          container
+          justifyContent="space-between"
+          alignItems="center"
+          sx={{ mb: 4 }}
+        >
+          <Typography variant="h5" component="h2">
+            {t("generative:rag.promptView.table.currentPrompts")}
+          </Typography>
+          <Button
+            variant="contained"
+            color="primary"
+            onClick={() => setNewPromptModalOpen(true)}
+            startIcon={<AddIcon />}
+          >
+            {t("generative:rag.promptView.table.newPrompt")}
+          </Button>
+        </Grid>
+      )}
+      {!showTableTitle && (
+        <Grid
+          container
+          justifyContent="space-between"
+          alignItems="center"
+          sx={{ mb: 4 }}
+        >
+          <Typography variant="subtitle1" component="p" sx={{ mb: 2 }}>
+            {t("generative:rag.promptView.table.choosePrompt")}
+          </Typography>
+          <Button
+            variant="contained"
+            color="primary"
+            onClick={() => setNewPromptModalOpen(true)}
+            startIcon={<AddIcon />}
+          >
+            {t("generative:rag.promptView.table.newPrompt")}
+          </Button>
+        </Grid>
+      )}
+      <Box sx={{ height: "100%" }}>
+        <DataGrid
+          rows={promptRows}
+          columns={columns}
+          hideFooter
+          initialState={{
+            pagination: {
+              paginationModel: {
+                pageSize: 5,
+              },
+            },
+          }}
+          pageSizeOptions={[5, 10, 25, 50]}
+          autoHeight
+          loading={loading}
+        />
+        <PromptViewModal
+          key={selectedPrompt?.id ?? "no-prompt"}
+          open={modalOpen}
+          handleClose={handleCloseModal}
+          prompt={selectedPrompt}
+        />
+        <NewPromptModal
+          open={newPromptModalOpen}
+          handleClose={() => setNewPromptModalOpen(false)}
+          onPromptCreated={handlePromptCreated}
+          existingPrompts={rawPrompts}
+        />
+      </Box>
+    </Paper>
+  );
+}
+
+PromptSelectionTable.propTypes = {
+  prompts: PropTypes.array,
+  loading: PropTypes.bool,
+  rowSelectionModel: PropTypes.array,
+  onRowSelectionModelChange: PropTypes.func,
+  showTableTitle: PropTypes.bool,
+  setSessionData: PropTypes.func,
+};
diff --git a/DashAI/front/src/components/generative/RAG/PromptViewModal.jsx b/DashAI/front/src/components/generative/RAG/PromptViewModal.jsx
new file mode 100644
index 000000000..cd025e7f9
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/PromptViewModal.jsx
@@ -0,0 +1,190 @@
+import { useState, useMemo } from "react";
+import PropTypes from "prop-types";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  TextField,
+  MenuItem,
+  Typography,
+  Box,
+  useTheme,
+} from "@mui/material";
+import { useTranslation } from "react-i18next";
+import { renderTemplateWithHighlights } from "../../../pages/generative/RAGSession/components/sectionUtils";
+import { LANGUAGE_CODES } from "../../../constants/languages";
+
+/**
+ * Dialog that displays prompt content with optional language selection.
+ *
+ * Supports two prompt shapes:
+ *  - Single-template: content in `prompt.parameters.template` (string)
+ *  - Multi-template: content in `prompt.parameters.templates` (dict of `{ [lang]: string }`)
+ *
+ * For multi-template prompts, a language selector is enabled so the user can
+ * switch between available language versions. For single-template prompts,
+ * the selector is disabled and shows either the stored language or
+ * "Language not available".
+ *
+ * The parent must pass a unique `key` prop (e.g. `prompt.id`) to ensure state
+ * resets when a different prompt is displayed (component is always mounted).
+ *
+ * @param {object}  props
+ * @param {boolean} props.open                         - Whether the dialog is visible
+ * @param {function} props.handleClose                  - Callback when the dialog is closed
+ * @param {object}  [props.prompt]                      - The prompt object to display
+ * @param {string}  props.prompt.name                   - Display name (shown in title bar)
+ * @param {string}  props.prompt.class_name             - Component type (shown as "Type")
+ * @param {object}  props.prompt.parameters             - Parameters bag
+ * @param {string}  [props.prompt.parameters.template]  - Single template string
+ * @param {object}  [props.prompt.parameters.templates] - Multi-language template dict
+ * @param {string}  [props.prompt.parameters.language]  - Default language code
+ */
+export default function PromptViewModal({ open, handleClose, prompt }) {
+  const { t } = useTranslation(["generative"]);
+  const theme = useTheme();
+
+  const placeholderColors = useMemo(
+    () => ({
+      bg: theme.palette.placeholder?.bg || theme.palette.warning.light,
+      text: theme.palette.placeholder?.text || theme.palette.warning.dark,
+    }),
+    [theme],
+  );
+
+  const hasMultiTemplates = useMemo(
+    () =>
+      !!prompt?.parameters?.templates &&
+      Object.keys(prompt.parameters.templates).length > 0,
+    [prompt],
+  );
+
+  const [selectedLanguage, setSelectedLanguage] = useState(() => {
+    if (prompt?.parameters?.templates) {
+      return (
+        prompt?.parameters?.language ||
+        Object.keys(prompt.parameters.templates)[0] ||
+        ""
+      );
+    }
+    return "";
+  });
+
+  /** Language options derived from available template keys for multi-template prompts. */
+  const languageOptions = useMemo(() => {
+    if (!hasMultiTemplates || !prompt?.parameters?.templates) return [];
+    const codes = Object.keys(prompt.parameters.templates);
+    return codes.map((code) => ({
+      code,
+      name: t(`generative:rag.prompt.languages.${code}`) || code,
+    }));
+  }, [hasMultiTemplates, prompt, t]);
+
+  const currentTemplate = useMemo(() => {
+    if (hasMultiTemplates) {
+      return prompt?.parameters?.templates?.[selectedLanguage] || "";
+    }
+    return prompt?.parameters?.template || "";
+  }, [hasMultiTemplates, prompt, selectedLanguage]);
+
+  return (
+    <Dialog open={open} onClose={handleClose} maxWidth="md" fullWidth>
+      <DialogTitle>{prompt?.name || t("generative:rag.promptView.untitledPrompt")}</DialogTitle>
+      <DialogContent>
+        <TextField
+          label={t("generative:rag.promptView.type")}
+          value={prompt?.class_name || ""}
+          disabled
+          fullWidth
+          size="small"
+          sx={{ mt: 1, mb: 2 }}
+        />
+
+        {hasMultiTemplates ? (
+          <TextField
+            select
+            label={t("generative:rag.promptView.language")}
+            value={selectedLanguage}
+            onChange={(e) => setSelectedLanguage(e.target.value)}
+            fullWidth
+            size="small"
+            sx={{ mb: 2 }}
+          >
+            {languageOptions.map((opt) => (
+              <MenuItem key={opt.code} value={opt.code}>
+                {opt.name}
+              </MenuItem>
+            ))}
+          </TextField>
+        ) : (
+          <TextField
+            label={t("generative:rag.promptView.language")}
+            value={
+              prompt?.parameters?.language
+                ? t(`generative:rag.prompt.languages.${prompt.parameters.language}`) ||
+                  prompt.parameters.language
+                : t("generative:rag.promptView.languageNotAvailable")
+            }
+            disabled
+            fullWidth
+            size="small"
+            sx={{ mb: 2 }}
+          />
+        )}
+
+        <Typography variant="subtitle2" gutterBottom>
+          {t("generative:rag.promptView.templateContent")}
+        </Typography>
+        <Box
+          sx={{
+            p: 2,
+            backgroundColor: theme.palette.ui.box,
+            border: "1px solid",
+            borderColor: theme.palette.ui.border,
+            borderRadius: 1,
+            fontFamily: theme.typography.code.fontFamily,
+            lineHeight: 1.6,
+            whiteSpace: "pre-wrap",
+            wordBreak: "break-word",
+            maxHeight: "400px",
+            overflowY: "auto",
+          }}
+        >
+          {renderTemplateWithHighlights(currentTemplate, placeholderColors, theme.typography.code.fontFamily) ?? (
+            <Typography
+              variant="body2"
+              color="text.secondary"
+              fontStyle="italic"
+            >
+              {hasMultiTemplates
+                ? t("generative:rag.promptView.languageNotAvailable")
+                : t("generative:rag.promptView.noContent")}
+            </Typography>
+          )}
+        </Box>
+      </DialogContent>
+      <DialogActions>
+        <Button onClick={handleClose}>
+          {t("generative:rag.promptView.close")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
+PromptViewModal.propTypes = {
+  open: PropTypes.bool.isRequired,
+  handleClose: PropTypes.func.isRequired,
+  prompt: PropTypes.shape({
+    id: PropTypes.number,
+    name: PropTypes.string,
+    class_name: PropTypes.string,
+    parameters: PropTypes.shape({
+      template: PropTypes.string,
+      templates: PropTypes.objectOf(PropTypes.string),
+      language: PropTypes.string,
+    }),
+  }),
+};
diff --git a/DashAI/front/src/components/generative/RAG/RAGBreadcrumbs.jsx b/DashAI/front/src/components/generative/RAG/RAGBreadcrumbs.jsx
new file mode 100644
index 000000000..271cbd328
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/RAGBreadcrumbs.jsx
@@ -0,0 +1,151 @@
+import React from "react";
+import Breadcrumbs from "@mui/material/Breadcrumbs";
+import Typography from "@mui/material/Typography";
+import Link from "@mui/material/Link";
+import IconButton from "@mui/material/IconButton";
+import ArrowBackIcon from "@mui/icons-material/ArrowBack";
+import Box from "@mui/material/Box";
+import { useNavigate, useLocation } from "react-router-dom";
+import { useTranslation } from "react-i18next";
+import { useGenerative } from "../GenerativeContext";
+
+/**
+ * Breadcrumbs component for RAG navigation
+ * @param {string} sessionName - Optional session name to show in breadcrumbs
+ */
+function RAGBreadcrumbs({ sessionName }) {
+  const { t } = useTranslation(["generative"]);
+  const navigate = useNavigate();
+  const location = useLocation();
+  const {
+    setSelectedSessionId,
+    setSelectedTaskName,
+    setSelectedDisplayName,
+    setStepIndex,
+  } = useGenerative() ?? {};
+
+  const navigateToGenerative = () => {
+    setSelectedSessionId?.(null);
+    setSelectedTaskName?.(null);
+    setSelectedDisplayName?.(null);
+    setStepIndex?.(0);
+    navigate("/app/generative");
+  };
+
+  const getBreadcrumbs = () => {
+    const path = location.pathname;
+    if (!path.startsWith("/app/generative/RAG")) return [];
+
+    const base = [
+      { label: t("generative:rag.breadcrumbs.generative"), path: "/app/generative" },
+      { label: t("generative:rag.breadcrumbs.rag"), path: "/app/generative/RAG" },
+    ];
+
+    if (path === "/app/generative/RAG/documents") return [...base, { label: t("generative:rag.breadcrumbs.documents"), path: null, current: true }];
+    if (path === "/app/generative/RAG/prompts") return [...base, { label: t("generative:rag.breadcrumbs.prompts"), path: null, current: true }];
+
+    if (sessionName) return [...base, { label: sessionName, path: null, current: true, isSession: true }];
+
+    base[1] = { ...base[1], path: null, current: true };
+    return base;
+  };
+
+  const breadcrumbs = getBreadcrumbs();
+
+  const handleNavigate = (path) => {
+    if (!path) return;
+    if (path === "/app/generative") {
+      navigateToGenerative();
+      return;
+    }
+    if (path === "/app/generative/RAG") {
+      setSelectedSessionId?.(null);
+      setSelectedTaskName?.("RAGTask");
+      setSelectedDisplayName?.(null);
+      setStepIndex?.(0);
+    }
+    navigate(path);
+  };
+
+  const handleBack = () => {
+    if (breadcrumbs.length > 1) {
+      const parentBreadcrumb = breadcrumbs[breadcrumbs.length - 2];
+      handleNavigate(parentBreadcrumb.path);
+      return;
+    }
+    navigateToGenerative();
+  };
+
+  const showBackButton = breadcrumbs.length > 0;
+
+  return (
+    <Box 
+      sx={{ 
+        mb: 2,
+        minHeight: '24px', // Ensure consistent height
+        display: 'flex',
+        alignItems: 'center',
+        gap: 1
+      }}
+    >
+      {showBackButton && (
+        <IconButton
+          onClick={handleBack}
+          size="small"
+          sx={{
+            color: 'text.secondary',
+            '&:hover': {
+              color: 'primary.main',
+              backgroundColor: 'action.hover'
+            }
+          }}
+          aria-label={t("generative:rag.breadcrumbs.goBack")}
+        >
+          <ArrowBackIcon fontSize="small" />
+        </IconButton>
+      )}
+      <Breadcrumbs 
+        aria-label="breadcrumb" 
+        sx={{ 
+          minHeight: '24px', // Ensure consistent height
+          display: 'flex',
+          alignItems: 'center'
+        }}
+      >
+        {breadcrumbs.map((breadcrumb, index) => {
+          if (breadcrumb.current) {
+            return (
+              <Typography key={index} color="text.primary">
+                {breadcrumb.isSession ? (
+                  <>
+                    <em>{breadcrumb.label}</em> {t("generative:rag.breadcrumbs.sessionSuffix")}
+                  </>
+                ) : (
+                  breadcrumb.label
+                )}
+              </Typography>
+            );
+          } else {
+            return (
+              <Link
+                key={index}
+                underline="hover"
+                color="inherit"
+                href="#"
+                onClick={(e) => {
+                  e.preventDefault();
+                  handleNavigate(breadcrumb.path);
+                }}
+                sx={{ cursor: "pointer" }}
+              >
+                {breadcrumb.label}
+              </Link>
+            );
+          }
+        })}
+      </Breadcrumbs>
+    </Box>
+  );
+}
+
+export default RAGBreadcrumbs;
\ No newline at end of file
diff --git a/DashAI/front/src/components/generative/RAG/RAGDocumentsPanel.jsx b/DashAI/front/src/components/generative/RAG/RAGDocumentsPanel.jsx
new file mode 100644
index 000000000..8fca3e9d4
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/RAGDocumentsPanel.jsx
@@ -0,0 +1,34 @@
+import React from "react";
+import PropTypes from "prop-types";
+import DocumentsBar from "./DocumentsBar";
+
+/**
+ * Unified RAG Documents Panel wrapper
+ * Handles consistent DocumentsBar rendering across all RAG contexts
+ * 
+ * @param {string} selectedSessionId - Session ID for session-specific documents
+ * @param {boolean} isRagChatActive - Whether RAG chat is currently active
+ * @param {function} onDocumentChange - Optional callback for document changes
+ */
+export default function RAGDocumentsPanel({
+  selectedSessionId,
+  isRagChatActive = false,
+  onDocumentChange,
+  showSearch = false,
+}) {
+  return (
+    <DocumentsBar
+      selectedSessionId={selectedSessionId}
+      taskName="RAGTask"
+      onDocumentChange={onDocumentChange}
+      showSearch={showSearch}
+    />
+  );
+}
+
+RAGDocumentsPanel.propTypes = {
+  selectedSessionId: PropTypes.oneOfType([PropTypes.string, PropTypes.number]),
+  isRagChatActive: PropTypes.bool,
+  onDocumentChange: PropTypes.func,
+  showSearch: PropTypes.bool,
+};
diff --git a/DashAI/front/src/components/generative/RAG/RAGInfoBar.jsx b/DashAI/front/src/components/generative/RAG/RAGInfoBar.jsx
new file mode 100644
index 000000000..c0b163873
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/RAGInfoBar.jsx
@@ -0,0 +1,186 @@
+import { Box, Typography } from "@mui/material";
+import { useTranslation } from "react-i18next";
+
+function Section({ title, children }) {
+  return (
+    <Box>
+      <Typography variant="subtitle2" sx={{ mb: 1 }}>
+        {title}
+      </Typography>
+      {children}
+    </Box>
+  );
+}
+
+export default function RAGInfoBar() {
+  const { t } = useTranslation(["generative"]);
+
+  return (
+    <Box display="flex" flexDirection="column" gap={2}>
+      <Box>
+        <Typography variant="subtitle1">
+          {t("generative:rag.rightPanel.title")}
+        </Typography>
+        <Typography variant="body2" color="textSecondary">
+          {t("generative:rag.rightPanel.subtitle")}
+        </Typography>
+      </Box>
+
+      <Section title={t("generative:rag.rightPanel.howItWorksTitle")}>
+        <Box component="ul" sx={{ pl: 2, my: 0 }}>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.howItWorks.step1")}
+          </Typography>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.howItWorks.step2")}
+          </Typography>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.howItWorks.step3")}
+          </Typography>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.howItWorks.step4")}
+          </Typography>
+        </Box>
+      </Section>
+
+      <Section title={t("generative:rag.rightPanel.settingsTitle")}>
+        <Box display="flex" flexDirection="column" gap={2}>
+          <Box>
+            <Typography variant="body2">
+              {t("generative:rag.rightPanel.settings.documents.title")}
+            </Typography>
+            <Box component="ul" sx={{ pl: 2, my: 0 }}>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.documents.point1",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.documents.point2",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.documents.point3",
+                )}
+              </Typography>
+            </Box>
+          </Box>
+
+          <Box>
+            <Typography variant="body2">
+              {t("generative:rag.rightPanel.settings.chunking.title")}
+            </Typography>
+            <Box component="ul" sx={{ pl: 2, my: 0 }}>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.chunking.point1",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.chunking.point2",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.chunking.point3",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.chunking.point4",
+                )}
+              </Typography>
+            </Box>
+          </Box>
+
+          <Box>
+            <Typography variant="body2">
+              {t(
+                "generative:rag.rightPanel.settings.retriever.title",
+              )}
+            </Typography>
+            <Box component="ul" sx={{ pl: 2, my: 0 }}>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.retriever.point1",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.retriever.point2",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.retriever.point3",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.retriever.point4",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.retriever.point5",
+                )}
+              </Typography>
+            </Box>
+          </Box>
+
+          <Box>
+            <Typography variant="body2">
+              {t("generative:rag.rightPanel.settings.llm.title")}
+            </Typography>
+            <Box component="ul" sx={{ pl: 2, my: 0 }}>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t("generative:rag.rightPanel.settings.llm.point1")}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t("generative:rag.rightPanel.settings.llm.point2")}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t("generative:rag.rightPanel.settings.llm.point3")}
+              </Typography>
+            </Box>
+          </Box>
+
+          <Box>
+            <Typography variant="body2">
+              {t("generative:rag.rightPanel.settings.prompt.title")}
+            </Typography>
+            <Box component="ul" sx={{ pl: 2, my: 0 }}>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.prompt.point1",
+                )}
+              </Typography>
+              <Typography component="li" variant="body2" color="textSecondary">
+                {t(
+                  "generative:rag.rightPanel.settings.prompt.point2",
+                )}
+              </Typography>
+            </Box>
+          </Box>
+        </Box>
+      </Section>
+
+      <Section title={t("generative:rag.rightPanel.tipsTitle")}>
+        <Box component="ul" sx={{ pl: 2, my: 0 }}>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.tips.tip1")}
+          </Typography>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.tips.tip2")}
+          </Typography>
+          <Typography component="li" variant="body2" color="textSecondary">
+            {t("generative:rag.rightPanel.tips.tip3")}
+          </Typography>
+        </Box>
+      </Section>
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/RAGParamsPanel.jsx b/DashAI/front/src/components/generative/RAG/RAGParamsPanel.jsx
new file mode 100644
index 000000000..e8ab96691
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/RAGParamsPanel.jsx
@@ -0,0 +1,117 @@
+import React, { useEffect, useState } from "react";
+import { Box, Button, Typography } from "@mui/material";
+import SideBar from "../../threeSectionLayout/panelContainers/SideBar";
+import { useSnackbar } from "notistack";
+import { useTranslation } from "react-i18next";
+import {
+  getRAGSession,
+  updateGenerativeSessionParams,
+} from "../../../api/rag";
+import PromptParamsCard from "./PromptParamsCard";
+import GeneratorParamsCard from "./GeneratorParamsCard";
+
+export default function RAGParamsPanel({ selectedSessionId }) {
+  const { t } = useTranslation(["generative"]);
+  const { enqueueSnackbar } = useSnackbar();
+  const [promptModel, setPromptModel] = useState({ component: "", params: {} });
+  const [generatorModel, setGeneratorModel] = useState({ component: null, params: {} });
+  const [loading, setLoading] = useState(false);
+  const [isValid, setIsValid] = useState(true);
+  const [savedVersion, setSavedVersion] = useState(0);
+  const originalParamsRef = React.useRef(null);
+
+  useEffect(() => {
+    if (!selectedSessionId) return;
+    setLoading(true);
+    getRAGSession(selectedSessionId)
+      .then((session) => {
+        const params = session.parameters || {};
+        setPromptModel(params.prompt || { component: "", params: {} });
+        setGeneratorModel({
+          component: params.generation_model?.component || null,
+          params: params.generation_model?.params || {},
+        });
+        // Store original parameters snapshot
+        originalParamsRef.current = {
+          prompt: params.prompt || { component: "", params: {} },
+          generation_model: {
+            component: params.generation_model?.component || null,
+            params: params.generation_model?.params || {},
+          },
+        };
+      })
+      .catch((err) => {
+        console.error("Failed to load RAG session:", err);
+        enqueueSnackbar(t("generative:rag.paramsPanel.failedToLoad"), { variant: "error" });
+      })
+      .finally(() => setLoading(false));
+  }, [selectedSessionId, enqueueSnackbar]);
+
+  const hasParamChanges = React.useMemo(() => {
+    if (!originalParamsRef.current) return false;
+    const original = originalParamsRef.current;
+    const hasPromptChanged = JSON.stringify(original.prompt) !== JSON.stringify(promptModel);
+    const hasComponentChanged = original.generation_model.component !== generatorModel.component;
+    const hasParamsChanged = JSON.stringify(original.generation_model.params) !== JSON.stringify(generatorModel.params);
+    return hasPromptChanged || hasComponentChanged || hasParamsChanged;
+  }, [promptModel, generatorModel, savedVersion]);
+
+  const handleSave = async () => {
+    if (!selectedSessionId || !hasParamChanges) return;
+    const payload = {
+      parameters: {
+        prompt: promptModel,
+        generation_model: {
+          component: generatorModel.component,
+          params: generatorModel.params,
+        },
+      },
+    };
+
+    try {
+      await updateGenerativeSessionParams(selectedSessionId, payload.parameters);
+      enqueueSnackbar(t("generative:rag.paramsPanel.updated"), { variant: "success" });
+      // Update the snapshot after successful save and trigger recalculation
+      originalParamsRef.current = payload.parameters;
+      setSavedVersion((v) => v + 1);
+    } catch (err) {
+      console.error("Failed to update RAG session:", err);
+      enqueueSnackbar(t("generative:rag.paramsPanel.failedToUpdate"), { variant: "error" });
+    }
+  };
+
+  return (
+    <SideBar>
+      <Box sx={{ p: 2, height: "100%", display: "flex", flexDirection: "column", gap: 2 }}>
+        <Typography variant="h6">{t("generative:rag.paramsPanel.title")}</Typography>
+
+        <Box sx={{ overflow: "auto", flex: 1 }}>
+          <Box sx={{ mb: 2 }}>
+            <PromptParamsCard promptModel={promptModel} setPromptModel={setPromptModel} onTokenCountChange={() => {}} />
+          </Box>
+
+          <Box sx={{ mt: 2 }}>
+            <GeneratorParamsCard
+              generatorModel={generatorModel}
+              setGeneratorModel={setGeneratorModel}
+              chunkSize={0}
+              topK={0}
+              promptTokenCount={0}
+              setIsValid={setIsValid}
+            />
+          </Box>
+        </Box>
+
+        <Box sx={{ display: "flex", gap: 1, justifyContent: "flex-end", pt: 1 }}>
+          <Button 
+            variant="contained" 
+            onClick={handleSave} 
+            disabled={loading || !isValid || !selectedSessionId || !hasParamChanges}
+          >
+            {t("generative:rag.paramsPanel.save")}
+          </Button>
+        </Box>
+      </Box>
+    </SideBar>
+  );
+}
diff --git a/DashAI/front/src/components/generative/RAG/RAGSessionSummary.jsx b/DashAI/front/src/components/generative/RAG/RAGSessionSummary.jsx
new file mode 100644
index 000000000..2b93b11f6
--- /dev/null
+++ b/DashAI/front/src/components/generative/RAG/RAGSessionSummary.jsx
@@ -0,0 +1,582 @@
+import { useEffect, useMemo, useRef, useState } from "react";
+import { 
+  Box, 
+  Typography, 
+  Button, 
+  Chip, 
+  Grid, 
+  Card, 
+  CardContent,
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  IconButton,
+  Collapse,
+  TextField
+} from "@mui/material";
+import ChatIcon from "@mui/icons-material/Chat";
+import SettingsIcon from "@mui/icons-material/Settings";
+import InfoIcon from "@mui/icons-material/Info";
+import CloseIcon from "@mui/icons-material/Close";
+import ExpandMoreIcon from "@mui/icons-material/ExpandMore";
+import ContentCutIcon from "@mui/icons-material/ContentCut";
+import LeaderboardIcon from "@mui/icons-material/Leaderboard";
+import BoltIcon from "@mui/icons-material/Bolt";
+import ContentCopyIcon from "@mui/icons-material/ContentCopy";
+import EditIcon from "@mui/icons-material/Edit";
+import SaveIcon from "@mui/icons-material/Save";
+import { useSnackbar } from "notistack";
+import { useTranslation } from "react-i18next";
+import { getGenerativeSession } from "../../../api/generativeTask";
+import { updateGenerativeSession, getSessions } from "../../../api/session";
+import RAGBreadcrumbs from "./RAGBreadcrumbs";
+
+export default function RAGSessionSummary({
+  sessionId,
+  onStartChat,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [sessionData, setSessionData] = useState(null);
+  const [loading, setLoading] = useState(true);
+  const [modalOpen, setModalOpen] = useState(false);
+  const [modalContent, setModalContent] = useState(null);
+  const [isEditing, setIsEditing] = useState(false);
+  const [editingName, setEditingName] = useState("");
+  const [nameError, setNameError] = useState("");
+  const [editingDescription, setEditingDescription] = useState("");
+  const [isSaving, setIsSaving] = useState(false);
+  const [expandedSections, setExpandedSections] = useState({
+    chunking: false,
+    retriever: false,
+    generation: false
+  });
+  const { enqueueSnackbar } = useSnackbar();
+  const originalMetadataRef = useRef({ name: "", description: "" });
+
+  const toggleSection = (section) => {
+    setExpandedSections(prev => ({
+      ...prev,
+      [section]: !prev[section]
+    }));
+  };
+
+  const hasMetadataChanges = useMemo(() => {
+    const normalizedName = (editingName || "").trim();
+    const normalizedDescription = editingDescription || "";
+    const originalName = originalMetadataRef.current.name || "";
+    const originalDescription = originalMetadataRef.current.description || "";
+
+    return normalizedName !== originalName || normalizedDescription !== originalDescription;
+  }, [editingDescription, editingName]);
+
+  // Toggle edit mode
+  const handleToggleEditMode = () => {
+    if (!isEditing && sessionData) {
+      const currentName = sessionData.name || "";
+      const currentDescription = sessionData.description || "";
+      originalMetadataRef.current = {
+        name: currentName,
+        description: currentDescription,
+      };
+      setEditingName(currentName);
+      setEditingDescription(currentDescription);
+      setNameError("");
+    }
+    setIsEditing(!isEditing);
+  };
+
+  const handleSaveMetadata = async () => {
+    if (!sessionData) return;
+    if (!hasMetadataChanges || isSaving) return;
+
+    const newName = (editingName || "").trim();
+    if (!newName) {
+      setNameError(t("generative:rag.validation.nameRequired"));
+      enqueueSnackbar(t("generative:rag.validation.nameRequired"), { variant: "error" });
+      return;
+    }
+
+    try {
+      setIsSaving(true);
+
+      // Validate uniqueness: no other session (different id) should have the same name
+      const allSessions = await getSessions();
+      const duplicate = allSessions.find((s) => s.name === newName && s.id !== sessionData.id);
+      if (duplicate) {
+        setNameError(t("generative:rag.validation.nameUnique"));
+        enqueueSnackbar(t("generative:rag.validation.nameUnique"), { variant: "error" });
+        setIsSaving(false);
+        return;
+      }
+
+      await updateGenerativeSession({
+        id: String(sessionData.id),
+        formData: {
+          name: newName,
+          description: editingDescription,
+        },
+      });
+
+      setSessionData({
+        ...sessionData,
+        name: newName,
+        description: editingDescription,
+      });
+      originalMetadataRef.current = {
+        name: newName,
+        description: editingDescription || "",
+      };
+
+      setIsEditing(false);
+      enqueueSnackbar(t("generative:rag.summary.sessionUpdated"), {
+        variant: "success",
+      });
+    } catch (error) {
+      enqueueSnackbar(t("generative:rag.summary.failedToUpdateSession"), {
+        variant: "error",
+      });
+      console.error("Failed to update session metadata:", error);
+    } finally {
+      setIsSaving(false);
+    }
+  };
+
+  // Helper function to determine if a parameter should be shown inline or in modal
+  const isSimpleParameter = (value) => {
+    return (
+      typeof value === 'string' ||
+      typeof value === 'number' ||
+      typeof value === 'boolean' ||
+      value === null ||
+      value === undefined ||
+      (Array.isArray(value) && value.length <= 3 && value.every(item => typeof item !== 'object'))
+    );
+  };
+
+  const formatSimpleValue = (value) => {
+    if (value === null || value === undefined) return 'null';
+    if (typeof value === 'boolean') return value.toString();
+    if (Array.isArray(value)) return `[${value.join(', ')}]`;
+    return String(value);
+  };
+
+  const isApiKey = (key) => {
+    return key.toLowerCase().includes('api');
+  };
+
+  const maskApiKey = (value) => {
+    const stringValue = String(value);
+    if (stringValue.length <= 10) return stringValue;
+    return stringValue.substring(0, 15) + '...';
+  };
+
+  const handleCopyToClipboard = (value) => {
+    navigator.clipboard.writeText(String(value)).then(() => {
+      enqueueSnackbar(t("generative:rag.summary.apiKeyCopied"), { variant: "success" });
+    }).catch(() => {
+      enqueueSnackbar(t("generative:rag.summary.failedToCopyApiKey"), { variant: "error" });
+    });
+  };
+
+  const handleParameterClick = (paramName, paramValue, componentName) => {
+    setModalContent({
+      title: t("generative:rag.summary.parameterDetailTitle", { component: componentName, param: paramName }),
+      content: paramValue
+    });
+    setModalOpen(true);
+  };
+
+  const renderParametersList = (params, componentName) => {
+    if (!params || Object.keys(params).length === 0) {
+      return (
+        <Typography variant="caption" color="text.primary" sx={{ ml: 1 }}>
+          {t("generative:rag.summary.noParameters")}
+        </Typography>
+      );
+    }
+
+    return Object.entries(params).map(([key, value]) => {
+      const isSimple = isSimpleParameter(value);
+      const isAPI = isApiKey(key);
+      
+      return (
+        <Box key={key} display="flex" alignItems="center" sx={{ ml: 1, mb: 0.25 }}>
+          <Typography variant="caption" color="text.secondary">
+            • {key}: 
+          </Typography>
+          {isSimple ? (
+            <Box sx={{ display: 'flex', alignItems: 'center', gap: 0.5, ml: 0.5 }}>
+              <Typography variant="caption" color="text.secondary">
+                {isAPI ? maskApiKey(value) : formatSimpleValue(value)}
+              </Typography>
+              {isAPI && (
+                <IconButton
+                  size="small"
+                  onClick={() => handleCopyToClipboard(value)}
+                  sx={{
+                    p: 0.25,
+                    minWidth: 'auto',
+                    color: 'text.secondary',
+                    '&:hover': {
+                      color: 'primary.main',
+                      backgroundColor: 'action.hover'
+                    }
+                  }}
+                  title={t("generative:rag.summary.apiKeyCopied")}
+                >
+                  <ContentCopyIcon sx={{ fontSize: '0.75rem' }} />
+                </IconButton>
+              )}
+            </Box>
+          ) : (
+            <Button
+              variant="text"
+              size="small"
+              startIcon={<InfoIcon sx={{ fontSize: '0.75rem' }} />}
+              onClick={() => handleParameterClick(key, value, componentName)}
+              sx={{ 
+                ml: 0.5, 
+                p: 0.25,
+                minHeight: 'auto',
+                textTransform: 'none',
+                color: 'primary.main',
+                '&:hover': {
+                  backgroundColor: 'action.hover'
+                }
+              }}
+            >
+              {t("generative:rag.summary.viewDetails")}
+            </Button>
+          )}
+        </Box>
+      );
+    });
+  };
+
+  const CollapsibleParameterCard = ({ icon, title, component, params, sectionKey, componentName }) => (
+    <Box 
+      sx={{ 
+        border: 1, 
+        borderColor: 'divider', 
+        borderRadius: 1, 
+        backgroundColor: 'background.box',
+        mt: 2,
+        overflow: 'hidden',
+        mb: 2
+      }}
+    >
+      {/* Header */}
+      <Box 
+        sx={{ 
+          display: 'flex', 
+          alignItems: 'center', 
+          justifyContent: 'space-between',
+          p: 2,
+          backgroundColor: 'background.box',
+          cursor: 'pointer'
+        }}
+        onClick={() => toggleSection(sectionKey)}
+      >
+        <Box sx={{ display: 'flex', alignItems: 'center', gap: 2 }}>
+          <Typography variant="body2">{icon}</Typography>
+          <Typography variant="subtitle1">
+            {title}: <Typography component="span" color="text.primary">{component}</Typography>
+          </Typography>
+        </Box>
+        
+        <IconButton size="small">
+          <ExpandMoreIcon 
+            sx={{ 
+              transform: expandedSections[sectionKey] ? 'rotate(180deg)' : 'rotate(0deg)',
+              transition: 'transform 0.3s'
+            }} 
+          />
+        </IconButton>
+      </Box>
+
+      <Collapse in={expandedSections[sectionKey]} timeout="auto">
+        <Box sx={{ p: 2, pt: 1, backgroundColor: 'background.box' }}>
+          {renderParametersList(params, componentName)}
+        </Box>
+      </Collapse>
+    </Box>
+  );
+
+  useEffect(() => {
+    const fetchSessionData = async () => {
+      if (!sessionId) return;
+
+      try {
+        setLoading(true);
+        const session = await getGenerativeSession(sessionId);
+        setSessionData(session);
+      } catch (error) {
+        enqueueSnackbar(t("generative:error.failedToFetchSessionInfo"), {
+          variant: "error",
+        });
+        console.error("Failed to fetch session data:", error);
+      } finally {
+        setLoading(false);
+      }
+    };
+
+    fetchSessionData();
+  }, [sessionId, enqueueSnackbar, t]);
+
+  if (loading) {
+    return (
+      <Box
+        display="flex"
+        alignItems="center"
+        justifyContent="center"
+        height="100%"
+      >
+        <Typography variant="h6" color="text.secondary">
+          {t("generative:rag.summary.loading")}
+        </Typography>
+      </Box>
+    );
+  }
+
+  if (!sessionData) {
+    return (
+      <Box
+        display="flex"
+        alignItems="center"
+        justifyContent="center"
+        height="100%"
+      >
+        <Typography variant="h6" color="text.secondary">
+          {t("generative:rag.summary.sessionNotFound")}
+        </Typography>
+      </Box>
+    );
+  }
+
+  const parameters = sessionData.parameters || {};
+
+  return (
+    <Box
+      display="flex"
+      flexDirection="column"
+      height="100%"
+      px={3}
+      gap={3}
+      overflow="auto"
+    >
+      {/* RAG Breadcrumbs */}
+      <RAGBreadcrumbs sessionName={sessionData?.name} />
+
+      <Box 
+        display="flex" 
+        width="100%"
+        justifyContent="space-between"
+        alignItems="flex-start"
+        paddingBottom={2}
+        gap={2}
+      >
+        {/* Left: Session Info (Editable/Readable) */}
+        <Box flexGrow={1} display="flex" flexDirection="column">
+          {isEditing ? (
+            <Box display="flex" flexDirection="column" gap={2}>
+              <TextField
+                fullWidth
+                label={t("generative:rag.summary.sessionName")}
+                value={editingName}
+                onChange={(e) => {
+                  setEditingName(e.target.value);
+                  // clear inline error while typing
+                  if (nameError) setNameError("");
+                }}
+                error={Boolean(nameError)}
+                helperText={nameError}
+                variant="outlined"
+                size="small"
+              />
+              <TextField
+                fullWidth
+                label={t("generative:rag.summary.sessionDescription")}
+                value={editingDescription}
+                onChange={(e) => setEditingDescription(e.target.value)}
+                variant="outlined"
+                size="small"
+                multiline
+                rows={2}
+              />
+            </Box>
+          ) : (
+            <>
+              <Box display="flex" alignItems="center" gap={1}>
+                <Typography variant="h4" gutterBottom sx={{ mr: 1 }}>
+                  {sessionData.name}
+                </Typography>
+                <IconButton
+                  size="small"
+                  onClick={handleToggleEditMode}
+                  sx={{ color: 'text.secondary' }}
+                  aria-label={t("generative:rag.summary.edit")}
+                >
+                  <EditIcon fontSize="small" />
+                </IconButton>
+              </Box>
+              <Typography variant="body1" color="text.secondary" mb={2}>
+                {sessionData.description || t("generative:rag.summary.descriptionPlaceholder")}
+              </Typography>
+              <Box display="flex" gap={1}>
+                <Chip 
+                  label={t("generative:rag.summary.title")} 
+                  color="text.secondary" 
+                  variant="outlined"
+                  size="small"
+                  sx={{ color: 'text.secondary' }}
+                />
+                <Chip 
+                  label={`${t("generative:rag.summary.createdLabel")} ${new Date(sessionData.created).toLocaleDateString()}`}
+                  variant="outlined"
+                  size="small"
+                  color="text.secondary"
+                  sx={{ color: 'text.secondary' }}
+                />
+              </Box>
+            </>
+          )}
+        </Box>
+
+        {/* Right: Action Buttons */}
+        <Box display="flex" flexDirection="column" gap={1} alignItems="flex-end">
+          {isEditing ? (
+            <>
+              <Button
+                variant="contained"
+                color="primary"
+                onClick={handleSaveMetadata}
+                disabled={isSaving || !hasMetadataChanges}
+                aria-label={t("generative:rag.summary.save")}
+                sx={{ minWidth: 40, width: 40, height: 40, padding: 0 }}
+              >
+                <SaveIcon fontSize="small" />
+              </Button>
+              <IconButton
+                size="small"
+                onClick={handleToggleEditMode}
+                disabled={isSaving}
+                aria-label={t("generative:rag.summary.cancel")}
+                sx={{ color: 'text.secondary' }}
+              >
+                <CloseIcon fontSize="small" />
+              </IconButton>
+            </>
+          ) : (
+            <Button
+              variant="contained"
+              size="large"
+              startIcon={<ChatIcon />}
+              onClick={onStartChat}
+              sx={{
+                py: 1.5,
+                px: 4,
+                borderRadius: 2,
+                boxShadow: 3,
+                textTransform: 'none',
+                '&:hover': {
+                  boxShadow: 6,
+                  transform: 'translateY(-1px)',
+                },
+                transition: 'all 0.2s ease-in-out'
+              }}
+            >
+              {t("generative:rag.summary.openChatButton")}
+            </Button>
+          )}
+        </Box>
+      </Box>
+
+      <Grid container spacing={3} sx={{ width: '100%' }}>
+        <Grid item sx={{ backgroundColor: 'background.box', width: '100%' }}>
+          <Card sx={{ backgroundColor: 'background.box' }}>
+            <CardContent>
+              <Box display="flex" alignItems="center" mb={3}>
+                <SettingsIcon sx={{ mr: 1, color: 'primary.main' }} />
+                <Typography variant="h6">
+                  {t("generative:rag.summary.detailedConfig")}
+                </Typography>
+              </Box>
+              
+              {parameters.chunking_model && (
+                <CollapsibleParameterCard
+                  icon={<ContentCutIcon fontSize="small" sx={{ color: 'text.secondary' }} />}
+                  title={t("generative:rag.summary.chunkingModel")}
+                  component={parameters.chunking_model.component}
+                  params={parameters.chunking_model.params}
+                  sectionKey="chunking"
+                  componentName={t("generative:rag.summary.chunkingModel")}
+                />
+              )}
+
+              {parameters.retriever_model && (
+                <CollapsibleParameterCard
+                  icon={<LeaderboardIcon fontSize="small" sx={{ color: 'text.secondary' }} />}
+                  title={t("generative:rag.summary.retrieverModel")}
+                  component={parameters.retriever_model.component}
+                  params={parameters.retriever_model.params}
+                  sectionKey="retriever"
+                  componentName={t("generative:rag.summary.retrieverModel")}
+                />
+              )}
+
+              {parameters.generation_model && (
+                <CollapsibleParameterCard
+                  icon={<BoltIcon fontSize="small" sx={{ color: 'text.secondary' }} />}
+                  title={t("generative:rag.summary.generationModel")}
+                  component={parameters.generation_model.component}
+                  params={parameters.generation_model.params}
+                  sectionKey="generation"
+                  componentName={t("generative:rag.summary.generationModel")}
+                />
+              )}
+            </CardContent>
+          </Card>
+        </Grid>
+      </Grid>
+
+      <Dialog
+        open={modalOpen}
+        onClose={() => setModalOpen(false)}
+        maxWidth="md"
+        fullWidth
+      >
+        <DialogTitle>
+          <Box display="flex" justifyContent="space-between" alignItems="center">
+            <Typography variant="h6">
+              {modalContent?.title || t("generative:rag.summary.parameterDetails")}
+            </Typography>
+            <IconButton onClick={() => setModalOpen(false)} size="small">
+              <CloseIcon />
+            </IconButton>
+          </Box>
+        </DialogTitle>
+        <DialogContent>
+          <Box>
+            <Typography variant="body2" color="text.secondary" gutterBottom>
+              {t("generative:rag.summary.completeConfig")}
+            </Typography>
+            <Box
+              component="pre"
+              sx={{
+                bgcolor: 'background.paper',
+                color: 'text.secondary',
+                p: 2,
+              }}
+            >
+              {modalContent?.content ? JSON.stringify(modalContent.content, null, 2) : ''}
+            </Box>
+          </Box>
+        </DialogContent>
+        <DialogActions>
+          <Button onClick={() => setModalOpen(false)}>{t("generative:rag.summary.close")}</Button>
+        </DialogActions>
+      </Dialog>
+    </Box>
+  );
+}
diff --git a/DashAI/front/src/components/generative/SessionBar.jsx b/DashAI/front/src/components/generative/SessionBar.jsx
index b8ad8a2c5..c038bc96c 100644
--- a/DashAI/front/src/components/generative/SessionBar.jsx
+++ b/DashAI/front/src/components/generative/SessionBar.jsx
@@ -13,11 +13,27 @@ import SideBar from "../threeSectionLayout/panelContainers/SideBar";
 import { useTranslation } from "react-i18next";
 import { useGenerative } from "./GenerativeContext";
 
-export default function SessionBar({ onToggle }) {
+export default function SessionBar({
+  onToggle,
+  sessions: sessionsProp,
+  selectedSessionId: selectedSessionIdProp,
+  handleSessionClick: handleSessionClickProp,
+  handleNewSessionButton: handleNewSessionButtonProp,
+  handleSessionDelete: handleSessionDeleteProp,
+  showSearch = true,
+}) {
   const theme = useTheme();
   const navigate = useNavigate();
-  const { tasks, sessions, selectedSessionId, deleteSessionById, editSession } =
-    useGenerative();
+  const {
+    tasks,
+    sessions: sessionsCtx,
+    selectedSessionId: selectedSessionIdCtx,
+    deleteSessionById,
+    editSession,
+  } = useGenerative();
+
+  const sessions = sessionsProp ?? sessionsCtx ?? [];
+  const selectedSessionId = selectedSessionIdProp ?? selectedSessionIdCtx;
   const [searchQuery, setSearchQuery] = useState("");
   const [filteredSessions, setFilteredSessions] = useState(sessions);
   const [selectedInfoSession, setSelectedInfoSession] = useState(null);
@@ -58,7 +74,7 @@ export default function SessionBar({ onToggle }) {
       if (prevKeys === newKeys) return prev;
       return initialOpenState;
     });
-  }, [sessions, tasks]);
+  }, [sessions, tasks, taskDisplayNameMap, t]);
 
   useEffect(() => {
     if (searchQuery.trim() === "") {
@@ -84,14 +100,32 @@ export default function SessionBar({ onToggle }) {
   };
 
   const handleNewSessionButton = () => {
+    if (handleNewSessionButtonProp) {
+      handleNewSessionButtonProp();
+      return;
+    }
     navigate("/app/generative");
   };
 
   const handleSessionClick = (sessionId) => {
+    const session = sessions.find((s) => s.id === sessionId);
+    if (!session) return;
+
+    if (handleSessionClickProp) {
+      const displayName =
+        taskDisplayNameMap[session.task_name] || t("common:other");
+      handleSessionClickProp(sessionId, session.task_name, displayName);
+      return;
+    }
+
     navigate(`/app/generative/sessions/${sessionId}`);
   };
 
   const handleSessionDelete = async (id) => {
+    if (handleSessionDeleteProp) {
+      await handleSessionDeleteProp(id);
+      return;
+    }
     const wasSelected = id === selectedSessionId;
     const ok = await deleteSessionById(id);
     if (ok && wasSelected) {
@@ -157,8 +191,8 @@ export default function SessionBar({ onToggle }) {
         </Box>
 
         {/* Search Bar */}
-        {sessions.length > SEARCH_THRESHOLD && (
-          <Box px={4} pb={4} flex={"0 0 auto"}>
+        {showSearch && sessions.length > SEARCH_THRESHOLD && (
+          <Box px={2} pb={2} flex={"0 0 auto"}>
             <SearchBar
               placeholder={t("generative:label.searchSessions")}
               value={searchQuery}
diff --git a/DashAI/front/src/components/generative/SourcesDisplay.jsx b/DashAI/front/src/components/generative/SourcesDisplay.jsx
new file mode 100644
index 000000000..7301fa889
--- /dev/null
+++ b/DashAI/front/src/components/generative/SourcesDisplay.jsx
@@ -0,0 +1,215 @@
+import React, { useState } from 'react';
+import { useTranslation } from 'react-i18next';
+import {
+  Box,
+  Typography,
+  Collapse,
+  IconButton,
+  List,
+  ListItem,
+  Divider,
+  Button,
+} from '@mui/material';
+import {
+  ExpandMore,
+  ExpandLess,
+  Source as SourceIcon,
+  Description as DescriptionIcon,
+  Visibility as VisibilityIcon,
+} from '@mui/icons-material';
+import DocumentReferencesModal from './DocumentReferencesModal';
+
+const SourcesDisplay = ({ references, onOpenReference, isUser = false }) => {
+  const { t } = useTranslation('generative');
+  const [expanded, setExpanded] = useState(false);
+  const [documentModalOpen, setDocumentModalOpen] = useState(false);
+  const [selectedDocument, setSelectedDocument] = useState(null);
+  const [selectedChunks, setSelectedChunks] = useState([]);
+  const referenceEntries = references ? Object.entries(references) : [];
+  const sourceCount = referenceEntries.length;
+
+  const handleToggleExpanded = () => {
+    setExpanded(!expanded);
+  };
+
+  const handleOpenDocumentModal = (docId, chunks) => {
+    setSelectedDocument({ id: docId });
+    setSelectedChunks(chunks);
+    setDocumentModalOpen(true);
+  };
+
+  const handleCloseDocumentModal = () => {
+    setDocumentModalOpen(false);
+    setSelectedDocument(null);
+    setSelectedChunks([]);
+  };
+
+  // Group references by document ID
+  const groupedReferences = referenceEntries.reduce((acc, [key, ref]) => {
+    const docId = ref.document_id;
+    if (!acc[docId]) {
+      acc[docId] = [];
+    }
+    acc[docId].push({ key, ...ref });
+    return acc;
+  }, {});
+
+  // Get document title - fallback to "Document X" if no title available
+  const getDocumentTitle = (docId, chunks) => {
+    const firstChunk = chunks[0];
+    if (firstChunk.document_title) return firstChunk.document_title;
+    if (firstChunk.document_name) return firstChunk.document_name;
+    if (firstChunk.title) return firstChunk.title;
+    if (firstChunk.name) return firstChunk.name;
+    
+    return t('sourcesDisplay.fallbackTitle', { id: docId, defaultValue: `Document ${docId}` });
+  };
+
+  return (
+    <Box 
+      sx={{ 
+        mt: 1,
+        ml: isUser ? 0 : '40px',
+        mr: isUser ? '40px' : 0,
+        maxWidth: isUser ? 'calc(80% - 40px)' : 'calc(80% - 40px)',
+        border: 1, 
+        borderColor: 'divider', 
+        borderRadius: 1, 
+        backgroundColor: 'background.box',
+        overflow: 'hidden'
+      }}
+    >
+      {/* Header */}
+      <Box 
+        sx={{ 
+          display: 'flex', 
+          alignItems: 'center', 
+          justifyContent: 'space-between',
+          p: 1.5,
+          backgroundColor: 'action.hover',
+          cursor: 'pointer'
+        }}
+        onClick={handleToggleExpanded}
+      >
+        <Box sx={{ display: 'flex', alignItems: 'center', gap: 1.5 }}>
+          <SourceIcon fontSize="small" color="primary" sx={{ flexShrink: 0 }} />
+          <Typography variant="body2" sx={{ fontWeight: 'medium', flexShrink: 0 }}>
+            {t('sourcesDisplay.viewSources')}
+          </Typography>
+          <Box
+            sx={{
+              backgroundColor: 'primary.main',
+              color: 'primary.contrastText',
+              borderRadius: '10px',
+              minWidth: 20,
+              height: 20,
+              fontSize: '0.75rem',
+              fontWeight: 'bold',
+              display: 'flex',
+              alignItems: 'center',
+              justifyContent: 'center',
+              px: 0.5,
+              flexShrink: 0
+            }}
+          >
+            {sourceCount}
+          </Box>
+        </Box>
+        
+        <IconButton size="small">
+          {expanded ? <ExpandLess /> : <ExpandMore />}
+        </IconButton>
+      </Box>
+
+      {/* Sources List */}
+      <Collapse in={expanded} timeout="auto">
+        <List dense disablePadding>
+          {Object.entries(groupedReferences).map(([docId, chunks], index, array) => (
+            <React.Fragment key={docId}>
+              <ListItem disablePadding>
+                <Box sx={{ 
+                    width: '100%', 
+                    p: 2,
+                    display: 'flex',
+                    alignItems: 'center',
+                    justifyContent: 'space-between',
+                    backgroundColor: "background.box",
+                    gap: 2
+                }}>
+                  <Box sx={{ 
+                    display: 'flex', 
+                    alignItems: 'center', 
+                    gap: 1,
+                    minWidth: 0,
+                    flex: 1
+                  }}>
+                    <DescriptionIcon fontSize="small" color="primary" sx={{ flexShrink: 0 }} />
+
+                    <Box sx={{ display: 'flex', alignItems: 'center', minWidth: 0, flex: 1, gap: 0.5 }}>
+                      <Typography 
+                        component="span"
+                        variant="body2"
+                        sx={{
+                          fontWeight: 'bold',
+                          overflow: 'hidden',
+                          textOverflow: 'ellipsis',
+                          whiteSpace: 'nowrap',
+                          flexShrink: 1,
+                          minWidth: 0
+                        }}
+                      >
+                        {getDocumentTitle(docId, chunks)}
+                      </Typography>
+                      <Typography 
+                        component="span"
+                        variant="body2"
+                        color="text.secondary"
+                        sx={{
+                          flexShrink: 0,
+                          whiteSpace: 'nowrap'
+                        }}
+                      >
+                        {t('sourcesDisplay.provided')}{' '}
+                        <strong>{chunks.length}</strong>{' '}
+                        {t('sourcesDisplay.chunk', { count: chunks.length })}
+                      </Typography>
+                    </Box>
+                  </Box>
+                  
+                  <Button
+                    size="small"
+                    variant="outlined"
+                    startIcon={<VisibilityIcon />}
+                    onClick={() => handleOpenDocumentModal(docId, chunks)}
+                    sx={{
+                      fontSize: '0.75rem',
+                      py: 0.5,
+                      px: 1,
+                      whiteSpace: 'nowrap',
+                      minWidth: '140px',
+                      flexShrink: 0
+                    }}
+                  >
+                    {t('sourcesDisplay.viewChunks')}
+                  </Button>
+                </Box>
+              </ListItem>
+
+              {index < array.length - 1 && <Divider />}
+            </React.Fragment>
+          ))}
+        </List>
+      </Collapse>
+
+      <DocumentReferencesModal
+        open={documentModalOpen}
+        onClose={handleCloseDocumentModal}
+        document={selectedDocument}
+        chunks={selectedChunks}
+        onOpenReference={onOpenReference}
+      />
+    </Box>
+  );
+};
+
+export default SourcesDisplay;
diff --git a/DashAI/front/src/components/models/modelSession/EditModelDialog.jsx b/DashAI/front/src/components/models/modelSession/EditModelDialog.jsx
index dc28b6347..e22d39d71 100644
--- a/DashAI/front/src/components/models/modelSession/EditModelDialog.jsx
+++ b/DashAI/front/src/components/models/modelSession/EditModelDialog.jsx
@@ -45,12 +45,12 @@ function EditModelDialog({
         }}
       >
         <FormSchemaWithSelectedModel
+          modelToConfigure={modelToConfigure}
+          initialValues={paramsInitialValues}
           onFormSubmit={(values) => {
             updateParameters(values);
             setOpen(false);
           }}
-          modelToConfigure={modelToConfigure}
-          initialValues={paramsInitialValues}
           onCancel={() => setOpen(false)}
         />
       </FormSchemaDialog>
diff --git a/DashAI/front/src/components/shared/FormSchema.jsx b/DashAI/front/src/components/shared/FormSchema.jsx
index f2269a85a..26e91128f 100644
--- a/DashAI/front/src/components/shared/FormSchema.jsx
+++ b/DashAI/front/src/components/shared/FormSchema.jsx
@@ -34,6 +34,7 @@ function FormSchema({
   onValuesChange,
   showBorder = true,
   hideButtons = false,
+  excludeFields = [],
 }) {
   const { formik, modelSchema, loading, handleUpdateSchema } = useFormSchema({
     model,
@@ -61,6 +62,7 @@ function FormSchema({
           onFormSubmit={onFormSubmit}
           setError={setError}
           errorsMessage={errorsMessage}
+          excludeFields={excludeFields}
         />
       </FormSchemaParameterContainer>
 
@@ -92,6 +94,7 @@ FormSchema.propTypes = {
   onValuesChange: PropTypes.func,
   showBorder: PropTypes.bool,
   hideButtons: PropTypes.bool,
+  excludeFields: PropTypes.arrayOf(PropTypes.string),
 };
 
 export default FormSchema;
diff --git a/DashAI/front/src/components/shared/FormSchemaDialog.jsx b/DashAI/front/src/components/shared/FormSchemaDialog.jsx
index 966397b7d..433cca23e 100644
--- a/DashAI/front/src/components/shared/FormSchemaDialog.jsx
+++ b/DashAI/front/src/components/shared/FormSchemaDialog.jsx
@@ -46,7 +46,7 @@ function FormSchemaDialog({
             onFormSubmit={onFormSubmit}
           />
         </DialogTitle>
-        <DialogContent>
+        <DialogContent sx={{ overflow: "visible" }}>
           {React.cloneElement(children, { onClose: handleClose })}
         </DialogContent>
       </FormSchemaContainer>
diff --git a/DashAI/front/src/components/shared/FormSchemaFieldWithParent.jsx b/DashAI/front/src/components/shared/FormSchemaFieldWithParent.jsx
index 5c35bf03c..2905a55f5 100644
--- a/DashAI/front/src/components/shared/FormSchemaFieldWithParent.jsx
+++ b/DashAI/front/src/components/shared/FormSchemaFieldWithParent.jsx
@@ -1,8 +1,17 @@
-import { MenuItem, Tooltip, IconButton } from "@mui/material";
+import {
+  MenuItem,
+  Tooltip,
+  IconButton,
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+} from "@mui/material";
 import PropTypes from "prop-types";
 import { Input } from "../configurableObject/Inputs/InputStyles";
-import React from "react";
-import { useFormSchemaStore } from "../../contexts/schema";
+import React, { useState, useRef } from "react";
+import { useFormSchemaStore, FormSchemaProvider } from "../../contexts/schema";
 import {
   formattedModel,
   formattedSubform,
@@ -12,6 +21,9 @@ import useModelParents from "../../hooks/useModelParents";
 import { Settings } from "@mui/icons-material";
 import { useTranslation } from "react-i18next";
 import FormSchemaFieldCard from "./FormSchemaFieldCard";
+import FormSchema from "./FormSchema";
+import FormSchemaLayout from "./FormSchemaLayout";
+import useSchema from "../../hooks/useSchema";
 
 /**
  * Renders a parent-model selector field as a card.
@@ -25,21 +37,44 @@ function FormSchemaFieldWithParent({
   description,
   errorMessage,
 }) {
-  const { addProperty, getModelFromCurrentProperty } = useFormSchemaStore();
+  const { formValues, setFormValues, getModelFromCurrentProperty } =
+    useFormSchemaStore();
+  const parentComponent =
+    field?.value?.properties?.component ?? field?.value?.component ?? null;
   const { models } = useModelParents({
-    parent: field.value?.properties.component,
+    parent: parentComponent,
   });
   const { t } = useTranslation(["common"]);
 
+  // State for handling the sub-modal
+  const [openSubModal, setOpenSubModal] = useState(false);
+  const formikRef = useRef(null);
+
+  // Get the currently selected model name
+  const selectedModelName = getModelFromCurrentProperty(name);
+
+  // Get default values for the sub-model
+  const { defaultValues: subModelDefaultValues } = useSchema({
+    modelName: selectedModelName,
+  });
+
   const handleOnChange = async (event) => {
+    if (!parentComponent) {
+      return;
+    }
+
     const model = models?.find((model) => model.name === event.target.value);
+    if (!model?.schema) {
+      return;
+    }
+
     const { initialValues } = generateYupSchema(
       await formattedModel(model?.schema),
     );
 
     field.onChange(
       formattedSubform({
-        parent: field.value?.properties.component,
+        parent: parentComponent,
         model: model?.name,
         params: initialValues,
       }),
@@ -47,7 +82,50 @@ function FormSchemaFieldWithParent({
   };
 
   const handleClick = () => {
-    addProperty({ key: name, label });
+    if (selectedModelName) {
+      setOpenSubModal(true);
+    }
+  };
+
+  // Function to get current values for the sub-model
+  const getSubModelInitialValues = () => {
+    if (!formValues || !formValues[name]) {
+      return subModelDefaultValues || {};
+    }
+
+    const fieldValue = formValues[name];
+    if (fieldValue?.properties?.params?.comp?.params) {
+      return fieldValue.properties.params.comp.params;
+    }
+
+    return subModelDefaultValues || {};
+  };
+
+  // Handle saving the sub-model configuration
+  const handleSubModelSave = (values) => {
+    if (formValues && formValues[name]) {
+      setFormValues((prevFormValues) => {
+        const updatedFormValues = { ...prevFormValues };
+        if (!updatedFormValues[name]) {
+          updatedFormValues[name] = {};
+        }
+        if (!updatedFormValues[name].properties) {
+          updatedFormValues[name].properties = {};
+        }
+        if (!updatedFormValues[name].properties.params) {
+          updatedFormValues[name].properties.params = {};
+        }
+        if (!updatedFormValues[name].properties.params.comp) {
+          updatedFormValues[name].properties.params.comp = { params: {} };
+        }
+        updatedFormValues[name].properties.params.comp.params = {
+          ...updatedFormValues[name].properties.params.comp.params,
+          ...values,
+        };
+        return updatedFormValues;
+      });
+    }
+    setOpenSubModal(false);
   };
 
   const configureButton = (
@@ -59,36 +137,97 @@ function FormSchemaFieldWithParent({
   );
 
   return (
-    <FormSchemaFieldCard
-      label={label}
-      paramKey={name}
-      description={description}
-      errorMessage={errorMessage}
-      headerRight={configureButton}
-    >
-      <Input
-        select
-        value={getModelFromCurrentProperty(name)}
-        onChange={handleOnChange}
-        size="small"
-        sx={{ width: "100%" }}
+    <>
+      <FormSchemaFieldCard
+        label={label}
+        paramKey={name}
+        description={description}
+        errorMessage={errorMessage}
+        headerRight={configureButton}
+      >
+        <Input
+          select
+          value={getModelFromCurrentProperty(name)}
+          onChange={handleOnChange}
+          size="small"
+          sx={{ width: "100%" }}
+        >
+          {models?.map((model) => (
+            <MenuItem key={model.name} value={model.name}>
+              {model.name}
+            </MenuItem>
+          ))}
+        </Input>
+      </FormSchemaFieldCard>
+
+      <Dialog
+        open={openSubModal}
+        onClose={() => setOpenSubModal(false)}
+        maxWidth="sm"
+        fullWidth
+        PaperProps={{
+          sx: {
+            maxHeight: "80vh",
+            zIndex: 1400,
+          },
+        }}
+        BackdropProps={{
+          sx: {
+            zIndex: 1399,
+          },
+        }}
       >
-        {models?.map((model) => (
-          <MenuItem key={model.name} value={model.name}>
-            {model.name}
-          </MenuItem>
-        ))}
-      </Input>
-    </FormSchemaFieldCard>
+        <DialogTitle>{`${selectedModelName} Configuration`}</DialogTitle>
+        <FormSchemaProvider key={`subform-provider-${selectedModelName}`}>
+          <DialogContent dividers sx={{ minHeight: 200 }}>
+            {selectedModelName && (
+              <FormSchemaLayout>
+                <FormSchema
+                  model={selectedModelName}
+                  initialValues={getSubModelInitialValues()}
+                  onFormSubmit={handleSubModelSave}
+                  autoSave={false}
+                  onCancel={() => setOpenSubModal(false)}
+                  formSubmitRef={formikRef}
+                  hideButtons={true}
+                />
+              </FormSchemaLayout>
+            )}
+          </DialogContent>
+        </FormSchemaProvider>
+        <DialogActions>
+          <Button onClick={() => setOpenSubModal(false)} color="inherit">
+            {t("common:cancel")}
+          </Button>
+          <Button
+            onClick={() => {
+              if (formikRef.current && formikRef.current.values) {
+                handleSubModelSave(formikRef.current.values);
+              } else {
+                handleSubModelSave(getSubModelInitialValues());
+              }
+            }}
+            variant="contained"
+            color="primary"
+          >
+            {t("common:save")}
+          </Button>
+        </DialogActions>
+      </Dialog>
+    </>
   );
 }
 
 FormSchemaFieldWithParent.propTypes = {
   name: PropTypes.string.isRequired,
   label: PropTypes.string.isRequired,
+  field: PropTypes.shape({
+    value: PropTypes.any,
+    onChange: PropTypes.func.isRequired,
+    error: PropTypes.any,
+  }).isRequired,
   description: PropTypes.string,
   errorMessage: PropTypes.string,
-  field: PropTypes.object.isRequired,
 };
 
 export default FormSchemaFieldWithParent;
diff --git a/DashAI/front/src/components/shared/FormSchemaRenderFields.jsx b/DashAI/front/src/components/shared/FormSchemaRenderFields.jsx
index 3834579fa..db3ba05ce 100644
--- a/DashAI/front/src/components/shared/FormSchemaRenderFields.jsx
+++ b/DashAI/front/src/components/shared/FormSchemaRenderFields.jsx
@@ -56,7 +56,8 @@ function FormSchemaRenderFields({
   onFormSubmit,
   setError,
   errorsMessage,
-  spacing = 2,
+  spacing = 1,
+  excludeFields = [],
 }) {
   if (!modelSchema) return null;
 
@@ -76,6 +77,7 @@ function FormSchemaRenderFields({
     const fields = [];
 
     for (const key in modelSchema) {
+      if (excludeFields.includes(key)) continue;
       const fieldSchema = modelSchema[key];
       const objName = key;
       const value = formik?.values?.[objName];
@@ -192,6 +194,7 @@ FormSchemaRenderFields.propTypes = {
   setError: PropTypes.func,
   errorsMessage: PropTypes.object,
   spacing: PropTypes.number,
+  excludeFields: PropTypes.arrayOf(PropTypes.string),
 };
 
 export default FormSchemaRenderFields;
diff --git a/DashAI/front/src/components/shared/FormSchemaWithSelectedModel.jsx b/DashAI/front/src/components/shared/FormSchemaWithSelectedModel.jsx
index 67357e678..aab49eb3d 100644
--- a/DashAI/front/src/components/shared/FormSchemaWithSelectedModel.jsx
+++ b/DashAI/front/src/components/shared/FormSchemaWithSelectedModel.jsx
@@ -83,6 +83,7 @@ function FormSchemaWithSelectedModel({
         </>
       )}
 
+
       <FormSchema
         model={selectedModel}
         initialValues={defaultValues}
diff --git a/DashAI/front/src/components/shared/Upload.jsx b/DashAI/front/src/components/shared/Upload.jsx
new file mode 100644
index 000000000..78579dbbe
--- /dev/null
+++ b/DashAI/front/src/components/shared/Upload.jsx
@@ -0,0 +1,241 @@
+import React, { useState, useRef } from "react";
+import PropTypes from "prop-types";
+import {
+  Box,
+  Button,
+  CircularProgress,
+  DialogContentText,
+  Grid,
+  IconButton,
+  Paper,
+  Typography,
+} from "@mui/material";
+
+import TextSnippetIcon from "@mui/icons-material/TextSnippet";
+import ClearIcon from "@mui/icons-material/Clear";
+
+/**
+ * Renders a drag and drop to upload a file (dataset).
+ * The upload (send to API) doesn't happen here, this component just adds the file "uploaded" to the
+ * newDataset state in the modal
+ * @param {function} onFileUpload function to handle when the user "uploads" a dataset
+ */
+function Upload({ onFileUpload, emptyUploadText, multiple = false }) {
+  const [EMPTY, LOADING, LOADED] = [0, 1, 2];
+  const [datasetState, setDatasetState] = useState(EMPTY);
+  const [dragActive, setDragActive] = useState(false);
+  const [fileNames, setFileNames] = useState([]);
+  const inputRef = useRef(null);
+
+  const uploadDataset = async (files) => {
+    const fileArray = Array.isArray(files) ? files : Array.from(files);
+    setDatasetState(LOADING);
+    const url = "";
+    await onFileUpload(fileArray, url);
+    setDatasetState(LOADED);
+    setFileNames(fileArray.map((f) => f.name));
+  };
+
+  const handleDrag = (e) => {
+    e.preventDefault();
+    e.stopPropagation();
+    if (datasetState === EMPTY) {
+      if (e.type === "dragenter" || e.type === "dragover") {
+        setDragActive(true);
+      } else if (e.type === "dragleave") {
+        setDragActive(false);
+      }
+    }
+  };
+
+  const handleSelect = (e) => {
+    if (datasetState === EMPTY) {
+      const files = multiple ? e.target.files : [e.target.files[0]];
+      uploadDataset(files);
+    }
+  };
+
+  const handleDrop = (e) => {
+    e.preventDefault();
+    e.stopPropagation();
+    if (datasetState === EMPTY) {
+      setDragActive(false);
+      const files = multiple ? e.dataTransfer.files : [e.dataTransfer.files[0]];
+      if (files && files.length > 0) {
+        uploadDataset(files);
+      }
+    }
+  };
+
+  const handleButtonClick = () => {
+    inputRef.current.click();
+  };
+
+  const handleDeleteDataset = () => {
+    onFileUpload(null, "");
+    setDatasetState(EMPTY);
+    setFileNames([]);
+  };
+
+  const stateContent = (state) => {
+    switch (state) {
+      case EMPTY:
+        return (
+          <React.Fragment>
+            <Grid>
+              <input
+                type="file"
+                ref={inputRef}
+                style={{ display: "none" }}
+                onChange={handleSelect}
+                multiple={multiple}
+              />
+            </Grid>
+            {dragActive ? (
+              <Grid>
+                <Typography variant="subtitle1">
+                  Drop the file{multiple ? "s" : ""} here.
+                </Typography>
+              </Grid>
+            ) : (
+              <React.Fragment>
+                <Grid>
+                  <Typography variant="subtitle1">
+                    Drag and drop {multiple ? "your files" : "a file"} here, or
+                  </Typography>
+                </Grid>
+                <Grid>
+                  <Button variant="contained">Upload a file</Button>
+                </Grid>
+              </React.Fragment>
+            )}
+          </React.Fragment>
+        );
+
+      case LOADING:
+        return <CircularProgress color="inherit" />;
+
+      case LOADED:
+        return (
+          <>
+            <TextSnippetIcon sx={{ fontSize: "58px" }} />
+            <Grid>
+              {fileNames.map((name, index) => (
+                <Typography
+                  key={index}
+                  variant="subtitle1"
+                  sx={{
+                    color: "gray",
+                    textAlign: "center",
+                  }}
+                >
+                  {name}
+                </Typography>
+              ))}
+            </Grid>
+          </>
+        );
+    }
+  };
+
+  return (
+    <Paper
+      sx={{
+        p: 4,
+        borderRadius: 2,
+        display: "flex",
+        flexDirection: "column",
+        minHeight: "400px",
+      }}
+    >
+      {/* state text */}
+      <Box sx={{ textAlign: "center", mb: 2 }}>
+        <DialogContentText>
+          {datasetState === LOADING && "Loading..."}
+          {datasetState === LOADED && "Loaded"}
+          {datasetState === EMPTY && emptyUploadText}
+        </DialogContentText>
+      </Box>
+
+      {/* Drag and drop */}
+      <Box
+        sx={{
+          flex: 1,
+          border: 1,
+          borderRadius: 2,
+          cursor: datasetState === EMPTY ? "pointer" : "auto",
+          borderStyle: "dashed",
+          borderColor: "divider",
+          overflow: "auto",
+          position: "relative",
+          display: "flex",
+          alignItems: "center",
+          justifyContent: "center",
+          minHeight: "300px",
+        }}
+        onClick={datasetState === EMPTY ? handleButtonClick : null}
+        onDragEnter={datasetState === EMPTY ? handleDrag : null}
+        onDragLeave={datasetState === EMPTY ? handleDrag : null}
+        onDragOver={datasetState === EMPTY ? handleDrag : null}
+        onDrop={datasetState === EMPTY ? handleDrop : null}
+      >
+        <input
+          type="file"
+          ref={inputRef}
+          style={{ display: "none" }}
+          onChange={handleSelect}
+          multiple={multiple}
+        />
+
+        <Box sx={{ textAlign: "center", flex: 1, display: "flex", flexDirection: "column", justifyContent: "center", alignItems: "center" }}>
+          {datasetState === EMPTY &&
+            (dragActive ? (
+              <Typography variant="subtitle1">
+                Drop the file{multiple ? "s" : ""} here.
+              </Typography>
+            ) : (
+              <>
+                <Typography variant="subtitle1" sx={{ mb: 2 }}>
+                  Drag and drop {multiple ? "your files" : "a file"} here, or
+                </Typography>
+                <Button variant="contained">Upload a file</Button>
+              </>
+            ))}
+
+          {datasetState === LOADING && <CircularProgress color="inherit" />}
+
+          {datasetState === LOADED && (
+            <>
+              <TextSnippetIcon sx={{ fontSize: "58px", mb: 1 }} />
+              {fileNames.map((name, index) => (
+                <Typography
+                  key={index}
+                  variant="subtitle1"
+                  sx={{
+                    color: "gray",
+                    textAlign: "center",
+                  }}
+                >
+                  {name}
+                </Typography>
+              ))}
+              <IconButton
+                onClick={handleDeleteDataset}
+                sx={{ position: "absolute", right: 0, top: 0 }}
+              >
+                <ClearIcon />
+              </IconButton>
+            </>
+          )}
+        </Box>
+      </Box>
+    </Paper>
+  );
+}
+
+Upload.propTypes = {
+  onFileUpload: PropTypes.func.isRequired,
+  emptyUploadText: PropTypes.oneOfType([PropTypes.string, PropTypes.node]),
+};
+
+export default Upload;
diff --git a/DashAI/front/src/constants/languages.js b/DashAI/front/src/constants/languages.js
new file mode 100644
index 000000000..bed11dcb9
--- /dev/null
+++ b/DashAI/front/src/constants/languages.js
@@ -0,0 +1,5 @@
+/**
+ * Language codes available for RAG prompt templates.
+ * Display names are loaded from i18n locale files (generative:rag.prompt.languages.*).
+ */
+export const LANGUAGE_CODES = ["en", "es", "pt"];
diff --git a/DashAI/front/src/contexts/schema.js b/DashAI/front/src/contexts/schema.js
index ab9e4da5f..e78aac0fc 100644
--- a/DashAI/front/src/contexts/schema.js
+++ b/DashAI/front/src/contexts/schema.js
@@ -19,6 +19,15 @@ import { useSnackbar } from "notistack";
 // Create the FormSchema context
 const FormSchemaContext = createContext();
 
+const fallbackContextValue = {
+  formValues: {},
+  setFormValues: () => {},
+  properties: [],
+  setProperties: () => {},
+  errorForm: false,
+  setErrorForm: () => {},
+};
+
 // Create the FormSchema provider
 export const FormSchemaProvider = ({ children }) => {
   // Define the default values for the form
@@ -59,6 +68,7 @@ FormSchemaProvider.propTypes = {
 
 // Custom hook to obtain the state
 export const useFormSchemaStore = () => {
+  const context = useContext(FormSchemaContext) || fallbackContextValue;
   const {
     formValues,
     setFormValues,
@@ -66,7 +76,7 @@ export const useFormSchemaStore = () => {
     setProperties,
     errorForm,
     ...rest
-  } = useContext(FormSchemaContext);
+  } = context;
 
   const { enqueueSnackbar } = useSnackbar();
 
@@ -76,16 +86,24 @@ export const useFormSchemaStore = () => {
     let formValuesByProperties = { ...formValues };
 
     for (const property of properties) {
+      if (!formValuesByProperties) {
+        return null;
+      }
+
       if (property.key in formValuesByProperties) {
         formValuesByProperties = formValuesByProperties[property.key];
       } else {
-        formValuesByProperties =
-          formValuesByProperties.properties.params.comp.params[property.key];
+        if (formValuesByProperties.properties?.params?.comp?.params) {
+          formValuesByProperties =
+            formValuesByProperties.properties.params.comp.params[property.key];
+        } else {
+          return null;
+        }
       }
     }
 
     return formValuesByProperties;
-  }, [JSON.stringify(formValues), properties]);
+  }, [formValues, properties]);
 
   const handleUpdateSchema = (values, onSubmit) => {
     if (!properties.length) {
@@ -149,21 +167,33 @@ export const useFormSchemaStore = () => {
   };
 
   const getModelFromCurrentProperty = (property) => {
-    if (formValues === null) return null;
+    if (formValues === null || !property) return null;
+
+    //console.log("formValues", formValues);
+    //console.log("properties", properties);
 
-    if (properties.length === 0)
-      return getModelFromSubform(formValues[property]);
+    if (properties.length === 0) {
+      return formValues[property]
+        ? getModelFromSubform(formValues[property])
+        : null;
+    }
 
     let params = null;
     for (const prop of properties) {
       if (params === null) {
         params = formValues[prop.key];
+        if (!params) return null;
         continue;
       }
 
-      params = getParamsFromSubform(params)[prop.key];
+      const propParams = getParamsFromSubform(params);
+      if (!propParams || !propParams[prop.key]) return null;
+      params = propParams[prop.key];
     }
-    return getModelFromSubform(params[property]);
+
+    return params && params[property]
+      ? getModelFromSubform(params[property])
+      : null;
   };
 
   const propertyData = useMemo(() => {
@@ -190,6 +220,7 @@ export const useFormSchemaStore = () => {
 
   return {
     formValues,
+    setFormValues,
     properties,
     propertyData,
     valuesByProperties,
diff --git a/DashAI/front/src/pages/generative/RAG/RAGDocumentsPage.jsx b/DashAI/front/src/pages/generative/RAG/RAGDocumentsPage.jsx
new file mode 100644
index 000000000..71162a31c
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAG/RAGDocumentsPage.jsx
@@ -0,0 +1,159 @@
+import { useState, useEffect, useCallback } from "react";
+import { useNavigate } from "react-router-dom";
+import { useTranslation } from "react-i18next";
+import { Box, Typography } from "@mui/material";
+import ModuleContainer from "../../../components/layout/ModuleContainer";
+import LeftPanel from "../../../components/threeSectionLayout/panels/LeftPanel";
+import CenterPanel from "../../../components/threeSectionLayout/panels/CenterPanel";
+import RightPanel from "../../../components/threeSectionLayout/panels/RightPanel";
+import SessionBar from "../../../components/generative/SessionBar";
+import RAGBreadcrumbs from "../../../components/generative/RAG/RAGBreadcrumbs";
+import DocumentTable from "../../../components/generative/RAG/DocumentTable";
+import {
+  loadDocuments,
+  deleteDocument,
+} from "../../../api/rag";
+import { getSessions, removeSession } from "../../../api/session";
+import { useThreePanelLayout } from "../../../hooks/useThreePanelsLayout";
+import { ThreePanelLayoutContext } from "../../../components/threeSectionLayout/panels/ThreePanelLayoutContext";
+import { FormSchemaProvider } from "../../../contexts/schema";
+
+function RAGDocumentsPage() {
+  const navigate = useNavigate();
+  const threePanelLayout = useThreePanelLayout();
+  const { t } = useTranslation(["generative"]);
+  const [allDocuments, setAllDocuments] = useState([]);
+  const [documentsLoading, setDocumentsLoading] = useState(true);
+  const [sessions, setSessions] = useState([]);
+
+  const loadSessions = useCallback(async () => {
+    try {
+      const allSessions = await getSessions();
+      setSessions(allSessions);
+    } catch (error) {
+      console.error("RAGDocumentsPage: Error loading sessions:", error);
+    }
+  }, []);
+
+  const fetchAllDocuments = useCallback(async () => {
+    setDocumentsLoading(true);
+    try {
+      const docs = await loadDocuments();
+      setAllDocuments(docs);
+    } catch (error) {
+      console.error("RAGDocumentsPage: Error loading all documents:", error);
+    } finally {
+      setDocumentsLoading(false);
+    }
+  }, []);
+
+  useEffect(() => {
+    loadSessions();
+    fetchAllDocuments();
+  }, [loadSessions, fetchAllDocuments]);
+
+  const handleRemoveDocumentFromTable = useCallback(
+    async (id) => {
+      try {
+        await deleteDocument(id);
+        await fetchAllDocuments();
+      } catch (error) {
+        console.error("RAGDocumentsPage: Failed to delete document:", error);
+      }
+    },
+    [fetchAllDocuments],
+  );
+
+  const handleAddDocument = useCallback((newDoc) => {
+    setAllDocuments((prev) => [newDoc, ...prev]);
+  }, []);
+
+  const handleSessionClick = (sessionId, taskName, taskDisplayName) => {
+    navigate("/app/generative", {
+      state: {
+        selectedSessionId: sessionId,
+        selectedTaskName: taskName,
+        selectedDisplayName: taskDisplayName
+      }
+    });
+  };
+
+  const handleNewSessionButton = () => {
+    navigate("/app/generative");
+  };
+
+  const handleSessionDelete = async (id) => {
+    setSessions((prev) => prev.filter((s) => s.id !== id));
+    await removeSession(id);
+  };
+
+  return (
+    <FormSchemaProvider>
+      <ThreePanelLayoutContext.Provider value={threePanelLayout}>
+        <ModuleContainer>
+          <LeftPanel data-tour="sessions-left-panel">
+            <SessionBar
+              sessions={sessions}
+              handleSessionClick={handleSessionClick}
+              handleNewSessionButton={handleNewSessionButton}
+              handleSessionDelete={handleSessionDelete}
+              stepIndex={0}
+              onToggle={threePanelLayout.handleToggleLeft}
+              showSearch={false}
+            />
+          </LeftPanel>
+
+          <CenterPanel data-tour="documents-center-panel">
+            <RAGBreadcrumbs />
+            <Typography variant="h5" component="h1">
+              {t("generative:ragDocumentsPage.title")}
+            </Typography>
+            <Typography variant="subtitle1" component="p" sx={{ mb: 2 }}>
+              {t("generative:ragDocumentsPage.description")}
+            </Typography>
+            <DocumentTable
+              documents={[...allDocuments]
+                .sort(
+                  (a, b) =>
+                    new Date(b.created || b.createdAt || 0) -
+                    new Date(a.created || a.createdAt || 0),
+                )
+                .map((doc) => ({
+                  ...doc,
+                  id: String(doc.id),
+                  name: doc.file_name,
+                  createdAt: doc.created || doc.createdAt || "",
+                  preview: doc.file_url,
+                  file_type: doc.file_name
+                    ? doc.file_name.split(".").pop().toLowerCase()
+                    : "",
+                }))}
+              onRemove={handleRemoveDocumentFromTable}
+              onAddDocument={handleAddDocument}
+              isLoading={documentsLoading}
+              showTableTitle={false}
+            />
+          </CenterPanel>
+
+          <RightPanel toggleButtonTop="50%" data-tour="documents-right-panel">
+            <Box
+              width="100%"
+              height="100%"
+              sx={{
+                backgroundColor: "background.box",
+                borderRadius: 2,
+                minWidth: 0,
+                maxWidth: "100%",
+                overflow: "auto",
+                p: 2,
+              }}
+            >
+            </Box>
+          </RightPanel>
+        </ModuleContainer>
+      </ThreePanelLayoutContext.Provider>
+    </FormSchemaProvider>
+  );
+}
+
+export default RAGDocumentsPage;
diff --git a/DashAI/front/src/pages/generative/RAG/RAGPromptsPage.jsx b/DashAI/front/src/pages/generative/RAG/RAGPromptsPage.jsx
new file mode 100644
index 000000000..6634951cc
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAG/RAGPromptsPage.jsx
@@ -0,0 +1,124 @@
+import { useState, useEffect, useCallback } from "react";
+import { useNavigate } from "react-router-dom";
+import { useTranslation } from "react-i18next";
+import { Box, Typography } from "@mui/material";
+import ModuleContainer from "../../../components/layout/ModuleContainer";
+import LeftPanel from "../../../components/threeSectionLayout/panels/LeftPanel";
+import CenterPanel from "../../../components/threeSectionLayout/panels/CenterPanel";
+import RightPanel from "../../../components/threeSectionLayout/panels/RightPanel";
+import SessionBar from "../../../components/generative/SessionBar";
+import RAGBreadcrumbs from "../../../components/generative/RAG/RAGBreadcrumbs";
+import RAGDocumentsPanel from "../../../components/generative/RAG/RAGDocumentsPanel";
+import PromptSelectionTable from "../../../components/generative/RAG/PromptSelectionTable";
+import { getSessions, removeSession } from "../../../api/session";
+import { useThreePanelLayout } from "../../../hooks/useThreePanelsLayout";
+import { ThreePanelLayoutContext } from "../../../components/threeSectionLayout/panels/ThreePanelLayoutContext";
+import { FormSchemaProvider } from "../../../contexts/schema";
+
+function RAGPromptsPage() {
+  const navigate = useNavigate();
+  const threePanelLayout = useThreePanelLayout();
+  const { t } = useTranslation(["generative"]);
+  const [rowSelectionModel, setRowSelectionModel] = useState([]);
+  const [sessions, setSessions] = useState([]);
+  const [documentRefreshTrigger, setDocumentRefreshTrigger] = useState(0);
+
+  const loadSessions = useCallback(async () => {
+    try {
+      const allSessions = await getSessions();
+      setSessions(allSessions);
+    } catch (error) {
+      console.error("RAGPromptsPage: Error loading sessions:", error);
+    }
+  }, []);
+
+  useEffect(() => {
+    loadSessions();
+  }, [loadSessions]);
+
+  const handleSessionClick = (sessionId, taskName, taskDisplayName) => {
+    navigate("/app/generative", {
+      state: {
+        selectedSessionId: sessionId,
+        selectedTaskName: taskName,
+        selectedDisplayName: taskDisplayName,
+      },
+    });
+  };
+
+  const handleNewSessionButton = () => {
+    navigate("/app/generative");
+  };
+
+  const handleSessionDelete = async (id) => {
+    setSessions((prev) => prev.filter((s) => s.id !== id));
+    await removeSession(id);
+  };
+
+  const handleRowSelectionModelChange = (newSelection) => {
+    setRowSelectionModel(newSelection);
+  };
+
+  const handleDocumentChange = () => {
+    setDocumentRefreshTrigger((prev) => prev + 1);
+  };
+
+  return (
+    <FormSchemaProvider>
+      <ThreePanelLayoutContext.Provider value={threePanelLayout}>
+        <ModuleContainer>
+          <LeftPanel data-tour="sessions-left-panel">
+            <SessionBar
+              sessions={sessions}
+              handleSessionClick={handleSessionClick}
+              handleNewSessionButton={handleNewSessionButton}
+              handleSessionDelete={handleSessionDelete}
+              stepIndex={0}
+              onToggle={threePanelLayout.handleToggleLeft}
+              showSearch={false}
+            />
+          </LeftPanel>
+
+          <CenterPanel data-tour="prompts-center-panel">
+            <RAGBreadcrumbs />
+            <Typography variant="h5" component="h1">
+              {t("generative:ragPromptsPage.title")}
+            </Typography>
+            <Typography variant="subtitle1" component="p" sx={{ mb: 2 }}>
+              {t("generative:ragPromptsPage.description")}
+            </Typography>
+            <PromptSelectionTable
+              showTableTitle={false}
+              loading={false}
+              rowSelectionModel={rowSelectionModel}
+              onRowSelectionModelChange={handleRowSelectionModelChange}
+            />
+          </CenterPanel>
+
+          <RightPanel toggleButtonTop="50%" data-tour="documents-right-panel">
+            <Box
+              width="100%"
+              height="100%"
+              sx={{
+                backgroundColor: "background.box",
+                borderRadius: 2,
+                minWidth: 0,
+                maxWidth: "100%",
+                overflow: "auto",
+                p: 2,
+              }}
+            >
+              <RAGDocumentsPanel
+                selectedSessionId={null}
+                isRagChatActive={false}
+                onDocumentChange={handleDocumentChange}
+              />
+            </Box>
+          </RightPanel>
+        </ModuleContainer>
+      </ThreePanelLayoutContext.Provider>
+    </FormSchemaProvider>
+  );
+}
+
+export default RAGPromptsPage;
diff --git a/DashAI/front/src/pages/generative/RAGSession/RAGSessionPage.jsx b/DashAI/front/src/pages/generative/RAGSession/RAGSessionPage.jsx
new file mode 100644
index 000000000..692fc2ccc
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/RAGSessionPage.jsx
@@ -0,0 +1,247 @@
+import { useState, useEffect, useCallback } from "react";
+import { useNavigate, useParams } from "react-router-dom";
+import { Box } from "@mui/material";
+import ModuleContainer from "../../../components/layout/ModuleContainer";
+import LeftPanel from "../../../components/threeSectionLayout/panels/LeftPanel";
+import CenterPanel from "../../../components/threeSectionLayout/panels/CenterPanel";
+import RightPanel from "../../../components/threeSectionLayout/panels/RightPanel";
+import SessionBar from "../../../components/generative/SessionBar";
+import GenerativeChat from "../../../components/generative/GenerativeChat";
+import RAGSessionSummary from "../../../components/generative/RAG/RAGSessionSummary";
+import { removeSession } from "../../../api/session";
+import { getGenerativeSession } from "../../../api/generativeTask";
+import { useGenerative } from "../../../components/generative/GenerativeContext";
+import { useThreePanelLayout } from "../../../hooks/useThreePanelsLayout";
+import { ThreePanelLayoutContext } from "../../../components/threeSectionLayout/panels/ThreePanelLayoutContext";
+import { FormSchemaProvider } from "../../../contexts/schema";
+import RAGSessionSetup from "./RAGSessionSetup";
+import RAGDocumentsPanel from "../../../components/generative/RAG/RAGDocumentsPanel";
+import RAGInfoBar from "../../../components/generative/RAG/RAGInfoBar";
+import RAGParamsPanel from "../../../components/generative/RAG/RAGParamsPanel";
+
+function RAGSessionPage({ onSessionSelect, sessions, setSessions }) {
+  const navigate = useNavigate();
+  const { id: urlSessionId } = useParams();
+  const threePanelLayout = useThreePanelLayout();
+  const generative = useGenerative() ?? {};
+
+  const [showRagChat, setShowRagChat] = useState(false);
+  const [activeRagChatSessionId, setActiveRagChatSessionId] = useState(null);
+
+  const {
+    sessions: contextSessions,
+    setSessions: setContextSessions,
+    selectedSessionId: globalSelectedSessionId,
+    setSelectedSessionId: setGlobalSelectedSessionId,
+    selectedTaskName,
+    setSelectedTaskName,
+    setSelectedDisplayName,
+    setStepIndex,
+    fetchSessions,
+  } = generative;
+
+  const currentSessions = sessions || contextSessions || [];
+  const currentSetSessions = setSessions || setContextSessions;
+
+  useEffect(() => {
+    if (currentSessions.length === 0 && fetchSessions) {
+      fetchSessions();
+    }
+  }, []);
+
+  const isRagSessionSelected =
+    selectedTaskName === "RAGTask" && Boolean(globalSelectedSessionId);
+  const isRagChatActive = showRagChat && activeRagChatSessionId === globalSelectedSessionId;
+
+  const [setupKey, setSetupKey] = useState(0);
+
+  const resolveSessionId = useCallback(
+    (rawId) => {
+      const num = Number(rawId);
+      return Number.isFinite(num) && num > 0 ? num : null;
+    },
+    [],
+  );
+
+  useEffect(() => {
+    const sid = resolveSessionId(urlSessionId);
+    if (!sid) return;
+    if (sid === globalSelectedSessionId) return;
+
+    const load = async () => {
+      try {
+        const session = await getGenerativeSession(sid);
+        if (!session) return;
+        setGlobalSelectedSessionId?.(sid);
+        setSelectedTaskName?.(session.task_name);
+        setStepIndex?.(0);
+        setShowRagChat(false);
+        setActiveRagChatSessionId(null);
+      } catch {
+        // session not found or error — ignore
+      }
+    };
+    load();
+  }, [urlSessionId, globalSelectedSessionId, resolveSessionId]);
+
+  useEffect(() => {
+    if (!isRagSessionSelected) {
+      setShowRagChat(false);
+      setActiveRagChatSessionId(null);
+    }
+  }, [isRagSessionSelected, globalSelectedSessionId]);
+
+  useEffect(() => {
+    setShowRagChat(false);
+    setActiveRagChatSessionId(null);
+  }, [globalSelectedSessionId]);
+
+  const handleStartRagChat = () => {
+    setShowRagChat(true);
+    setActiveRagChatSessionId(globalSelectedSessionId);
+  };
+
+  const handleSessionClick = useCallback((sessionId, taskName, taskDisplayName) => {
+    if (onSessionSelect) {
+      onSessionSelect(sessionId, taskName, taskDisplayName);
+      return;
+    }
+
+    setGlobalSelectedSessionId?.(sessionId);
+    setSelectedTaskName?.(taskName);
+    setSelectedDisplayName?.(taskDisplayName);
+    setStepIndex?.(0);
+    setShowRagChat(false);
+    setActiveRagChatSessionId(null);
+
+    navigate(`/app/generative/sessions/${sessionId}`, { replace: true });
+  }, [onSessionSelect, navigate, setGlobalSelectedSessionId, setSelectedTaskName, setSelectedDisplayName, setStepIndex]);
+
+  const handleNewSessionButton = async () => {
+    setGlobalSelectedSessionId?.(null);
+    setSelectedTaskName?.("");
+    setSelectedDisplayName?.(null);
+    setStepIndex?.(0);
+    setShowRagChat(false);
+    setActiveRagChatSessionId(null);
+    if (fetchSessions) await fetchSessions();
+    setSetupKey((prev) => prev + 1);
+  };
+
+  const handleSessionDelete = async (id) => {
+    currentSetSessions?.((prev) => prev.filter((s) => s.id !== id));
+    if (id === globalSelectedSessionId) {
+      setGlobalSelectedSessionId?.(null);
+    }
+    await removeSession(id);
+  };
+
+  const handleCloseSetup = () => {
+    setSetupKey((prev) => prev + 1);
+  };
+
+  const handleSessionCreated = (createdSession) => {
+    if (!createdSession?.id) return;
+
+    currentSetSessions?.((prev) => {
+      const nextSessions = Array.isArray(prev) ? prev.slice() : [];
+      const exists = nextSessions.some((session) => session.id === createdSession.id);
+      if (!exists) {
+        nextSessions.unshift(createdSession);
+      }
+      return nextSessions;
+    });
+
+    navigate(`/app/generative/sessions/${createdSession.id}`, { replace: true });
+  };
+
+  const centerContent = isRagSessionSelected ? (
+    isRagChatActive ? (
+      <GenerativeChat />
+    ) : (
+      <RAGSessionSummary
+        sessionId={globalSelectedSessionId}
+        onStartChat={handleStartRagChat}
+      />
+    )
+  ) : (
+    <RAGSessionSetup
+      key={setupKey}
+      onClose={handleCloseSetup}
+      onSessionCreated={handleSessionCreated}
+      existingSessions={currentSessions}
+    />
+  );
+
+  return (
+    <FormSchemaProvider>
+      <ThreePanelLayoutContext.Provider value={threePanelLayout}>
+        <ModuleContainer>
+          <LeftPanel data-tour="sessions-left-panel">
+            {isRagSessionSelected ? (
+              <Box sx={{ display: "flex", flexDirection: "column", height: "100%", gap: 1 }}>
+                <Box sx={{ flex: "0 0 60%", minHeight: 0 }}>
+                  <RAGDocumentsPanel
+                    selectedSessionId={globalSelectedSessionId}
+                    isRagChatActive={isRagChatActive}
+                  />
+                </Box>
+
+                <Box sx={{ flex: "0 0 40%", overflow: "auto", minHeight: 0 }}>
+                  <SessionBar
+                    sessions={currentSessions}
+                    selectedSessionId={globalSelectedSessionId}
+                    handleSessionClick={handleSessionClick}
+                    handleNewSessionButton={handleNewSessionButton}
+                    handleSessionDelete={handleSessionDelete}
+                    stepIndex={0}
+                    onToggle={threePanelLayout.handleToggleLeft}
+                    showSearch={false}
+                  />
+                </Box>
+              </Box>
+            ) : (
+              <SessionBar
+                sessions={currentSessions}
+                selectedSessionId={globalSelectedSessionId}
+                handleSessionClick={handleSessionClick}
+                handleNewSessionButton={handleNewSessionButton}
+                handleSessionDelete={handleSessionDelete}
+                stepIndex={0}
+                onToggle={threePanelLayout.handleToggleLeft}
+                showSearch={false}
+              />
+            )}
+          </LeftPanel>
+
+          <CenterPanel data-tour="task-gallery">
+            {centerContent}
+          </CenterPanel>
+
+          <RightPanel toggleButtonTop="50%" data-tour="parameters-right-panel">
+            <Box
+              width="100%"
+              height="100%"
+              sx={{
+                backgroundColor: "background.box",
+                borderRadius: 2,
+                minWidth: 0,
+                maxWidth: "100%",
+                overflow: "auto",
+                p: 2,
+              }}
+            >
+              {isRagSessionSelected ? (
+                <RAGParamsPanel selectedSessionId={globalSelectedSessionId} />
+              ) : (
+                <RAGInfoBar />
+              )}
+            </Box>
+          </RightPanel>
+        </ModuleContainer>
+      </ThreePanelLayoutContext.Provider>
+    </FormSchemaProvider>
+  );
+}
+
+export default RAGSessionPage;
diff --git a/DashAI/front/src/pages/generative/RAGSession/RAGSessionSetup.jsx b/DashAI/front/src/pages/generative/RAGSession/RAGSessionSetup.jsx
new file mode 100644
index 000000000..b4eaf3106
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/RAGSessionSetup.jsx
@@ -0,0 +1,500 @@
+import { useEffect, useMemo, useState, useCallback } from "react";
+import {
+  Box,
+  Button,
+  TextField,
+  Typography,
+  CircularProgress,
+  Alert,
+} from "@mui/material";
+import { ViewList as ViewListIcon } from "@mui/icons-material";
+import { useNavigate } from "react-router-dom";
+import DocumentSelector from "../../../components/generative/RAG/DocumentSelector";
+import ChunkingSection from "./sections/ChunkingSection";
+import RetrieverSection from "./sections/RetrieverSection";
+import GeneratorSection from "./sections/GeneratorSection";
+import PromptSection from "./sections/PromptSection";
+import RAGCard from "./components/RAGCard";
+import SectionCard from "./components/SectionCard";
+import { useSnackbar } from "notistack";
+import { useTranslation } from "react-i18next";
+import { createRAGSession } from "../../../api/rag";
+import { getSessions } from "../../../api/session";
+import { generateSequentialName } from "../../../utils/nameGenerator";
+import RAGSectionColumn from "./components/RAGSectionColumn";
+
+const defaultSessionData = {
+  name: "",
+  description: "",
+  documents: [],
+  parameters: {
+    chunking_model: {
+      component: "",
+      params: {},
+    },
+    retriever_model: {
+      component: "",
+      params: {},
+    },
+    generator_model: {
+      component: "",
+      params: {},
+    },
+    prompt: {
+      component: "",
+      params: {},
+    },
+  },
+};
+
+export default function RAGSessionSetup({
+  initialData,
+  onClose,
+  onSessionCreated,
+  existingSessions = [],
+}) {
+  const { enqueueSnackbar } = useSnackbar();
+  const { t } = useTranslation(["generative"]);
+  const navigate = useNavigate();
+  const goToPromptsDetail = () => navigate("/app/generative/RAG/prompts");
+  const goToDocumentsDetail = () => navigate("/app/generative/RAG/documents");
+
+  const [suggestedName, setSuggestedName] = useState(() => {
+    const sessionsList = Array.isArray(existingSessions) ? existingSessions : [];
+    const ragSessions = sessionsList.filter(s => s?.task_name === "RAGTask");
+    const { defaultName } = generateSequentialName({
+      base: "RAG_Session",
+      items: ragSessions,
+      getName: (session) => session?.name,
+    });
+    return defaultName || "RAG_Session_1";
+  });
+
+  useEffect(() => {
+    let cancelled = false;
+    getSessions().then((allSessions) => {
+      if (cancelled) return;
+      const ragSessions = (allSessions || []).filter(s => s?.task_name === "RAGTask");
+      const { defaultName } = generateSequentialName({
+        base: "RAG_Session",
+        items: ragSessions,
+        getName: (session) => session?.name,
+      });
+      setSuggestedName(defaultName || "RAG_Session_1");
+    }).catch(() => {});
+    return () => { cancelled = true; };
+  }, []);
+
+  const [isNameTouched, setIsNameTouched] = useState(Boolean(initialData?.name));
+  const [sessionData, setSessionData] = useState({
+    ...defaultSessionData,
+    name: initialData?.name || suggestedName,
+    description: initialData?.description || "",
+    documents: initialData?.documents || [],
+  });
+  const [saving, setSaving] = useState(false);
+  const [nameError, setNameError] = useState("");
+  const [isDuplicateName, setIsDuplicateName] = useState(false);
+  const [expandedSections, setExpandedSections] = useState({
+    sessionDetails: true,
+    documents: true,
+    chunking: true,
+    retriever: true,
+    generator: true,
+    prompt: true,
+  });
+
+  const [isGeneratorValidState, setIsGeneratorValidState] = useState(false);
+
+  let chunkSize = sessionData.parameters.chunking_model?.params?.chunk_size || 0;
+  // If chunking model is character based, use 1 token = 4 characters as a rough estimate
+  if (sessionData.parameters.chunking_model?.component?.toLowerCase().includes("character")) {
+    chunkSize = Math.ceil(chunkSize / 4);
+  }
+  const topK = sessionData.parameters.retriever_model?.params?.top_k || 0;
+
+  const [promptTokenCount, setPromptTokenCount] = useState(0);
+
+  useEffect(() => {
+    if (initialData?.name) return;
+    if (isNameTouched) return;
+    setSessionData((prev) => ({ ...prev, name: suggestedName }));
+  }, [initialData?.name, isNameTouched, suggestedName]);
+
+  useEffect(() => {
+    const trimmedValue = sessionData.name.trim().toLowerCase();
+    if (!trimmedValue) {
+      setIsDuplicateName(false);
+      return;
+    }
+
+    const sessionsList = Array.isArray(existingSessions) ? existingSessions : [];
+    const isDuplicate = sessionsList.some(
+      (session) => session?.name?.toLowerCase() === trimmedValue
+    );
+    setIsDuplicateName(isDuplicate);
+  }, [sessionData.name, existingSessions]);
+
+  const handleSessionNameChange = (event) => {
+    const value = event.target.value;
+    setIsNameTouched(true);
+    setSessionData((prev) => ({
+      ...prev,
+      name: value,
+    }));
+
+    if (value.trim() === "") {
+      setNameError(t("generative:rag.validation.nameRequired"));
+    } else {
+      setNameError("");
+    }
+  };
+
+  const handleSessionDescriptionChange = (event) => {
+    const value = event.target.value;
+    setSessionData((prev) => ({
+      ...prev,
+      description: value,
+    }));
+  };
+
+  const handleDocumentSelectionChange = useCallback((selectedDocs) => {
+    setSessionData((prev) => ({
+      ...prev,
+      documents: selectedDocs.map((doc) => doc.id),
+    }));
+  }, []);
+
+  const updateChunkingModel = (model) => {
+    setSessionData((prev) => ({
+      ...prev,
+      parameters: {
+        ...prev.parameters,
+        chunking_model: { ...model },
+      },
+    }));
+  };
+
+  const updateRetrieverModel = (model) => {
+    setSessionData((prev) => ({
+      ...prev,
+      parameters: {
+        ...prev.parameters,
+        retriever_model: { ...model },
+      },
+    }));
+  };
+
+  const updateGeneratorModel = (model) => {
+    setSessionData((prev) => ({
+      ...prev,
+      parameters: {
+        ...prev.parameters,
+        generator_model: { ...model },
+      },
+    }));
+  };
+
+  const updatePrompt = (prompt) => {
+    setSessionData((prev) => ({
+      ...prev,
+      parameters: {
+        ...prev.parameters,
+        prompt,
+      },
+    }));
+  };
+
+  const handleSectionChange = (section) => (event, isExpanded) => {
+    setExpandedSections((prev) => ({
+      ...prev,
+      [section]: isExpanded,
+    }));
+  };
+
+  const validateConfiguration = () => {
+    if (!sessionData.name.trim()) {
+      enqueueSnackbar(t("generative:rag.validation.nameRequired"), { variant: "warning" });
+      return false;
+    }
+    if (isDuplicateName) {
+      enqueueSnackbar(t("generative:rag.validation.nameUnique"), { variant: "warning" });
+      return false;
+    }
+    if (sessionData.documents.length === 0) {
+      enqueueSnackbar(t("generative:rag.validation.documentsRequired"), {
+        variant: "warning",
+      });
+      return false;
+    }
+    if (!sessionData.parameters.chunking_model?.component) {
+      enqueueSnackbar(t("generative:rag.validation.chunkingRequired"), { variant: "warning" });
+      return false;
+    }
+    if (!sessionData.parameters.retriever_model?.component) {
+      enqueueSnackbar(t("generative:rag.validation.retrieverRequired"), { variant: "warning" });
+      return false;
+    }
+    if (!sessionData.parameters.generator_model?.component) {
+      enqueueSnackbar(t("generative:rag.validation.generatorRequired"), { variant: "warning" });
+      return false;
+    }
+    if (!isGeneratorValidState) {
+      enqueueSnackbar(t("generative:rag.validation.generatorInvalid"), { variant: "error" });
+      return false;
+    }
+    if (!sessionData.parameters.prompt?.component) {
+      enqueueSnackbar(t("generative:rag.validation.promptRequired"), { variant: "warning" });
+      return false;
+    }
+    return true;
+  };
+
+  const isConfigurationComplete = useMemo(() => {
+    const isNameValid = Boolean(sessionData.name?.trim()) && !isDuplicateName;
+    const areDocsValid = Array.isArray(sessionData.documents) && sessionData.documents.length > 0;
+    const isChunkingValid = Boolean(sessionData.parameters.chunking_model?.component);
+    const isRetrieverValid = Boolean(sessionData.parameters.retriever_model?.component);
+    const isGeneratorSelected = Boolean(sessionData.parameters.generator_model?.component);
+    const isPromptValid = sessionData.parameters.prompt?.component;
+
+    return isNameValid && areDocsValid && isChunkingValid && isRetrieverValid && isGeneratorValidState && isGeneratorSelected && isPromptValid;
+  }, [sessionData, isGeneratorValidState, isDuplicateName]);
+
+  const handleSave = async () => {
+    if (!validateConfiguration()) {
+      return;
+    }
+
+    setSaving(true);
+    try {
+      const finalSessionData = {
+        name: sessionData.name.trim(),
+        description: sessionData.description.trim(),
+        task_name: "RAGTask",
+        model_name: "RAGPipeline",
+        parameters: {
+          documents: sessionData.documents,
+          chunking_model: sessionData.parameters.chunking_model,
+          retriever_model: sessionData.parameters.retriever_model,
+          generation_model: sessionData.parameters.generator_model,
+          prompt: sessionData.parameters.prompt,
+        },
+      };
+
+      const createdSession = await createRAGSession(finalSessionData);
+      enqueueSnackbar(t("generative:rag.messages.success"), { variant: "success" });
+
+      if (onSessionCreated) {
+        onSessionCreated(createdSession);
+      } else {
+        onClose?.();
+      }
+    } catch (error) {
+      console.error("Error creating RAG session:", error);
+      enqueueSnackbar(
+        error.message || t("generative:error.failedToCreateSession"),
+        { variant: "error" }
+      );
+    } finally {
+      setSaving(false);
+    }
+  };
+
+  return (
+      <Box
+        width="100%"
+        height="100%"
+        display="flex"
+        flexDirection="column"
+      >
+        <Box flex={1} overflow="auto" sx={{display: "flex", flexDirection: "column", gap: 3}}>
+
+          <RAGSectionColumn>
+            <Typography variant="h5" component="h3">
+              {t("generative:rag.setup.title")}
+            </Typography>
+            <Typography variant="body2" color="textSecondary">
+              {t("generative:rag.setup.subtitle")}
+            </Typography>
+          </RAGSectionColumn>
+
+
+          <RAGCard
+            title={t("generative:rag.setup.sessionDetails")}
+            expanded={expandedSections.sessionDetails}
+            onChange={handleSectionChange("sessionDetails")}
+          >
+            <SectionCard>
+              <RAGSectionColumn>
+                <TextField
+                  fullWidth
+                  label={t("generative:rag.setup.sessionName")}
+                  variant="outlined"
+                  value={sessionData.name}
+                  onChange={handleSessionNameChange}
+                  placeholder={t("generative:rag.setup.sessionNamePlaceholder")}
+                  error={Boolean(nameError) || isDuplicateName}
+                  helperText={nameError || (isDuplicateName ? t("generative:rag.setup.sessionNameDuplicate") : "")}
+                  size="medium"
+                  disabled={saving}
+                  />
+
+                {isDuplicateName && (
+                  <Alert
+                  severity="warning"
+                  sx={{
+                    p: 2,
+                    backgroundColor: "action.hover",
+                    border: "1px solid",
+                    borderColor: "warning.main",
+                    borderRadius: 1,
+                    display: "flex",
+                    alignItems: "center",
+                    gap: 1,
+                  }}
+                  >
+                    <Typography variant="body2">
+                      {t("generative:rag.setup.sessionNameDuplicateAlert")}
+                    </Typography>
+                  </Alert>
+                )}
+
+                <TextField
+                  fullWidth
+                  label={t("generative:rag.setup.description")}
+                  variant="outlined"
+                  value={sessionData.description}
+                  onChange={handleSessionDescriptionChange}
+                  placeholder={t("generative:rag.setup.descriptionPlaceholder")}
+                  multiline
+                  rows={3}
+                  size="medium"
+                  disabled={saving}
+                  />
+              </RAGSectionColumn>
+            </SectionCard>
+          </RAGCard>
+
+          <RAGCard
+            title={t("generative:rag.setup.selectDocuments")}
+            expanded={expandedSections.documents}
+            onChange={handleSectionChange("documents")}
+            actions={[
+              {
+                icon: <ViewListIcon fontSize="small" />,
+                tooltip: t("generative:rag.setup.openDocumentsLibrary"),
+                onClick: goToDocumentsDetail,
+                ariaLabel: "open-documents-library",
+              },
+            ]}
+          >
+            <SectionCard>
+              <DocumentSelector
+                selectedIds={sessionData.documents}
+                onSelect={handleDocumentSelectionChange}
+              />
+            </SectionCard>
+          </RAGCard>
+
+          <RAGCard
+            title={t("generative:rag.setup.chunkingStrategy")}
+            expanded={expandedSections.chunking}
+            onChange={handleSectionChange("chunking")}
+          >
+            <SectionCard>  
+              <ChunkingSection
+                chunkingModel={sessionData.parameters.chunking_model}
+                setChunkingModel={updateChunkingModel}
+                />
+            </SectionCard>
+          </RAGCard>
+
+          <RAGCard
+            title={t("generative:rag.setup.retrieverModel")}
+            expanded={expandedSections.retriever}
+            onChange={handleSectionChange("retriever")}
+          >
+            <SectionCard>
+              <RetrieverSection
+                retrieverModel={sessionData.parameters.retriever_model}
+                setRetrieverModel={updateRetrieverModel}
+              />
+
+            </SectionCard>
+          </RAGCard>
+
+          <RAGCard
+            title={t("generative:rag.setup.promptTemplate")}
+            expanded={expandedSections.prompt}
+            onChange={handleSectionChange("prompt")}
+            actions={[
+              {
+                icon: <ViewListIcon fontSize="small" />,
+                tooltip: t("generative:rag.prompt.openPrompts"),
+                onClick: goToPromptsDetail,
+                ariaLabel: "open-prompt-library",
+              },
+            ]}
+          >
+            <SectionCard>
+              <PromptSection
+                promptModel={sessionData.parameters.prompt}
+                setPromptModel={updatePrompt}
+                onTokenCountChange={setPromptTokenCount}
+                />
+            </SectionCard>
+          </RAGCard>
+
+          <RAGCard
+            title={t("generative:rag.setup.languageModel")}
+            expanded={expandedSections.generator}
+            onChange={handleSectionChange("generator")}
+          >
+            <SectionCard>
+              <GeneratorSection
+                generatorModel={sessionData.parameters.generator_model}
+                setGeneratorModel={updateGeneratorModel}
+                chunkSize={chunkSize}
+                topK={topK}
+                promptTokenCount={promptTokenCount}
+                setIsValid={setIsGeneratorValidState}
+              />
+            </SectionCard>
+          </RAGCard>
+
+        </Box>
+
+        <Box
+          flexShrink={0}
+          display="flex"
+          justifyContent="flex-end"
+          gap={2}
+          sx={{
+            pt: 2,
+            pb: 0,
+            borderTop: "1px solid",
+            borderColor: "divider",
+          }}
+        >
+          <Button
+            variant="outlined"
+            color="inherit"
+            onClick={onClose}
+            disabled={saving}
+          >
+            {t("generative:rag.setup.cancel")}
+          </Button>
+          <Button
+            variant="contained"
+            color="primary"
+            onClick={handleSave}
+            disabled={saving || !isConfigurationComplete}
+            sx={{ minWidth: 120 }}
+          >
+            {saving ? <CircularProgress size={20} /> : t("generative:rag.setup.saveSession")}
+          </Button>
+        </Box>
+      </Box>
+    );
+}
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingAdvancedModal.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingAdvancedModal.jsx
new file mode 100644
index 000000000..b42aa131c
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingAdvancedModal.jsx
@@ -0,0 +1,88 @@
+import { useState } from "react";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  IconButton,
+} from "@mui/material";
+import CloseIcon from "@mui/icons-material/Close";
+import { useTranslation } from "react-i18next";
+import { FormSchemaProvider } from "../../../../contexts/schema";
+import ChunkingConfigurationStep from "./ChunkingConfigurationStep";
+import { getModelFromSubform } from "../../../../utils/schema";
+
+export default function ChunkingAdvancedModal({
+  open,
+  onClose,
+  chunkingModel,
+  setChunkingModel,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [stepValid, setStepValid] = useState(false);
+  const modelName = getModelFromSubform(chunkingModel);
+
+  const handleClose = () => {
+    onClose();
+  };
+
+  const handleSave = () => {
+    onClose();
+  };
+
+  return (
+    <Dialog
+      open={open}
+      onClose={handleClose}
+      maxWidth="md"
+      fullWidth
+      PaperProps={{
+        sx: { minHeight: "500px" },
+      }}
+    >
+      <DialogTitle
+        sx={{
+          bgcolor: "background.paper",
+          display: "flex",
+          justifyContent: "space-between",
+          alignItems: "center",
+        }}
+      >
+        {t("generative:rag.advanced.chunkingTitle")}
+        <IconButton
+          onClick={handleClose}
+          size="small"
+          sx={{ color: "text.secondary" }}
+        >
+          <CloseIcon />
+        </IconButton>
+      </DialogTitle>
+
+      <DialogContent dividers sx={{ bgcolor: "background.paper", minHeight: 400 }}>
+        <FormSchemaProvider key={`chunking-advanced-${modelName}`}>
+          <ChunkingConfigurationStep
+            chunkingModel={chunkingModel}
+            setChunkingModel={setChunkingModel}
+            setNextEnabled={setStepValid}
+          />
+        </FormSchemaProvider>
+      </DialogContent>
+
+      <DialogActions sx={{ p: 2, bgcolor: "background.paper" }}>
+        <Button onClick={handleClose} variant="outlined">
+          {t("generative:rag.advanced.close")}
+        </Button>
+        <Button
+          onClick={handleSave}
+          variant="contained"
+          color="primary"
+          disabled={!stepValid}
+        >
+          {t("generative:rag.advanced.done")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingConfigurationStep.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingConfigurationStep.jsx
new file mode 100644
index 000000000..fa3e58cb4
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/ChunkingConfigurationStep.jsx
@@ -0,0 +1,181 @@
+import React, { useState, useEffect, useMemo } from "react";
+import {
+  Box,
+  Autocomplete,
+  TextField,
+  Typography,
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getChunkingComponents } from "../../../../api/rag";
+import FormSchema from "../../../../components/shared/FormSchema";
+import FormSchemaContainer from "../../../../components/shared/FormSchemaContainer";
+import { resolveDefaults, getModelFromSubform, getParamsFromSubform } from "../../../../utils/schema";
+
+export default function ChunkingConfigurationStep({
+  chunkingModel,
+  setChunkingModel,
+  setNextEnabled,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [chunkingOptions, setChunkingOptions] = useState([]);
+  const [selectedChunking, setSelectedChunking] = useState(null);
+  const [error, setError] = useState(null);
+
+  const [fetchedDefaults, setFetchedDefaults] = useState(null);
+
+  useEffect(() => {
+    if (!selectedChunking) return;
+    let cancelled = false;
+    (async () => {
+      const d = await resolveDefaults(selectedChunking.name);
+      if (!cancelled) setFetchedDefaults(d);
+    })();
+    return () => { cancelled = true; };
+  }, [selectedChunking]);
+
+  const formInitialValues = useMemo(
+    () => {
+      if (selectedChunking) {
+        const existingModelName = getModelFromSubform(chunkingModel);
+        const existingParams = getParamsFromSubform(chunkingModel) ?? chunkingModel?.params;
+        if (
+          existingModelName === selectedChunking.name &&
+          existingParams &&
+          Object.keys(existingParams).length > 0
+        ) {
+          return existingParams;
+        }
+      }
+      return fetchedDefaults || {};
+    },
+    [selectedChunking, chunkingModel, fetchedDefaults],
+  );
+
+  useEffect(() => {
+    let isMounted = true;
+    const fetchChunkingModels = async () => {
+      const data = await getChunkingComponents();
+      if (isMounted) {
+        setChunkingOptions(data || []);
+        console.log("[ChunkingConfigStep] Fetched chunking options:", data);
+      }
+    };
+    fetchChunkingModels();
+    return () => {
+      isMounted = false;
+    };
+  }, []);
+
+  useEffect(() => {
+    if (!chunkingOptions.length) return;
+    const modelName = getModelFromSubform(chunkingModel);
+    console.log("[ChunkingConfigStep] Syncing selectedChunking:", {
+      modelName,
+      chunkingModel,
+      chunkingOptions: chunkingOptions.map(c => c.name),
+    });
+    if (modelName) {
+      const found = chunkingOptions.find(
+        (c) => c.name === modelName,
+      );
+      if (found) {
+        console.log("[ChunkingConfigStep] Found chunking model:", found.name);
+        setSelectedChunking(found);
+        setNextEnabled(true);
+      }
+    } else {
+      setNextEnabled(false);
+    }
+  }, [chunkingOptions, chunkingModel]);
+
+  const handleChunkingSelectionChange = async (event, newValue) => {
+    setSelectedChunking(newValue);
+    setError(null);
+
+    if (newValue) {
+      const defaults = await resolveDefaults(newValue.name);
+      setChunkingModel({
+        component: newValue.name,
+        params: defaults,
+      });
+      setNextEnabled(true);
+    } else {
+      setChunkingModel({ component: "", params: {} });
+      setNextEnabled(false);
+    }
+  };
+
+  const handleFormSubmit = (values) => {
+    if (
+      values.chunk_size !== undefined &&
+      values.chunk_overlap !== undefined &&
+      values.chunk_overlap >= values.chunk_size
+    ) {
+      const errorMsg =
+        "chunk_overlap must be less than chunk_size. " +
+        `Got chunk_overlap=${values.chunk_overlap} and chunk_size=${values.chunk_size}`;
+      setError(errorMsg);
+      setNextEnabled(false);
+      return;
+    }
+
+    setError(null);
+    setChunkingModel({
+      component: selectedChunking.name,
+      params: values,
+    });
+    setNextEnabled(true);
+  };
+
+  return (
+    <Box sx={{ display: "flex", flexDirection: "column", gap: 3 }}>
+      <Typography variant="subtitle2" sx={{ mb: 0 }}>
+        {t("generative:rag.chunkingConfig.configureTitle")}
+      </Typography>
+
+      <Autocomplete
+        disablePortal
+        options={chunkingOptions}
+        getOptionLabel={(option) => option.name}
+        value={selectedChunking}
+        onChange={handleChunkingSelectionChange}
+        isOptionEqualToValue={(option, value) => option.name === value?.name}
+        renderInput={(params) => (
+          <TextField {...params} label={t("generative:rag.chunkingConfig.modelLabel")} />
+        )}
+      />
+
+      {error && (
+        <Typography variant="body2" color="error" sx={{ px: 1 }}>
+          {error}
+        </Typography>
+      )}
+
+      {selectedChunking && (
+        <FormSchemaContainer key={`chunking-form-${selectedChunking.name}`}>
+          <FormSchema
+            autoSave
+            model={selectedChunking.name}
+            initialValues={formInitialValues}
+            onFormSubmit={handleFormSubmit}
+            setError={(err) => {
+              if (err) {
+                console.error("FormSchema error:", err);
+                setError(err?.message || "Validation error");
+                setNextEnabled(false);
+              }
+            }}
+            hideButtons
+          />
+        </FormSchemaContainer>
+      )}
+    </Box>
+  );
+}
+
+ChunkingConfigurationStep.propTypes = {
+  chunkingModel: PropTypes.object,
+  setChunkingModel: PropTypes.func.isRequired,
+  setNextEnabled: PropTypes.func,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/CompositeRetrieverBuilder.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/CompositeRetrieverBuilder.jsx
new file mode 100644
index 000000000..3b0fa89f4
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/CompositeRetrieverBuilder.jsx
@@ -0,0 +1,418 @@
+import { useState, useEffect, useCallback } from "react";
+import {
+  Box,
+  Typography,
+  IconButton,
+  Tooltip,
+} from "@mui/material";
+import { useTheme } from "@mui/material/styles";
+import AddCircleOutlineIcon from "@mui/icons-material/AddCircleOutline";
+import RemoveCircleOutlineIcon from "@mui/icons-material/RemoveCircleOutline";
+import PsychologyIcon from "@mui/icons-material/Psychology";
+import AccountTreeIcon from "@mui/icons-material/AccountTree";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getRetrieverComponents, getRetrievalParadigm } from "../../../../api/rag";
+import RetrieverNodeConfig from "./RetrieverNodeConfig";
+
+const COMPOSITE_TYPES = ["SequentialRetriever", "ParallelRetriever", "MMRRerankerRetriever"];
+let _nodeIdCounter = 0;
+function nextId() {
+  return `n_${++_nodeIdCounter}`;
+}
+
+export default function CompositeRetrieverBuilder({
+  rootComponent,
+  rootParams,
+  onChange,
+}) {
+  const theme = useTheme();
+  const { t } = useTranslation(["generative"]);
+  const [allComponents, setAllComponents] = useState([]);
+  const [leafRegistry, setLeafRegistry] = useState({});
+  const [editing, setEditing] = useState(null);
+  const [tree, setTree] = useState(null);
+
+  const [ready, setReady] = useState(false);
+
+  useEffect(() => {
+    (async () => {
+      const paradigms = await getRetrievalParadigm();
+      const leaves = paradigms.filter((p) => !COMPOSITE_TYPES.includes(p.name));
+      const leafMap = {};
+      for (const leaf of leaves) {
+        const children = await getRetrieverComponents(leaf.name);
+        leafMap[leaf.name] = (children || []).filter((c) => c.configurable_object !== false);
+      }
+      setAllComponents(paradigms);
+      setLeafRegistry(leafMap);
+      setReady(true);
+    })();
+  }, []);
+
+  useEffect(() => {
+    if (!ready) return;
+
+    _nodeIdCounter = 0;
+    const fromParams = rootParams?.children || [];
+
+    const buildFromParams = (child) => {
+      const id = nextId();
+      const comp = findComponent(child.component);
+      const node = { nodeId: id, component: child.component, params: child.params || {}, children: [] };
+      if (COMPOSITE_TYPES.includes(child.component) && child.children?.length) {
+        node.children = child.children.map(buildFromParams).filter(Boolean);
+      }
+      return comp ? node : null;
+    };
+
+    const existingChildren = fromParams.map(buildFromParams).filter(Boolean);
+
+    setTree({
+      nodeId: nextId(),
+      component: rootComponent,
+      params: rootParams || {},
+      children: existingChildren,
+    });
+  }, [ready, rootComponent]);
+
+  const findComponent = (name) => {
+    const direct = allComponents.find((c) => c.name === name);
+    if (direct) return direct;
+    for (const key of Object.keys(leafRegistry)) {
+      const found = (leafRegistry[key] || []).find((c) => c.name === name);
+      if (found) return found;
+    }
+    return null;
+  };
+
+  const emit = useCallback(
+    (t) => {
+      const ser = (n) => {
+        const s = { component: n.component, params: n.params || {} };
+        if (COMPOSITE_TYPES.includes(n.component) && n.children.length > 0) {
+          s.children = n.children.map(ser);
+        }
+        return s;
+      };
+      onChange({
+        component: t.component,
+        params: { ...t.params, children: t.children.map(ser) },
+      });
+    },
+    [onChange],
+  );
+
+  const updateAt = (root, nodeId, fn) => {
+    const walk = (n) => {
+      if (n.nodeId === nodeId) return fn(n);
+      if (n.children) return { ...n, children: n.children.map(walk) };
+      return n;
+    };
+    return walk(root);
+  };
+
+  const handleSaveNode = (nodeId, component, params) => {
+    const isComposite = COMPOSITE_TYPES.includes(component);
+    setTree((prev) => {
+      const updated = updateAt(prev, nodeId, (n) => ({
+        ...n,
+        component,
+        params: { ...params },
+        children: isComposite ? n.children : [],
+      }));
+      emit(updated);
+      return updated;
+    });
+    setEditing(null);
+  };
+
+  const handleAddChild = (parentId) => {
+    const id = nextId();
+    setTree((prev) => {
+      const updated = updateAt(prev, parentId, (n) => ({
+        ...n,
+        children: [...n.children, { nodeId: id, component: "", params: {}, children: [] }],
+      }));
+      return updated;
+    });
+    setEditing(id);
+  };
+
+  const handleRemoveChild = (parentId, childId) => {
+    setTree((prev) => {
+      const updated = updateAt(prev, parentId, (n) => ({
+        ...n,
+        children: n.children.filter((c) => c.nodeId !== childId),
+      }));
+      emit(updated);
+      return updated;
+    });
+  };
+
+  if (!tree) {
+    return (
+      <Box sx={{ py: 4, textAlign: "center" }}>
+        <Typography variant="body2" color="text.secondary">
+          {t("generative:rag.common.loading")}
+        </Typography>
+      </Box>
+    );
+  }
+
+  return (
+    <Box sx={{ display: "flex", flexDirection: "row", gap: 2, alignItems: "flex-start" }}>
+      <Box
+        sx={{
+          display: "flex",
+          flexDirection: "column",
+          alignItems: "center",
+          gap: 0.5,
+          mt: 1,
+          minWidth: 90,
+          flexShrink: 0,
+        }}
+      >
+        <Typography
+          variant="caption"
+          sx={{
+            color: "text.secondary",
+            textAlign: "center",
+            lineHeight: 1.2,
+          }}
+        >
+          {t("generative:rag.composite.allChunks")}
+        </Typography>
+
+        <Box
+          sx={{
+            display: "flex",
+            flexDirection: "column",
+            alignItems: "center",
+            my: 1,
+          }}
+        >
+          <Box
+            sx={{
+              width: 0,
+              height: 0,
+              borderLeft: "14px solid transparent",
+              borderRight: "14px solid transparent",
+              borderTop: `20px solid ${theme.palette.divider}`,
+            }}
+          />
+          <Box sx={{ width: 2, height: 24, bgcolor: "divider" }} />
+          <Box
+            sx={{
+              width: 0,
+              height: 0,
+              borderLeft: "14px solid transparent",
+              borderRight: "14px solid transparent",
+              borderTop: `20px solid ${theme.palette.divider}`,
+            }}
+          />
+          <Box sx={{ width: 2, height: 24, bgcolor: "divider" }} />
+          <Box
+            sx={{
+              width: 0,
+              height: 0,
+              borderLeft: "14px solid transparent",
+              borderRight: "14px solid transparent",
+              borderTop: `20px solid ${theme.palette.divider}`,
+            }}
+          />
+        </Box>
+
+        <Typography
+          variant="caption"
+          sx={{
+            color: "text.secondary",
+            textAlign: "center",
+            lineHeight: 1.2,
+          }}
+        >
+          {t("generative:rag.composite.selectedChunks")}
+        </Typography>
+      </Box>
+
+      <Box sx={{ flex: 1, minWidth: 0 }}>
+        <TreeNodeView
+          node={tree}
+          depth={0}
+          isRoot
+          parentId={null}
+          findComponent={findComponent}
+          onEdit={(id) => setEditing(id)}
+          onAddChild={handleAddChild}
+          onRemoveChild={handleRemoveChild}
+          theme={theme}
+          t={t}
+        />
+      </Box>
+
+      {editing && (
+        <RetrieverNodeConfig
+          open
+          nodeId={editing}
+          nodeData={findNodeData(tree, editing)}
+          allComponents={allComponents}
+          leafRegistry={leafRegistry}
+          onSave={handleSaveNode}
+          onClose={() => setEditing(null)}
+        />
+      )}
+    </Box>
+  );
+}
+
+CompositeRetrieverBuilder.propTypes = {
+  rootComponent: PropTypes.string.isRequired,
+  rootParams: PropTypes.object,
+  onChange: PropTypes.func.isRequired,
+};
+
+
+function findNodeData(tree, nodeId) {
+  if (tree.nodeId === nodeId) return tree;
+  if (tree.children) {
+    for (const c of tree.children) {
+      const found = findNodeData(c, nodeId);
+      if (found) return found;
+    }
+  }
+  return null;
+}
+
+
+function TreeNodeView({
+  node,
+  depth,
+  isRoot,
+  parentId,
+  findComponent,
+  onEdit,
+  onAddChild,
+  onRemoveChild,
+  theme,
+  t,
+  children,
+}) {
+  if (!node || !node.component) return null;
+
+  const info = findComponent(node.component);
+  const getString = (val) => {
+    if (!val) return null;
+    if (typeof val === "string") return val;
+    if (val.en) return val.en;
+    return String(val);
+  };
+  const name = getString(info?.display_name) || getString(info?.name) || node.component || t("generative:rag.composite.configureNode");
+  const isComposite = COMPOSITE_TYPES.includes(node.component);
+  const Indent = depth * 3;
+
+  return (
+    <Box>
+      <Box sx={{ display: "flex", alignItems: "center", gap: 1, ml: Indent, position: "relative" }}>
+        {depth > 0 && (
+          <Box
+            sx={{
+              position: "absolute",
+              left: -Indent + (depth - 1) * 3 + 20,
+              top: "50%",
+              width: Indent - (depth - 1) * 3 - 20,
+              borderTop: "1px solid",
+              borderColor: "divider",
+            }}
+          />
+        )}
+
+        <Box
+          onClick={() => onEdit(node.nodeId)}
+          sx={{
+            display: "flex",
+            alignItems: "center",
+            gap: 1,
+            px: 1.5,
+            py: 1,
+            border: "1px solid",
+            borderColor: isComposite ? "secondary.main" : "primary.main",
+            borderRadius: 1,
+            cursor: "pointer",
+            backgroundColor: "background.paper",
+            "&:hover": { backgroundColor: "action.hover" },
+          }}
+        >
+          {isComposite ? (
+            <AccountTreeIcon sx={{ fontSize: 18, color: "secondary.main" }} />
+          ) : (
+            <PsychologyIcon sx={{ fontSize: 18, color: "primary.main" }} />
+          )}
+          <Typography variant="body2">
+            {name}
+          </Typography>
+        </Box>
+
+        {isComposite && (
+          <Tooltip title={t("generative:rag.composite.addChild")}>
+            <IconButton size="small" onClick={(e) => { e.stopPropagation(); onAddChild(node.nodeId); }} sx={{ color: "primary.main" }}>
+              <AddCircleOutlineIcon fontSize="small" />
+            </IconButton>
+          </Tooltip>
+        )}
+
+        {!isRoot && (
+          <Tooltip title={t("generative:rag.composite.removeNode")}>
+            <IconButton size="small" onClick={(e) => { e.stopPropagation(); onRemoveChild(parentId, node.nodeId); }} sx={{ color: "error.main" }}>
+              <RemoveCircleOutlineIcon fontSize="small" />
+            </IconButton>
+          </Tooltip>
+        )}
+      </Box>
+
+      {node.children?.length > 0 && (
+        <Box sx={{ position: "relative" }}>
+          {node.children.filter(Boolean).map((child) => (
+            <Box key={child.nodeId} sx={{ position: "relative" }}>
+              <Box
+                sx={{
+                  position: "absolute",
+                  left: Indent + 12,
+                  top: 0,
+                  bottom: 0,
+                  width: 1,
+                  borderLeft: "1px solid",
+                  borderColor: "divider",
+                }}
+              />
+              <TreeNodeView
+                node={child}
+                depth={depth + 1}
+                isRoot={false}
+                parentId={node.nodeId}
+                findComponent={findComponent}
+                onEdit={onEdit}
+                onAddChild={onAddChild}
+                onRemoveChild={onRemoveChild}
+                theme={theme}
+                t={t}
+              />
+            </Box>
+          ))}
+        </Box>
+      )}
+    </Box>
+  );
+}
+
+TreeNodeView.propTypes = {
+  node: PropTypes.object.isRequired,
+  depth: PropTypes.number.isRequired,
+  isRoot: PropTypes.bool.isRequired,
+  parentId: PropTypes.string,
+  findComponent: PropTypes.func.isRequired,
+  onEdit: PropTypes.func.isRequired,
+  onAddChild: PropTypes.func.isRequired,
+  onRemoveChild: PropTypes.func.isRequired,
+  theme: PropTypes.object.isRequired,
+  t: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorAdvancedModal.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorAdvancedModal.jsx
new file mode 100644
index 000000000..f81297668
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorAdvancedModal.jsx
@@ -0,0 +1,87 @@
+import { useState } from "react";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  IconButton,
+} from "@mui/material";
+import CloseIcon from "@mui/icons-material/Close";
+import { useTranslation } from "react-i18next";
+import { FormSchemaProvider } from "../../../../contexts/schema";
+import GeneratorConfigurationStep from "./GeneratorConfigurationStep";
+
+export default function GeneratorAdvancedModal({
+  open,
+  onClose,
+  selectedGenerator,
+  generatorModel,
+  setGeneratorModel,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [stepValid, setStepValid] = useState(false);
+
+  const handleClose = () => {
+    onClose();
+  };
+
+  const handleSave = () => {
+    onClose();
+  };
+
+  return (
+    <Dialog
+      open={open}
+      onClose={handleClose}
+      maxWidth="md"
+      fullWidth
+      PaperProps={{
+        sx: { minHeight: "500px" },
+      }}
+    >
+      <DialogTitle
+        sx={{
+          bgcolor: "background.paper",
+          display: "flex",
+          justifyContent: "space-between",
+          alignItems: "center",
+        }}
+      >
+        {t("generative:rag.advanced.generatorTitle")}
+        <IconButton
+          onClick={handleClose}
+          size="small"
+          sx={{ color: "text.secondary" }}
+        >
+          <CloseIcon />
+        </IconButton>
+      </DialogTitle>
+
+      <DialogContent dividers sx={{ bgcolor: "background.paper", minHeight: 400 }}>
+        <FormSchemaProvider key={`generator-advanced-${generatorModel?.component}`}>
+          <GeneratorConfigurationStep
+            generatorModel={generatorModel}
+            setGeneratorModel={setGeneratorModel}
+            setNextEnabled={setStepValid}
+          />
+        </FormSchemaProvider>
+      </DialogContent>
+
+      <DialogActions sx={{ p: 2, bgcolor: "background.paper" }}>
+        <Button onClick={handleClose} variant="outlined">
+          {t("generative:rag.advanced.close")}
+        </Button>
+        <Button
+          onClick={handleSave}
+          variant="contained"
+          color="primary"
+          disabled={!stepValid}
+        >
+          {t("generative:rag.advanced.done")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorConfigurationStep.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorConfigurationStep.jsx
new file mode 100644
index 000000000..ab65db1a1
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/GeneratorConfigurationStep.jsx
@@ -0,0 +1,129 @@
+import { useState, useEffect, useMemo } from "react";
+import PropTypes from "prop-types";
+import {
+  Box,
+  Autocomplete,
+  TextField,
+  Typography,
+} from "@mui/material";
+import { useTranslation } from "react-i18next";
+import FormSchema from "../../../../components/shared/FormSchema";
+import FormSchemaContainer from "../../../../components/shared/FormSchemaContainer";
+import { getGeneratorComponents } from "../../../../api/rag";
+import { resolveDefaults } from "../../../../utils/schema";
+
+export default function GeneratorConfigurationStep({
+  generatorModel,
+  setGeneratorModel,
+  setNextEnabled,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [generators, setGenerators] = useState([]);
+  const [selectedGenerator, setSelectedGenerator] = useState(null);
+
+  const [fetchedDefaults, setFetchedDefaults] = useState(null);
+
+  useEffect(() => {
+    if (!selectedGenerator) return;
+    let cancelled = false;
+    (async () => {
+      const d = await resolveDefaults(selectedGenerator.name);
+      if (!cancelled) setFetchedDefaults(d);
+    })();
+    return () => { cancelled = true; };
+  }, [selectedGenerator]);
+
+  const formInitialValues = useMemo(
+    () => {
+      if (
+        generatorModel?.component === selectedGenerator?.name &&
+        generatorModel?.params &&
+        Object.keys(generatorModel.params).length > 0
+      ) {
+        return generatorModel.params;
+      }
+      return fetchedDefaults || {};
+    },
+    [selectedGenerator, generatorModel?.component, generatorModel?.params, fetchedDefaults],
+  );
+
+  useEffect(() => {
+    let isMounted = true;
+    const load = async () => {
+      const data = await getGeneratorComponents();
+      if (isMounted) setGenerators(data || []);
+    };
+    load();
+    return () => {
+      isMounted = false;
+    };
+  }, []);
+
+  useEffect(() => {
+    if (!generators.length) return;
+    if (generatorModel?.component) {
+      const found = generators.find((g) => g.name === generatorModel.component);
+      if (found) setSelectedGenerator(found);
+    }
+    setNextEnabled(!!generatorModel?.component);
+  }, [generators, generatorModel?.component]);
+
+  const handleSelection = async (event, newValue) => {
+    setSelectedGenerator(newValue);
+    if (newValue) {
+      const defaults = await resolveDefaults(newValue.name);
+      setGeneratorModel({
+        component: newValue.name,
+        params: defaults,
+      });
+      setNextEnabled(true);
+    } else {
+      setGeneratorModel({ component: "", params: {} });
+      setNextEnabled(false);
+    }
+  };
+
+  return (
+    <Box sx={{ display: "flex", flexDirection: "column", gap: 3 }}>
+      <Typography variant="subtitle2" sx={{ mb: 0 }}>
+        {t("generative:rag.generatorConfig.configureTitle")}
+      </Typography>
+
+      <Autocomplete
+        disablePortal
+        options={generators}
+        getOptionLabel={(option) => option.name}
+        value={selectedGenerator}
+        onChange={handleSelection}
+        isOptionEqualToValue={(option, value) => option.name === value?.name}
+        renderInput={(params) => (
+          <TextField {...params} label={t("generative:rag.generatorConfig.modelLabel")} />
+        )}
+      />
+
+      {selectedGenerator && (
+        <FormSchemaContainer key={`generator-form-${selectedGenerator.name}`}>
+          <FormSchema
+            autoSave
+            model={selectedGenerator.name}
+            initialValues={formInitialValues}
+            onFormSubmit={(values) => {
+              setGeneratorModel({
+                component: selectedGenerator.name,
+                params: values,
+              });
+            }}
+            setError={(err) => console.error("FormSchema error:", err)}
+            hideButtons
+          />
+        </FormSchemaContainer>
+      )}
+    </Box>
+  );
+}
+
+GeneratorConfigurationStep.propTypes = {
+  generatorModel: PropTypes.object,
+  setGeneratorModel: PropTypes.func.isRequired,
+  setNextEnabled: PropTypes.func,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/NewPromptModal.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/NewPromptModal.jsx
new file mode 100644
index 000000000..5789409af
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/NewPromptModal.jsx
@@ -0,0 +1,266 @@
+import React, { useState, useCallback, useEffect, useRef } from "react";
+import CloseIcon from "@mui/icons-material/Close";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  Typography,
+  Grid,
+  IconButton,
+  TextField,
+  MenuItem,
+  Box,
+} from "@mui/material";
+import { useSnackbar } from "notistack";
+import { useTranslation } from "react-i18next";
+import { generateSequentialName } from "../../../../utils/nameGenerator";
+import PlaceholdersList from "../../../../components/generative/RAG/PlaceholdersList";
+import HighlightedTextarea from "../../../../components/generative/RAG/HighlightedTextarea";
+import { getCustomPrompts, createRAGPrompt } from "../../../../api/rag";
+import { LANGUAGE_CODES } from "../../../../constants/languages";
+
+export default function NewPromptModal({
+  open,
+  handleClose,
+  onPromptCreated,
+  existingPrompts = [],
+}) {
+  const { enqueueSnackbar } = useSnackbar();
+  const { t } = useTranslation(["generative"]);
+  const [promptTypes, setPromptTypes] = useState([]);
+  const selectedPromptType = "CustomRAGGenerationPrompt";
+  const [promptName, setPromptName] = useState("");
+  const [promptTemplate, setPromptTemplate] = useState("");
+  const [promptLanguage, setPromptLanguage] = useState("en");
+  const [defaultPromptName, setDefaultPromptName] = useState("");
+  const [hasUnsavedChanges, setHasUnsavedChanges] = useState(false);
+  const textareaRef = useRef(null);
+
+  const allRequiredPresent = React.useMemo(() => {
+    if (!selectedPromptType) return false;
+    const selectedType = promptTypes.find(
+      (type) => type.name === selectedPromptType,
+    );
+    if (!selectedType || !selectedType.metadata) return false;
+    const required = selectedType.metadata.required_placeholders || [];
+    return required.every((ph) => promptTemplate.includes(ph));
+  }, [selectedPromptType, promptTypes, promptTemplate]);
+
+  const canSave = Boolean(
+    selectedPromptType && promptName.trim() && allRequiredPresent,
+  );
+
+  useEffect(() => {
+    if (open) {
+      getCustomPrompts()
+        .then((data) => setPromptTypes(data))
+        .catch(() => setPromptTypes([]));
+
+      const generatedName = generateSequentialName({
+        base: "Prompt",
+        items: existingPrompts,
+      });
+      setDefaultPromptName(generatedName.defaultName);
+      setPromptName(generatedName.defaultName);
+      setPromptTemplate("");
+      setPromptLanguage("en");
+      setHasUnsavedChanges(false);
+    }
+  }, [open, existingPrompts]);
+
+  const handlePromptNameChange = useCallback((e) => {
+    setPromptName(e.target.value);
+    setHasUnsavedChanges(true);
+  }, []);
+
+  const handlePromptTemplateChange = useCallback((e) => {
+    setPromptTemplate(e.target.value);
+    setHasUnsavedChanges(true);
+  }, []);
+
+  /**
+   * Insert a placeholder string (e.g., "{chunks}") at the current cursor
+   * position in the prompt template textarea. If text is selected, it
+   * is replaced by the placeholder.
+   */
+  const handleInsertPlaceholder = useCallback(
+    (placeholder) => {
+      const textarea = textareaRef.current;
+      if (!textarea) return;
+
+      const start = textarea.selectionStart;
+      const end = textarea.selectionEnd;
+      const before = promptTemplate.substring(0, start);
+      const after = promptTemplate.substring(end);
+
+      setPromptTemplate(before + placeholder + after);
+      setHasUnsavedChanges(true);
+
+      // Restore cursor position right after the inserted placeholder
+      requestAnimationFrame(() => {
+        const pos = start + placeholder.length;
+        textarea.selectionStart = pos;
+        textarea.selectionEnd = pos;
+        textarea.focus();
+      });
+    },
+    [promptTemplate],
+  );
+
+  const handleConfirmClose = useCallback(() => {
+    if (hasUnsavedChanges) {
+      const confirmed = window.confirm(
+        t("generative:rag.newPrompt.unsavedChanges"),
+      );
+      if (!confirmed) return;
+    }
+    setHasUnsavedChanges(false);
+    handleClose();
+  }, [hasUnsavedChanges, handleClose, t]);
+
+  const handleSave = useCallback(async () => {
+    try {
+      const result = await createRAGPrompt({
+        class_name: selectedPromptType,
+        name: promptName,
+        parameters: {
+          template: promptTemplate,
+          ...(promptLanguage ? { language: promptLanguage } : {}),
+        },
+      });
+      if (result && result.id) {
+        enqueueSnackbar(t("generative:rag.newPrompt.success"), {
+          variant: "success",
+        });
+        setHasUnsavedChanges(false);
+        await onPromptCreated(result.id);
+      }
+    } catch (error) {
+      console.error("Error creating prompt:", error);
+      enqueueSnackbar(t("generative:rag.newPrompt.error"), {
+        variant: "error",
+      });
+    }
+  }, [
+    promptName,
+    promptTemplate,
+    selectedPromptType,
+    onPromptCreated,
+    enqueueSnackbar,
+    t,
+  ]);
+
+  const selectedType = promptTypes.find(
+    (type) => type.name === selectedPromptType,
+  );
+
+  return (
+    <Dialog
+      open={open}
+      onClose={handleConfirmClose}
+      fullWidth
+      maxWidth="md"
+      aria-labelledby="new-prompt-dialog-title"
+      PaperProps={{ sx: { minHeight: 600, borderRadius: 2 } }}
+    >
+      <DialogTitle id="new-prompt-dialog-title">
+        <Grid container alignItems="center" justifyContent="space-between">
+          <Grid item>
+            <Typography variant="h6" component="h2">
+              {t("generative:rag.newPrompt.title")}
+            </Typography>
+          </Grid>
+          <Grid item>
+            <IconButton
+              edge="end"
+              color="inherit"
+              onClick={handleConfirmClose}
+            >
+              <CloseIcon />
+            </IconButton>
+          </Grid>
+        </Grid>
+      </DialogTitle>
+      <DialogContent dividers>
+        <Box sx={{ mb: 2 }}>
+          <Typography variant="body2" color="textSecondary" sx={{ mb: 2 }}>
+            {t("generative:rag.newPrompt.description")}
+          </Typography>
+          <Typography variant="body2" color="textSecondary">
+            {t("generative:rag.newPrompt.placeholdersInfo")}
+          </Typography>
+        </Box>
+
+        <TextField
+          fullWidth
+          label={t("generative:rag.newPrompt.nameLabel")}
+          value={promptName}
+          onChange={handlePromptNameChange}
+          sx={{ mt: 2, mb: 2 }}
+          required
+          error={!promptName.trim()}
+          helperText={
+            !promptName.trim()
+              ? t("generative:rag.newPrompt.nameRequired")
+              : ""
+          }
+          InputLabelProps={{ required: false }}
+        />
+
+        <TextField
+          select
+          fullWidth
+          label={t("generative:rag.newPrompt.languageLabel")}
+          value={promptLanguage}
+          onChange={(e) => {
+            setPromptLanguage(e.target.value);
+            setHasUnsavedChanges(true);
+          }}
+          sx={{ mb: 2 }}
+        >
+          <MenuItem value="">{t("generative:rag.promptView.languageNone")}</MenuItem>
+          {LANGUAGE_CODES.map((code) => (
+            <MenuItem key={code} value={code}>
+              {t(`generative:rag.prompt.languages.${code}`)}
+            </MenuItem>
+          ))}
+        </TextField>
+
+        {selectedType && selectedType.metadata && (
+          <PlaceholdersList
+            required={selectedType.metadata.required_placeholders || []}
+            descriptions={selectedType.metadata.placeholder_descriptions || {}}
+            template={promptTemplate}
+            onInsertPlaceholder={handleInsertPlaceholder}
+          />
+        )}
+
+        <HighlightedTextarea
+          ref={textareaRef}
+          label={t("generative:rag.newPrompt.promptLabel")}
+          placeholder={t(
+            "generative:rag.newPrompt.promptPlaceholder",
+          )}
+          value={promptTemplate}
+          onChange={handlePromptTemplateChange}
+          minRows={6}
+          sx={{ mt: 2 }}
+        />
+      </DialogContent>
+      <DialogActions>
+        <Button onClick={handleConfirmClose}>
+          {t("generative:rag.newPrompt.cancel")}
+        </Button>
+        <Button
+          variant="contained"
+          disabled={!canSave}
+          onClick={handleSave}
+        >
+          {t("generative:rag.newPrompt.save")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverAdvancedModal.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverAdvancedModal.jsx
new file mode 100644
index 000000000..863d9d4cd
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverAdvancedModal.jsx
@@ -0,0 +1,94 @@
+import { useState, useRef } from "react";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  IconButton,
+} from "@mui/material";
+import CloseIcon from "@mui/icons-material/Close";
+import { useTranslation } from "react-i18next";
+import { FormSchemaProvider } from "../../../../contexts/schema";
+import RetrieverConfigurationStep from "./RetrieverConfigurationStep";
+
+export default function RetrieverAdvancedModal({
+  open,
+  onClose,
+  selectedParadigm,
+  allParadigms,
+  retrieverModel,
+  setRetrieverModel,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const [stepValid, setStepValid] = useState(false);
+  const retrieverStepRef = useRef(null);
+
+  const handleClose = () => {
+    onClose();
+  };
+
+  const handleSave = () => {
+    if (retrieverStepRef.current) {
+      retrieverStepRef.current.saveFormValues();
+    }
+    onClose();
+  };
+
+  return (
+    <Dialog
+      open={open}
+      onClose={handleClose}
+      maxWidth="md"
+      fullWidth
+      PaperProps={{
+        sx: { minHeight: "500px" },
+      }}
+    >
+      <DialogTitle
+        sx={{
+          bgcolor: "background.paper",
+          display: "flex",
+          justifyContent: "space-between",
+          alignItems: "center",
+        }}
+      >
+        {t("generative:rag.advanced.retrieverTitle")}
+        <IconButton
+          onClick={handleClose}
+          size="small"
+          sx={{ color: "text.secondary" }}
+        >
+          <CloseIcon />
+        </IconButton>
+      </DialogTitle>
+
+      <DialogContent dividers sx={{ bgcolor: "background.paper", minHeight: 400 }}>
+        <FormSchemaProvider key={`retriever-advanced-${retrieverModel?.component}`}>
+          <RetrieverConfigurationStep
+            ref={retrieverStepRef}
+            allParadigms={allParadigms}
+            retrieverModel={retrieverModel}
+            setRetrieverModel={setRetrieverModel}
+            setNextEnabled={setStepValid}
+          />
+        </FormSchemaProvider>
+      </DialogContent>
+
+      <DialogActions sx={{ p: 2, bgcolor: "background.paper" }}>
+        <Button onClick={handleClose} variant="outlined">
+          {t("generative:rag.advanced.close")}
+        </Button>
+        <Button
+          onClick={handleSave}
+          variant="contained"
+          color="primary"
+          disabled={!stepValid}
+        >
+          {t("generative:rag.advanced.done")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverConfigurationStep.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverConfigurationStep.jsx
new file mode 100644
index 000000000..d4d2477fd
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverConfigurationStep.jsx
@@ -0,0 +1,331 @@
+import {
+  useState,
+  useEffect,
+  useCallback,
+  useRef,
+  useMemo,
+  forwardRef,
+  useImperativeHandle,
+} from "react";
+import PropTypes from "prop-types";
+import { Box, Autocomplete, TextField, Typography } from "@mui/material";
+import { useTranslation } from "react-i18next";
+import { getRetrievalParadigm, getRetrieverComponents } from "../../../../api/rag";
+import FormSchema from "../../../../components/shared/FormSchema";
+import CompositeRetrieverBuilder from "./CompositeRetrieverBuilder";
+import {
+  FormSchemaProvider,
+} from "../../../../contexts/schema";
+import { resolveDefaults } from "../../../../utils/schema";
+
+/** Parent class name for all dense embedding components in the backend ComponentRegistry. */
+const DENSE_EMBEDDING_PARENT = "DenseEmbedding";
+
+const COMPOSITE_NAMES = ["SequentialRetriever", "ParallelRetriever", "MMRRerankerRetriever"];
+
+function getDisplayName(component) {
+  if (!component) return "";
+  const dn = component.display_name;
+  if (!dn) return component.name || "";
+  if (typeof dn === "string") return dn;
+  if (dn.en) return dn.en;
+  if (dn.es) return dn.es;
+  // Fallback: try any available language key
+  const keys = Object.keys(dn);
+  for (const key of keys) {
+    if (typeof dn[key] === "string") return dn[key];
+  }
+  return component.name || "";
+}
+
+function AutoSaveFormSchema({
+  selectedRetriever,
+  retrieverModel,
+  onParametersChange,
+}) {
+  const formikRef = useRef(null);
+
+  const [fetchedDefaults, setFetchedDefaults] = useState(null);
+
+  useEffect(() => {
+    if (!selectedRetriever) return;
+    let cancelled = false;
+    (async () => {
+      const d = await resolveDefaults(selectedRetriever.name);
+      if (!cancelled) setFetchedDefaults(d);
+    })();
+    return () => { cancelled = true; };
+  }, [selectedRetriever]);
+
+  const initialValues = useMemo(() => {
+    if (
+      retrieverModel?.component === selectedRetriever?.name &&
+      retrieverModel?.params &&
+      Object.keys(retrieverModel.params).length > 0
+    ) {
+      return retrieverModel.params;
+    }
+    return fetchedDefaults || {};
+  }, [selectedRetriever, retrieverModel, fetchedDefaults]);
+
+  const handleFormSubmit = useCallback(
+    (values) => {
+      onParametersChange(values);
+    },
+    [onParametersChange],
+  );
+
+  return (
+    <FormSchema
+      model={selectedRetriever.name}
+      initialValues={initialValues}
+      autoSave={true}
+      onFormSubmit={handleFormSubmit}
+      formSubmitRef={formikRef}
+      hideButtons
+    />
+  );
+}
+
+const RetrieverConfigurationStep = forwardRef(
+  function RetrieverConfigurationStep(
+    { allParadigms, retrieverModel, setRetrieverModel, setNextEnabled },
+    ref,
+  ) {
+    const { t } = useTranslation(["generative"]);
+    const [allOptions, setAllOptions] = useState([]);
+    const [selectedRetriever, setSelectedRetriever] = useState(null);
+    const [openConfig, setOpenConfig] = useState(false);
+    const savedParamsRef = useRef(null);
+    const retrieversRef = useRef([]);
+
+    const saveCurrentFormValues = useCallback(() => {
+      const values = savedParamsRef.current;
+      if (values && Object.keys(values).length > 0 && selectedRetriever) {
+        setRetrieverModel({
+          component: selectedRetriever.name,
+          params: values,
+        });
+      }
+    }, [selectedRetriever, setRetrieverModel]);
+
+    useEffect(() => {
+      return () => {
+        saveCurrentFormValues();
+      };
+    }, [saveCurrentFormValues]);
+
+    useImperativeHandle(
+      ref,
+      () => ({ saveFormValues: saveCurrentFormValues }),
+      [saveCurrentFormValues],
+    );
+
+    useEffect(() => {
+      const load = async () => {
+        let retrievers = [];
+        let concreteEmbeddings = [];
+        try {
+          const results = await Promise.all([
+            getRetrievalParadigm(),
+            getRetrieverComponents(DENSE_EMBEDDING_PARENT),
+          ]);
+          retrievers = results[0] || [];
+          concreteEmbeddings = (results[1] || []).filter(
+            (c) => !(c.flags || []).includes("abstract"),
+          );
+        } catch (e) {
+          retrievers = await getRetrievalParadigm();
+        }
+        retrieversRef.current = retrievers;
+
+        const keywordOptions = retrievers
+          .filter((c) => (c.flags || []).includes("keyword"))
+          .map((c) => ({ ...c, _type: "retriever" }));
+        const denseOptions = concreteEmbeddings.map((c) => ({
+          ...c,
+          _type: "embedding",
+        }));
+        const compositeOptions = retrievers
+          .filter((c) => (c.flags || []).includes("composite"))
+          .map((c) => ({ ...c, _type: "retriever" }));
+        const merged = [
+          ...keywordOptions,
+          ...denseOptions,
+          ...compositeOptions,
+        ];
+        setAllOptions(merged);
+
+        if (retrieverModel?.component) {
+          let found = null;
+          if (retrieverModel.component === "DenseEmbeddingRetriever") {
+            const embName =
+              retrieverModel.params?.embedding_model?.component;
+            if (embName) {
+              found = merged.find((c) => c.name === embName);
+            }
+            if (!found) {
+              found =
+                retrievers.find(
+                  (c) => c.name === "DenseEmbeddingRetriever",
+                ) || null;
+            }
+          } else {
+            found = merged.find((c) => c.name === retrieverModel.component);
+            if (!found && allParadigms) {
+              found = allParadigms.find(
+                (c) => c.name === retrieverModel.component,
+              );
+            }
+          }
+          if (found) {
+            setSelectedRetriever(found);
+            setOpenConfig(true);
+            setNextEnabled(true);
+          }
+        }
+      };
+      load();
+    }, []);
+
+    const handleRetrieverChange = useCallback(async (_event, newValue) => {
+      savedParamsRef.current = null;
+
+      if (!newValue) {
+        setSelectedRetriever(null);
+        setRetrieverModel({ component: "", params: {} });
+        setOpenConfig(false);
+        setNextEnabled(false);
+        return;
+      }
+
+      if (newValue._type === "embedding") {
+        const embeddingDefaults = await resolveDefaults(newValue.name);
+        const denseRetrieverDefaults = await resolveDefaults(
+          "DenseEmbeddingRetriever",
+        );
+        const model = {
+          component: "DenseEmbeddingRetriever",
+          params: {
+            ...denseRetrieverDefaults,
+            embedding_model: {
+              component: newValue.name,
+              params: embeddingDefaults,
+            },
+          },
+        };
+        const found = retrieversRef.current.find(
+          (c) => c.name === "DenseEmbeddingRetriever",
+        );
+        setSelectedRetriever(
+          found
+            ? { ...found, _type: "retriever" }
+            : { ...newValue, _type: "retriever", name: "DenseEmbeddingRetriever" },
+        );
+        setRetrieverModel(model);
+        setOpenConfig(Boolean(found?.schema?.properties));
+        setNextEnabled(true);
+      } else {
+        const defaults = await resolveDefaults(newValue.name);
+        setSelectedRetriever(newValue);
+        setRetrieverModel({ component: newValue.name, params: defaults });
+        setOpenConfig(Boolean(newValue?.schema?.properties));
+        setNextEnabled(true);
+      }
+    }, [setRetrieverModel, setNextEnabled]);
+
+    const handleParametersSave = useCallback(
+      (newParams) => {
+        savedParamsRef.current = newParams;
+        setRetrieverModel({
+          component: selectedRetriever?.name || "",
+          params: newParams,
+        });
+        setNextEnabled(true);
+      },
+      [selectedRetriever?.name, setRetrieverModel, setNextEnabled],
+    );
+
+    const handleCompositeChange = useCallback(
+      (updated) => {
+        savedParamsRef.current = updated.params;
+        setRetrieverModel({
+          component: updated.component,
+          params: updated.params,
+        });
+        setNextEnabled(true);
+      },
+      [setRetrieverModel, setNextEnabled],
+    );
+
+    const isComposite =
+      selectedRetriever && COMPOSITE_NAMES.includes(selectedRetriever.name);
+
+    return (
+      <Box sx={{ display: "flex", flexDirection: "column", gap: 3, p: 2 }}>
+        <Typography variant="subtitle2" sx={{ mb: 0 }}>
+          {t("generative:rag.composite.selectModel")}
+        </Typography>
+        <Typography variant="body2" color="text.secondary" sx={{ mb: 1 }}>
+          {t("generative:rag.composite.retrieverDescription")}
+        </Typography>
+        <Autocomplete
+          options={allOptions}
+          getOptionLabel={(opt) => getDisplayName(opt)}
+          groupBy={(opt) => {
+            if (opt._type === "embedding")
+              return t("generative:rag.composite.denseGroup");
+            const flags = opt.flags || [];
+            if (flags.includes("composite"))
+              return t("generative:rag.composite.compositeGroup");
+            if (flags.includes("keyword"))
+              return t("generative:rag.composite.keywordGroup");
+            return t("generative:rag.composite.simpleGroup");
+          }}
+          value={selectedRetriever}
+          onChange={handleRetrieverChange}
+          isOptionEqualToValue={(a, b) => a.name === b.name}
+          renderInput={(params) => (
+            <TextField {...params} label={t("generative:rag.retrieverConfig.modelLabel")} />
+          )}
+        />
+
+        {selectedRetriever && openConfig && !isComposite && (
+          <FormSchemaProvider
+            key={`retriever-provider-${selectedRetriever.name}`}
+          >
+            <AutoSaveFormSchema
+              key={`retriever-form-${selectedRetriever.name}`}
+              selectedRetriever={selectedRetriever}
+              retrieverModel={retrieverModel}
+              onParametersChange={handleParametersSave}
+            />
+          </FormSchemaProvider>
+        )}
+
+        {selectedRetriever && openConfig && isComposite && (
+          <>
+            <Typography variant="body2" color="text.secondary" sx={{ mt: -1 }}>
+              {t("generative:rag.composite.compositeInstructions")}
+            </Typography>
+            <CompositeRetrieverBuilder
+              key={`composite-${selectedRetriever.name}`}
+              rootComponent={selectedRetriever.name}
+              rootParams={retrieverModel.params}
+              onChange={handleCompositeChange}
+            />
+          </>
+        )}
+      </Box>
+    );
+  },
+);
+
+RetrieverConfigurationStep.propTypes = {
+  allParadigms: PropTypes.array,
+  retrieverModel: PropTypes.object,
+  setRetrieverModel: PropTypes.func.isRequired,
+  setNextEnabled: PropTypes.func.isRequired,
+};
+
+export default RetrieverConfigurationStep;
diff --git a/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverNodeConfig.jsx b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverNodeConfig.jsx
new file mode 100644
index 000000000..e6c263438
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/advanced/RetrieverNodeConfig.jsx
@@ -0,0 +1,169 @@
+import { useState, useEffect, useMemo } from "react";
+import {
+  Dialog,
+  DialogTitle,
+  DialogContent,
+  DialogActions,
+  Button,
+  IconButton,
+  Autocomplete,
+  TextField,
+  Typography,
+  Box,
+  Divider,
+} from "@mui/material";
+import CloseIcon from "@mui/icons-material/Close";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import FormSchema from "../../../../components/shared/FormSchema";
+import FormSchemaContainer from "../../../../components/shared/FormSchemaContainer";
+import { resolveDefaults } from "../../../../utils/schema";
+
+const COMPOSITE_TYPES = ["SequentialRetriever", "ParallelRetriever", "MMRRerankerRetriever"];
+
+export default function RetrieverNodeConfig({
+  open,
+  nodeId,
+  nodeData,
+  allComponents,
+  leafRegistry,
+  onSave,
+  onClose,
+}) {
+  const { t } = useTranslation(["generative"]);
+
+  const allOptions = useMemo(() => {
+    const list = [...allComponents];
+    for (const key of Object.keys(leafRegistry)) {
+      list.push(...(leafRegistry[key] || []));
+    }
+    return list.filter((c) => c.configurable_object !== false);
+  }, [allComponents, leafRegistry]);
+
+  const [selectedModel, setSelectedModel] = useState(null);
+  const [params, setParams] = useState({});
+  const [initialParams, setInitialParams] = useState({});
+
+  useEffect(() => {
+    if (!nodeData || !allOptions.length) return;
+    const found = allOptions.find((c) => c.name === nodeData.component);
+    setSelectedModel(found || null);
+    const p = nodeData.params || {};
+    setParams(p);
+    setInitialParams(p);
+  }, [nodeData, allOptions]);
+
+  const isComposite = selectedModel && COMPOSITE_TYPES.includes(selectedModel.name);
+
+  const getDisplay = (opt) => {
+    if (!opt) return "";
+    const dn = opt.display_name;
+    if (!dn) return opt.name || "";
+    if (typeof dn === "string") return dn;
+    if (dn.en) return dn.en;
+    if (dn.es) return dn.es;
+    // Fallback: try any available language key
+    const keys = Object.keys(dn);
+    for (const key of keys) {
+      if (typeof dn[key] === "string") return dn[key];
+    }
+    return opt.name || "";
+  };
+
+  const handleModelChange = async (_e, newVal) => {
+    setSelectedModel(newVal);
+    if (newVal) {
+      const defaults = await resolveDefaults(newVal.name);
+      setParams(defaults);
+      setInitialParams(defaults);
+    } else {
+      setParams({});
+    }
+  };
+
+  const handleSave = () => {
+    if (!selectedModel) return;
+    onSave(nodeId, selectedModel.name, params);
+  };
+
+  return (
+    <Dialog open={open} onClose={onClose} maxWidth="sm" fullWidth>
+      <DialogTitle
+        sx={{
+          bgcolor: "background.paper",
+          display: "flex",
+          justifyContent: "space-between",
+          alignItems: "center",
+        }}
+      >
+        {t("generative:rag.composite.configureNode")}
+        <IconButton onClick={onClose} size="small" sx={{ color: "text.secondary" }}>
+          <CloseIcon />
+        </IconButton>
+      </DialogTitle>
+
+      <DialogContent dividers sx={{ bgcolor: "background.paper" }}>
+        <Box sx={{ display: "flex", flexDirection: "column", gap: 3 }}>
+          <Autocomplete
+            options={allOptions}
+            value={selectedModel}
+            onChange={handleModelChange}
+            getOptionLabel={(opt) => getDisplay(opt)}
+            isOptionEqualToValue={(a, b) => a.name === b.name}
+            groupBy={(opt) => {
+              if (COMPOSITE_TYPES.includes(opt.name)) return t("generative:rag.composite.compositeGroup");
+              return t("generative:rag.composite.simpleGroup");
+            }}
+            renderInput={(p) => (
+              <TextField {...p} label={t("generative:rag.composite.selectModel")} />
+            )}
+          />
+
+          {isComposite && (
+            <Typography variant="caption" color="text.secondary">
+              {t("generative:rag.composite.nestedCompositeNote")}
+            </Typography>
+          )}
+
+          {selectedModel && (
+            <>
+              <Divider />
+              <Typography variant="subtitle2">
+                {t("generative:rag.composite.parameters")}
+              </Typography>
+              <FormSchemaContainer key={`node-config-${nodeId}-${selectedModel.name}`}>
+                <FormSchema
+                  model={selectedModel.name}
+                  initialValues={initialParams}
+                  autoSave
+                  onFormSubmit={(values) => setParams(values)}
+                  hideButtons
+                  excludeFields={isComposite ? ["children"] : []}
+                />
+              </FormSchemaContainer>
+            </>
+          )}
+        </Box>
+      </DialogContent>
+
+      <DialogActions sx={{ p: 2, bgcolor: "background.paper" }}>
+        <Button onClick={onClose} variant="outlined">
+          {t("common:cancel")}
+        </Button>
+        <Button onClick={handleSave} variant="contained" disabled={!selectedModel}>
+          {t("common:save")}
+        </Button>
+      </DialogActions>
+    </Dialog>
+  );
+}
+
+RetrieverNodeConfig.propTypes = {
+  open: PropTypes.bool.isRequired,
+  nodeId: PropTypes.string.isRequired,
+  nodeData: PropTypes.object,
+  allComponents: PropTypes.array.isRequired,
+  leafRegistry: PropTypes.object.isRequired,
+  onSave: PropTypes.func.isRequired,
+  onClose: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/AdvancedConfigCard.jsx b/DashAI/front/src/pages/generative/RAGSession/components/AdvancedConfigCard.jsx
new file mode 100644
index 000000000..04391159f
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/AdvancedConfigCard.jsx
@@ -0,0 +1,21 @@
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import PresetCard from "./PresetCard";
+
+export default function AdvancedConfigCard({ modelName, onClick }) {
+  const { t } = useTranslation(["generative"]);
+
+  return (
+    <PresetCard
+      selected
+      label={t("generative:rag.retriever.advancedApplied")}
+      description={modelName}
+      onClick={onClick}
+    />
+  );
+}
+
+AdvancedConfigCard.propTypes = {
+  modelName: PropTypes.string.isRequired,
+  onClick: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/GeneratorBody.jsx b/DashAI/front/src/pages/generative/RAGSession/components/GeneratorBody.jsx
new file mode 100644
index 000000000..8a80c0a7b
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/GeneratorBody.jsx
@@ -0,0 +1,254 @@
+import { useState, useEffect, useMemo } from "react";
+import {
+  Box,
+  Typography,
+  Autocomplete,
+  TextField,
+  Button,
+  Alert,
+  AlertTitle,
+  useTheme
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getGeneratorComponents } from "../../../../api/rag";
+import { resolveDefaults } from "../../../../utils/schema";
+import GeneratorAdvancedModal from "../advanced/GeneratorAdvancedModal";
+import AdvancedConfigCard from "./AdvancedConfigCard";
+import { getDescription } from "./sectionUtils";
+
+export default function GeneratorBody({
+  generatorModel,
+  setGeneratorModel,
+  chunkSize = 0,
+  topK = 0,
+  promptTokenCount = 0,
+  setIsValid,
+  isAdvanced,
+  setInitialModelParams,
+  showDetails = true,
+}) {
+  const theme = useTheme();
+  const { t, i18n } = useTranslation(["generative"]);
+  const [generators, setGenerators] = useState([]);
+  const [selectedGenerator, setSelectedGenerator] = useState(null);
+  const [showAdvanced, setShowAdvanced] = useState(false);
+  const [loading, setLoading] = useState(true);
+  const DEFAULT_CONTEXT_WINDOW = 10000;
+  const DEFAULT_MAX_TOKENS = 1000;
+
+  const contextStats = useMemo(() => {
+    if (!generatorModel?.params || !selectedGenerator || !generatorModel.component) {
+      return { isValid: false, availableTokens: 0 };
+    }
+
+    const contextWindow = generatorModel.params.context_window ?? DEFAULT_CONTEXT_WINDOW;
+    const maxTokens = generatorModel.params.max_tokens ?? DEFAULT_MAX_TOKENS;
+    const chunkTokens = chunkSize * topK;
+    const availableForMessage = contextWindow - chunkTokens - promptTokenCount - maxTokens;
+    const isValid = availableForMessage > 0;
+
+    return {
+      isValid,
+      availableTokens: Math.max(0, Math.floor(availableForMessage))
+    };
+  }, [generatorModel?.params, generatorModel?.component, selectedGenerator, chunkSize, topK, promptTokenCount]);
+
+  const isRemoteModel = useMemo(() => {
+    if (!selectedGenerator || !generatorModel?.component) return false;
+    const name = (selectedGenerator.name || "").toLowerCase();
+    return name.includes("openai") || name.includes("deepseek");
+  }, [selectedGenerator, generatorModel?.component]);
+
+  const isApiKeyMissing = useMemo(() => {
+    if (!isRemoteModel || !generatorModel?.params) return false;
+    return !generatorModel.params.API_key;
+  }, [isRemoteModel, generatorModel?.params]);
+
+  const overallIsValid = contextStats.isValid && !isApiKeyMissing;
+
+  useEffect(() => {
+    setIsValid(overallIsValid);
+  }, [overallIsValid, setIsValid]);
+
+  useEffect(() => {
+    const loadGenerators = async () => {
+      try {
+        const data = await getGeneratorComponents();
+        setGenerators(data || []);
+      } catch (error) {
+        console.error("Error loading generators:", error);
+      } finally {
+        setLoading(false);
+      }
+    };
+    loadGenerators();
+  }, []);
+
+  useEffect(() => {
+    if (!generators.length) return;
+
+    if (!generatorModel?.component) {
+      if (setInitialModelParams) setInitialModelParams(null);
+      return;
+    }
+
+    const foundGenerator = generators.find((generator) => {
+      return generator.name === generatorModel.component;
+    });
+
+    if (!foundGenerator) return;
+
+    const selectedGeneratorName = selectedGenerator?.name || null;
+    const currentGeneratorName = generatorModel.component;
+
+    if (selectedGeneratorName === currentGeneratorName) {
+      return;
+    }
+
+    setSelectedGenerator(foundGenerator);
+    if (setInitialModelParams) setInitialModelParams({ ...generatorModel.params });
+  }, [generators, generatorModel?.component, generatorModel?.params]);
+
+  const handleGeneratorChange = async (event, newValue) => {
+    if (!newValue) {
+      setSelectedGenerator(null);
+      if (setInitialModelParams) setInitialModelParams(null);
+      setGeneratorModel({ component: "", params: {} });
+      return;
+    }
+
+    const initialParams = await resolveDefaults(newValue.name);
+    const overriddenParams = {
+      ...initialParams,
+      max_tokens: DEFAULT_MAX_TOKENS,
+      context_window: DEFAULT_CONTEXT_WINDOW,
+    };
+    if (setInitialModelParams) setInitialModelParams({ ...overriddenParams });
+
+    setSelectedGenerator(newValue);
+    setGeneratorModel({
+      component: newValue.name,
+      params: { ...overriddenParams },
+    });
+  };
+
+  if (loading) {
+    return null;
+  }
+
+  return (
+    <>
+      <Box>
+        <Autocomplete
+          options={generators}
+          value={selectedGenerator}
+          onChange={handleGeneratorChange}
+          getOptionLabel={(option) => getDescription(option.display_name, i18n) || option.name || ""}
+          isOptionEqualToValue={(option, value) => option?.name === value?.name}
+          renderInput={(params) => (
+            <TextField
+              {...params}
+              label={t("generative:rag.generator.selectModel")}
+              placeholder={t("generative:rag.generator.selectModelPlaceholder")}
+            />
+          )}
+        />
+      </Box>
+
+      {isAdvanced && selectedGenerator && (
+        <AdvancedConfigCard
+          modelName={selectedGenerator.name}
+          onClick={() => setShowAdvanced(true)}
+        />
+      )}
+
+      {showDetails && selectedGenerator && generatorModel?.params && (
+        <Box
+          sx={{
+            p: 2,
+            backgroundColor: "action.hover",
+            border: "1px solid",
+            borderColor: contextStats.isValid ? "divider" : "error.main",
+            borderRadius: 1,
+            display: "flex",
+            flexDirection: "column",
+            gap: 1
+          }}
+        >
+          <Typography variant="body2">
+            <Box component="span" sx={{ typography: "subtitle2" }}>
+              {t("generative:rag.generator.modelInfo")}
+            </Box> {selectedGenerator.name}
+          </Typography>
+
+          <Typography variant="body2" sx={{ color: contextStats.isValid ? "success.main" : "error.main" }}>
+            {t("generative:validation.contextSpace", { availableChars: contextStats.availableTokens?.toLocaleString() })}
+          </Typography>
+
+          {!contextStats.isValid && (
+            <Alert severity="error" sx={{ mt: 1 }}>
+              <AlertTitle>{t("generative:validation.insufficientContextTitle")}</AlertTitle>
+              {t("generative:validation.insufficientContextDescription")}
+            </Alert>
+          )}
+
+          {isApiKeyMissing && (
+            <Alert severity="warning" sx={{ mt: 1 }}>
+              <AlertTitle>{t("generative:validation.apiKeyMissingTitle")}</AlertTitle>
+              {t("generative:validation.apiKeyMissingDescription")}
+            </Alert>
+          )}
+
+          {getDescription(selectedGenerator.description, i18n) && (
+            <Typography variant="caption" display="block" sx={{ mt: 0.5 }}>
+              {getDescription(selectedGenerator.description, i18n)}
+            </Typography>
+          )}
+        </Box>
+      )}
+
+      {showDetails && (
+        <Button
+          variant="outlined"
+          color="primary"
+          size="small"
+          onClick={() => setShowAdvanced(true)}
+          disabled={!selectedGenerator}
+          sx={{
+            alignSelf: "flex-start", 
+            width: "fit-content",
+            border: "1px solid",
+            borderColor: theme.palette.primary.main,
+            backgroundColor: theme.palette.action.selected,
+            color: theme.palette.text.primary,
+          }}
+        >
+          ↗ {t("generative:rag.generator.advancedButton")}
+        </Button>
+      )}
+
+      {selectedGenerator && (
+        <GeneratorAdvancedModal
+          open={showAdvanced}
+          onClose={() => setShowAdvanced(false)}
+          selectedGenerator={selectedGenerator}
+          generatorModel={generatorModel}
+          setGeneratorModel={setGeneratorModel}
+        />
+      )}
+    </>
+  );
+}
+
+GeneratorBody.propTypes = {
+  generatorModel: PropTypes.object,
+  setGeneratorModel: PropTypes.func.isRequired,
+  chunkSize: PropTypes.number,
+  topK: PropTypes.number,
+  promptTokenCount: PropTypes.number,
+  setIsValid: PropTypes.func.isRequired,
+  isAdvanced: PropTypes.bool,
+  setInitialModelParams: PropTypes.func,
+  showDetails: PropTypes.bool,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/PresetCard.jsx b/DashAI/front/src/pages/generative/RAGSession/components/PresetCard.jsx
new file mode 100644
index 000000000..e986c7968
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/PresetCard.jsx
@@ -0,0 +1,58 @@
+import { Typography, Paper } from "@mui/material";
+import { useTheme } from "@mui/material/styles";
+import PropTypes from "prop-types";
+
+export default function PresetCard({
+  selected = false,
+  onClick,
+  label,
+  description,
+  sx: extraSx,
+}) {
+  const theme = useTheme();
+
+  return (
+    <Paper
+      variant="outlined"
+      onClick={onClick}
+      sx={{
+        flex: 1,
+        cursor: "pointer",
+        minHeight: 56,
+        px: 3,
+        py: 4,
+        display: "flex",
+        flexDirection: "column",
+        gap: 0.5,
+        justifyContent: "flex-start",
+        border: "1px solid",
+        borderColor: selected ? theme.palette.primary.main : theme.palette.ui.border,
+        borderRadius: 2,
+        backgroundColor: selected ? theme.palette.action.selected : theme.palette.background.paper,
+        transition: theme.transitions.create(["background-color", "border-color", "box-shadow"], {
+          duration: theme.transitions.duration.short,
+        }),
+        "&:hover": {
+          backgroundColor: selected ? theme.palette.action.selected : theme.palette.action.hover,
+          borderColor: theme.palette.primary.main,
+        },
+        ...extraSx,
+      }}
+    >
+      <Typography variant="subtitle2">
+        {label}
+      </Typography>
+      <Typography variant="caption">
+        {description}
+      </Typography>
+    </Paper>
+  );
+}
+
+PresetCard.propTypes = {
+  selected: PropTypes.bool,
+  onClick: PropTypes.func,
+  label: PropTypes.string.isRequired,
+  description: PropTypes.string,
+  sx: PropTypes.object,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/RAGCard.jsx b/DashAI/front/src/pages/generative/RAGSession/components/RAGCard.jsx
new file mode 100644
index 000000000..72c871f96
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/RAGCard.jsx
@@ -0,0 +1,98 @@
+import {
+  Accordion,
+  AccordionSummary,
+  AccordionDetails,
+  Box,
+  Typography,
+  Tooltip,
+  IconButton,
+} from "@mui/material";
+import ExpandMoreIcon from "@mui/icons-material/ExpandMore";
+import PropTypes from "prop-types";
+
+export default function RAGCard({
+  title,
+  expanded,
+  onChange,
+  actions = [],
+  children,
+}) {
+  return (
+    <Accordion expanded={expanded} onChange={onChange}
+      sx={{
+        "&::before": {
+          display: "none",
+        },
+        "&.Mui-expanded": {
+          marginTop: 0,
+          marginBottom: 0,
+        },
+        px: 4
+      }}>
+      <AccordionSummary
+        expandIcon={<ExpandMoreIcon />}
+        sx={{
+          minHeight: 48,
+          "&.Mui-expanded": {
+            minHeight: 48,
+          },
+        }}
+      >
+        <Box
+          sx={{
+            display: "flex",
+            alignItems: "center",
+            justifyContent: "space-between",
+            width: "100%",
+          }}
+        >
+          <Typography variant="subtitle1">
+            {title}
+          </Typography>
+          {actions.length > 0 && (
+            <Box sx={{ display: "flex" }}>
+              {actions.map((action, index) => (
+                <Tooltip key={index} title={action.tooltip}>
+                  <IconButton
+                    size="small"
+                    onClick={(e) => {
+                      e.stopPropagation();
+                      action.onClick?.();
+                    }}
+                    aria-label={action.ariaLabel || ""}
+                  >
+                    {action.icon}
+                  </IconButton>
+                </Tooltip>
+              ))}
+            </Box>
+          )}
+        </Box>
+      </AccordionSummary>
+      <AccordionDetails
+        sx={{
+          display: "flex",
+          flexDirection: "column",
+          pt: 0,
+        }}
+      >
+        {children}
+      </AccordionDetails>
+    </Accordion>
+  );
+}
+
+RAGCard.propTypes = {
+  title: PropTypes.string.isRequired,
+  expanded: PropTypes.bool.isRequired,
+  onChange: PropTypes.func.isRequired,
+  actions: PropTypes.arrayOf(
+    PropTypes.shape({
+      icon: PropTypes.node.isRequired,
+      tooltip: PropTypes.string,
+      onClick: PropTypes.func,
+      ariaLabel: PropTypes.string,
+    }),
+  ),
+  children: PropTypes.node,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/RAGSectionColumn.jsx b/DashAI/front/src/pages/generative/RAGSession/components/RAGSectionColumn.jsx
new file mode 100644
index 000000000..5e1d827bc
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/RAGSectionColumn.jsx
@@ -0,0 +1,27 @@
+import { Box } from "@mui/material";
+import PropTypes from "prop-types";
+
+export default function RAGSectionColumn({ children, gap = 4, sx, ...props }) {
+  return (
+    <Box
+      sx={[
+        {
+          display: "flex",
+          flexDirection: "column",
+          width: "100%",
+          gap,
+        },
+        sx,
+      ]}
+      {...props}
+    >
+      {children}
+    </Box>
+  );
+}
+
+RAGSectionColumn.propTypes = {
+  children: PropTypes.node,
+  gap: PropTypes.oneOfType([PropTypes.number, PropTypes.string]),
+  sx: PropTypes.oneOfType([PropTypes.array, PropTypes.object, PropTypes.func]),
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/SectionCard.jsx b/DashAI/front/src/pages/generative/RAGSession/components/SectionCard.jsx
new file mode 100644
index 000000000..7af7dbbb9
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/SectionCard.jsx
@@ -0,0 +1,56 @@
+import { Box } from "@mui/material";
+import PropTypes from "prop-types";
+import { useTheme } from "@mui/material/styles";
+
+export default function SectionCard({
+  children,
+  sx,
+  gap = 0,
+  alignItems = "stretch",
+  justifyContent,
+}) {
+  const theme = useTheme();
+  const sectionGap =
+    typeof gap === "number" ? theme.spacing(gap) : gap;
+
+  return (
+    <Box
+      sx={[
+        {
+          "--rag-section-gap": sectionGap,
+          display: "flex",
+          flexDirection: "column",
+          gap,
+          alignItems,
+          justifyContent,
+          pt: 0,
+          pb: 3,
+        },
+        sx,
+      ]}
+    >
+      {children}
+    </Box>
+  );
+}
+
+SectionCard.propTypes = {
+  children: PropTypes.node,
+  sx: PropTypes.oneOfType([PropTypes.array, PropTypes.object, PropTypes.func]),
+  gap: PropTypes.oneOfType([PropTypes.number, PropTypes.string]),
+  alignItems: PropTypes.oneOf([
+    "stretch",
+    "flex-start",
+    "center",
+    "flex-end",
+    "baseline",
+  ]),
+  justifyContent: PropTypes.oneOf([
+    "flex-start",
+    "center",
+    "flex-end",
+    "space-between",
+    "space-around",
+    "space-evenly",
+  ]),
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/components/sectionUtils.jsx b/DashAI/front/src/pages/generative/RAGSession/components/sectionUtils.jsx
new file mode 100644
index 000000000..ec64f0b4c
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/components/sectionUtils.jsx
@@ -0,0 +1,68 @@
+import { Box } from "@mui/material";
+
+export const getDescription = (desc, i18n) => {
+  if (!desc) return "";
+  if (typeof desc === "string") return desc;
+  if (typeof desc === "object" && (desc.en || desc.es)) {
+    return desc[i18n.language] || desc.en || desc.es || "";
+  }
+  return "";
+};
+
+/**
+ * Render a template string with placeholders ({chunks}, {input}, etc.) highlighted.
+ *
+ * @param {string} template - The template string to render
+ * @param {object}  [highlightColors] - Optional resolved color overrides (hex/rgba)
+ * @param {string} [highlightColors.bg] - Background color for placeholders
+ * @param {string} [highlightColors.text] - Text color for placeholders
+ */
+export const renderTemplateWithHighlights = (template, highlightColors = {}, fontFamily = "monospace") => {
+  if (!template) return null;
+
+  const { bg, text } = highlightColors;
+
+  const placeholderRegex = /\{([^}]+)\}/g;
+  const parts = [];
+  let lastIndex = 0;
+  let match;
+
+  while ((match = placeholderRegex.exec(template)) !== null) {
+    if (match.index > lastIndex) {
+      parts.push(template.substring(lastIndex, match.index));
+    }
+
+    parts.push({
+      type: "placeholder",
+      value: match[0],
+      label: match[1],
+    });
+
+    lastIndex = placeholderRegex.lastIndex;
+  }
+
+  if (lastIndex < template.length) {
+    parts.push(template.substring(lastIndex));
+  }
+
+  return parts.map((part, idx) => {
+    if (typeof part === "string") {
+      return <span key={idx}>{part}</span>;
+    }
+    return (
+      <Box
+        component="span"
+        key={idx}
+        sx={{
+          backgroundColor: bg || "warning.light",
+          color: text || "warning.dark",
+          padding: "2px 4px",
+          borderRadius: "3px",
+          fontFamily,
+        }}
+      >
+        {part.value}
+      </Box>
+    );
+  });
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/sections/ChunkingSection.jsx b/DashAI/front/src/pages/generative/RAGSession/sections/ChunkingSection.jsx
new file mode 100644
index 000000000..0d005c3cd
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/sections/ChunkingSection.jsx
@@ -0,0 +1,212 @@
+import { useState, useEffect, useMemo} from "react";
+import {
+  Box,
+  Typography,
+  Button,
+  CircularProgress,
+  useTheme
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getChunkingComponents } from "../../../../api/rag";
+import { resolveDefaults, getModelFromSubform, getParamsFromSubform } from "../../../../utils/schema";
+import ChunkingAdvancedModal from "../advanced/ChunkingAdvancedModal";
+import AdvancedConfigCard from "../components/AdvancedConfigCard";
+import PresetCard from "../components/PresetCard";
+import RAGSectionColumn from "../components/RAGSectionColumn";
+
+export default function ChunkingSection({
+  chunkingModel,
+  setChunkingModel,
+}) {
+  const { t } = useTranslation(["generative"]);
+  const theme = useTheme();
+
+  const CHUNKING_PRESETS = useMemo(() => [
+    {
+      value: "small",
+      label: t("generative:rag.chunking.presets.small.label"),
+      config: { chunk_size: 256, chunk_overlap: 25 },
+    },
+    {
+      value: "paragraph",
+      label: t("generative:rag.chunking.presets.paragraph.label"),
+      config: { chunk_size: 500, chunk_overlap: 50 },
+    },
+    {
+      value: "page",
+      label: t("generative:rag.chunking.presets.page.label"),
+      config: { chunk_size: 2000, chunk_overlap: 200 },
+    },
+    {
+      value: "large",
+      label: t("generative:rag.chunking.presets.large.label"),
+      config: { chunk_size: 4000, chunk_overlap: 400 },
+    },
+  ], [t]);
+
+  const getPresetDescription = (preset) => {
+    const chars = preset.config.chunk_size;
+    const tokens = Math.ceil(chars / 4);
+    return t("generative:rag.chunking.presets.chunkSizeFormat", { chars, tokens });
+  };
+
+  const [chunkers, setChunkers] = useState([]);
+  const [selectedChunker, setSelectedChunker] = useState(null);
+  const [selectedPreset, setSelectedPreset] = useState("paragraph");
+  const [showAdvanced, setShowAdvanced] = useState(false);
+  const [loading, setLoading] = useState(true);
+
+  useEffect(() => {
+    const loadChunkers = async () => {
+      try {
+        const data = await getChunkingComponents();
+        setChunkers(data || []);
+        
+        if (data && data.length > 0) {
+          const modelName = getModelFromSubform(chunkingModel);
+          const params = getParamsFromSubform(chunkingModel) ?? chunkingModel?.params;
+
+          if (modelName) {
+            const found = data.find((c) => c.name === modelName);
+            if (found) {
+              setSelectedChunker(found);
+              updatePresetFromParams(params);
+            } else {
+              selectDefaultChunker(data);
+            }
+          } else {
+            selectDefaultChunker(data);
+          }
+        }
+      } catch (error) {
+        console.error("Error loading chunkers:", error);
+      } finally {
+        setLoading(false);
+      }
+    };
+    loadChunkers();
+  }, []);
+
+  // Update preset when chunkingModel.params changes (e.g. from Advanced Modal)
+  useEffect(() => {
+    const params = getParamsFromSubform(chunkingModel) ?? chunkingModel?.params;
+    if (params) {
+      updatePresetFromParams(params);
+    }
+  }, [chunkingModel, CHUNKING_PRESETS]);
+
+  const updatePresetFromParams = (params) => {
+    const preset = CHUNKING_PRESETS.find(p => 
+      p.config.chunk_size === params?.chunk_size &&
+      p.config.chunk_overlap === params?.chunk_overlap
+    );
+    if (preset) {
+      setSelectedPreset(preset.value);
+    } else {
+      setSelectedPreset("custom");
+    }
+  };
+
+  const selectDefaultChunker = (availableChunkers) => {
+    const defaultChunker = availableChunkers.find(c => c.name === "CharacterChunkModel") || availableChunkers[0];
+    const preset = CHUNKING_PRESETS.find(p => p.value === "paragraph");
+    applyPreset(defaultChunker, preset);
+  };
+
+  const applyPreset = async (chunker, preset) => {
+    setSelectedChunker(chunker);
+    setSelectedPreset(preset?.value || "custom");
+    
+    const defaultParams = await resolveDefaults(chunker.name);
+    const params = {
+      ...defaultParams,
+      ...(preset?.config || {}),
+    };
+
+    setChunkingModel({
+      component: chunker.name,
+      params,
+    });
+  };
+
+  const handlePresetClick = (presetValue) => {
+    if (!selectedChunker) return;
+    if (presetValue === selectedPreset) return;
+    const preset = CHUNKING_PRESETS.find(p => p.value === presetValue);
+    if (preset) {
+      applyPreset(selectedChunker, preset);
+    }
+  };
+
+  if (loading) {
+    return (
+      <Box display="flex" justifyContent="center" alignItems="center" sx={{ minHeight: 120 }}>
+        <CircularProgress />
+      </Box>
+    );
+  }
+
+  const isCustom = selectedPreset === "custom";
+
+  return (
+      <RAGSectionColumn>
+        <Typography variant="body2" color="textSecondary">
+          {t("generative:rag.chunking.description")}
+        </Typography>
+
+        <Box sx={{ display: "flex", gap: 1, alignItems: "stretch", flexWrap: "wrap" }}>
+          {CHUNKING_PRESETS.map((preset) => (
+            <PresetCard
+            key={preset.value}
+            selected={selectedPreset === preset.value}
+            onClick={() => handlePresetClick(preset.value)}
+            label={preset.label}
+            description={getPresetDescription(preset)}
+            sx={{minWidth: 150  }}
+            />
+          ))}
+          {isCustom && selectedChunker && (
+            <AdvancedConfigCard
+            modelName={selectedChunker.name}
+            onClick={() => setShowAdvanced(true)}
+            />
+          )}
+        </Box>
+
+        <Button
+          variant="contained"
+          color="primary"
+          size="small"
+          onClick={() => setShowAdvanced(true)}
+          disabled={!selectedChunker}
+          sx=
+          {{
+            alignSelf: "flex-start", 
+            width: "fit-content",
+            border: "1px solid",
+            borderColor: theme.palette.primary.main,
+            backgroundColor: theme.palette.action.selected,
+            color: theme.palette.text.primary,
+          }}
+          >
+           ↗ {t("generative:rag.chunking.advancedButton")}
+        </Button>
+
+        {selectedChunker && (
+          <ChunkingAdvancedModal
+          open={showAdvanced}
+          onClose={() => setShowAdvanced(false)}
+          chunkingModel={chunkingModel}
+          setChunkingModel={setChunkingModel}
+          />
+        )}
+    </RAGSectionColumn>
+  );
+}
+
+
+ChunkingSection.propTypes = {
+  chunkingModel: PropTypes.object,
+  setChunkingModel: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/sections/GeneratorSection.jsx b/DashAI/front/src/pages/generative/RAGSession/sections/GeneratorSection.jsx
new file mode 100644
index 000000000..0e8576287
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/sections/GeneratorSection.jsx
@@ -0,0 +1,241 @@
+import { useState, useEffect, useMemo } from "react";
+import {
+  Box,
+  Typography,
+  Autocomplete,
+  TextField,
+  Button,
+  Alert,
+  AlertTitle,
+  useTheme
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getGeneratorComponents } from "../../../../api/rag";
+import { resolveDefaults } from "../../../../utils/schema";
+import GeneratorAdvancedModal from "../advanced/GeneratorAdvancedModal";
+import AdvancedConfigCard from "../components/AdvancedConfigCard";
+import RAGSectionColumn from "../components/RAGSectionColumn";
+import { getDescription } from "../components/sectionUtils";
+
+export default function GeneratorSection({
+  generatorModel,
+  setGeneratorModel,
+  chunkSize = 0,
+  topK = 0,
+  promptTokenCount = 0,
+  setIsValid,
+}) {
+  const { t, i18n } = useTranslation(["generative"]);
+  const [generators, setGenerators] = useState([]);
+  const [selectedGenerator, setSelectedGenerator] = useState(null);
+  const [showAdvanced, setShowAdvanced] = useState(false);
+  const [loading, setLoading] = useState(true);
+  const DEFAULT_CONTEXT_WINDOW = 10000;
+  const DEFAULT_MAX_TOKENS = 1000;
+  const theme = useTheme();
+
+  const contextStats = useMemo(() => {
+    if (!generatorModel?.params || !selectedGenerator || !generatorModel.component) {
+      return { isValid: false, availableTokens: 0 };
+    }
+
+    const contextWindow = generatorModel.params.context_window ?? DEFAULT_CONTEXT_WINDOW;
+    const maxTokens = generatorModel.params.max_tokens ?? DEFAULT_MAX_TOKENS;
+    const chunkTokens = chunkSize * topK;
+    const availableForMessage = contextWindow - chunkTokens - promptTokenCount - maxTokens;
+    const isValid = availableForMessage > 0;
+
+    return {
+      isValid,
+      availableTokens: Math.max(0, Math.floor(availableForMessage))
+    };
+  }, [generatorModel?.params, generatorModel?.component, selectedGenerator, chunkSize, topK, promptTokenCount]);
+
+  const isRemoteModel = useMemo(() => {
+    if (!selectedGenerator || !generatorModel?.component) return false;
+    const name = (selectedGenerator.name || "").toLowerCase();
+    return name.includes("openai") || name.includes("deepseek");
+  }, [selectedGenerator, generatorModel?.component]);
+
+  const isApiKeyMissing = useMemo(() => {
+    if (!isRemoteModel || !generatorModel?.params) return false;
+    return !generatorModel.params.API_key;
+  }, [isRemoteModel, generatorModel?.params]);
+
+  const overallIsValid = contextStats.isValid && !isApiKeyMissing;
+
+  useEffect(() => {
+    setIsValid(overallIsValid);
+  }, [overallIsValid, setIsValid]);
+
+  useEffect(() => {
+    const loadGenerators = async () => {
+      try {
+        const data = await getGeneratorComponents();
+        setGenerators(data || []);
+      } catch (error) {
+        console.error("Error loading generators:", error);
+      } finally {
+        setLoading(false);
+      }
+    };
+    loadGenerators();
+  }, []);
+
+  useEffect(() => {
+    if (!generators.length) return;
+
+    if (!generatorModel?.component) {
+      return;
+    }
+
+    const foundGenerator = generators.find((generator) => {
+      return generator.name === generatorModel.component;
+    });
+
+    if (!foundGenerator) return;
+
+    const selectedGeneratorName = selectedGenerator?.name || null;
+    const currentGeneratorName = generatorModel.component;
+
+    if (selectedGeneratorName === currentGeneratorName) {
+      return;
+    }
+
+    setSelectedGenerator(foundGenerator);
+  }, [generators, generatorModel?.component, generatorModel?.params]);
+
+  const handleGeneratorChange = async (event, newValue) => {
+    if (!newValue) {
+      setSelectedGenerator(null);
+      setGeneratorModel({ component: "", params: {} });
+      return;
+    }
+
+    const initialParams = await resolveDefaults(newValue.name);
+    const overriddenParams = {
+      ...initialParams,
+      max_tokens: DEFAULT_MAX_TOKENS,
+      context_window: DEFAULT_CONTEXT_WINDOW,
+    };
+
+    setSelectedGenerator(newValue);
+    setGeneratorModel({
+      component: newValue.name,
+      params: { ...overriddenParams },
+    });
+  };
+
+  if (loading) {
+    return null;
+  }
+
+  return (
+    <RAGSectionColumn>
+      <Typography variant="body2" color="textSecondary">
+        {t("generative:rag.generator.description")}
+      </Typography>
+
+      <Box>
+        <Autocomplete
+          options={generators}
+          value={selectedGenerator}
+          onChange={handleGeneratorChange}
+          getOptionLabel={(option) => getDescription(option.display_name, i18n) || option.name || ""}
+          isOptionEqualToValue={(option, value) => option?.name === value?.name}
+          renderInput={(params) => (
+            <TextField
+              {...params}
+              label={t("generative:rag.generator.selectModel")}
+              placeholder={t("generative:rag.generator.selectModelPlaceholder")}
+            />
+          )}
+        />
+      </Box>
+
+      {selectedGenerator && (
+        <Box
+          sx={{
+            p: 2,
+            backgroundColor: "action.hover",
+            border: "1px solid",
+            borderColor: contextStats.isValid ? "divider" : "error.main",
+            borderRadius: 1,
+            display: "flex",
+            flexDirection: "column",
+            gap: 1
+          }}
+        >
+          <Typography variant="body2">
+            <Box component="span" sx={{ typography: "subtitle2" }}>
+              {t("generative:rag.generator.modelInfo")}
+            </Box> {selectedGenerator.name}
+          </Typography>
+
+          <Typography variant="body2" sx={{ color: contextStats.isValid ? "success.main" : "error.main" }}>
+            {t("generative:validation.contextSpace", { availableChars: contextStats.availableTokens?.toLocaleString() })}
+          </Typography>
+
+          {!contextStats.isValid && (
+            <Alert severity="error" sx={{ mt: 1 }}>
+              <AlertTitle>{t("generative:validation.insufficientContextTitle")}</AlertTitle>
+              {t("generative:validation.insufficientContextDescription")}
+            </Alert>
+          )}
+
+          {isApiKeyMissing && (
+            <Alert severity="warning" sx={{ mt: 1 }}>
+              <AlertTitle>{t("generative:validation.apiKeyMissingTitle")}</AlertTitle>
+              {t("generative:validation.apiKeyMissingDescription")}
+            </Alert>
+          )}
+
+          {getDescription(selectedGenerator.description, i18n) && (
+            <Typography variant="caption" display="block" sx={{ mt: 0.5 }}>
+              {getDescription(selectedGenerator.description, i18n)}
+            </Typography>
+          )}
+        </Box>
+      )}
+      {selectedGenerator && (
+        <Button
+          variant="contained"
+          color="primary"
+          size="small"
+          onClick={() => setShowAdvanced(true)}
+          disabled={!selectedGenerator}
+          sx={{
+            alignSelf: "flex-start",
+            width: "fit-content",
+            border: "1px solid",
+            borderColor: theme.palette.primary.main,
+            backgroundColor: theme.palette.action.selected,
+            color: theme.palette.text.primary,
+          }}
+        >
+          ↗ {t("generative:rag.generator.advancedButton")}
+        </Button>
+      )}
+
+      {selectedGenerator && (
+        <GeneratorAdvancedModal
+          open={showAdvanced}
+          onClose={() => setShowAdvanced(false)}
+          selectedGenerator={selectedGenerator}
+          generatorModel={generatorModel}
+          setGeneratorModel={setGeneratorModel}
+        />
+      )}
+    </RAGSectionColumn>
+  );
+}
+
+GeneratorSection.propTypes = {
+  generatorModel: PropTypes.object,
+  setGeneratorModel: PropTypes.func.isRequired,
+  chunkSize: PropTypes.number,
+  topK: PropTypes.number,
+  promptTokenCount: PropTypes.number,
+  setIsValid: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/sections/PromptSection.jsx b/DashAI/front/src/pages/generative/RAGSession/sections/PromptSection.jsx
new file mode 100644
index 000000000..a666741f5
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/sections/PromptSection.jsx
@@ -0,0 +1,404 @@
+import { useState, useEffect, useCallback, useMemo, useRef } from "react";
+import {
+  Box,
+  Typography,
+  Autocomplete,
+  TextField,
+  Button,
+  MenuItem,
+  useTheme,
+} from "@mui/material";
+import { AddCircleOutline as AddIcon } from "@mui/icons-material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import {
+  getRAGPrompts,
+  getDefaultPrompts,
+  isGenerationPromptClass,
+} from "../../../../api/rag";
+import NewPromptModal from "../advanced/NewPromptModal";
+import RAGSectionColumn from "../components/RAGSectionColumn";
+import { getDescription, renderTemplateWithHighlights } from "../components/sectionUtils";
+
+import { LANGUAGE_CODES } from "../../../../constants/languages";
+
+const CREATE_NEW_ID = "__create-new__";
+const DEFAULT_IDS = {
+  DefaultRAGGenerationPrompt: "default-generation",
+  DefaultQARAGGenerationPrompt: "default-QA",
+};
+
+function getDefaultDisplayName(option, t) {
+  // Use class_name (set explicitly by our code) rather than name
+  // (raw API field) to avoid potential encoding / serialization mismatches.
+  const cname = option.class_name || option.name || "";
+  if (cname.includes("DefaultQARAGGenerationPrompt")) {
+    return t("generative:rag.prompt.defaultQAGenerationPrompt");
+  }
+  if (cname.includes("DefaultRAGGenerationPrompt")) {
+    return t("generative:rag.prompt.defaultGenerationPrompt");
+  }
+  return option.name || cname;
+}
+
+function getOptionLabel(option, t) {
+  if (option._isCreateNew) return option.name;
+  if (option._isDefault) return getDefaultDisplayName(option, t);
+  return option.name;
+}
+
+export default function PromptSection({
+  promptModel,
+  setPromptModel,
+  onTokenCountChange,
+}) {
+  const { t, i18n } = useTranslation(["generative"]);
+  const theme = useTheme();
+  const placeholderColors = useMemo(
+    () => ({
+      bg: theme.palette.placeholder?.bg || theme.palette.warning.light,
+      text: theme.palette.placeholder?.text || theme.palette.warning.dark,
+    }),
+    [theme],
+  );
+  const platformLang = useMemo(() => {
+    const lang = (i18n.language || "en").split("-")[0];
+    return ["en", "es", "pt"].includes(lang) ? lang : "en";
+  }, [i18n.language]);
+  const [customPrompts, setCustomPrompts] = useState([]);
+  const [defaultPrompts, setDefaultPrompts] = useState([]);
+  const [selectedPrompt, setSelectedPrompt] = useState(null);
+  const [newPromptModalOpen, setNewPromptModalOpen] = useState(false);
+  const [loading, setLoading] = useState(true);
+  const [selectedLanguage, setSelectedLanguage] = useState(platformLang);
+  const prevSelectedRef = useRef(null);
+  const isInitializedRef = useRef(false);
+
+  const loadPrompts = useCallback(async () => {
+    try {
+      const dbPrompts = await getRAGPrompts();
+      // Keep only user-created generation prompts:
+      // 1. Exclude system defaults (class_name starts with "Default")
+      // 2. Exclude augmentation prompts (wrong type for this selector)
+      setCustomPrompts(
+        (dbPrompts || []).filter(
+          (p) =>
+            !p.class_name.startsWith("Default") &&
+            isGenerationPromptClass(p.class_name),
+        ),
+      );
+    } catch (error) {
+      console.error("Error loading custom RAG prompts:", error);
+      setCustomPrompts([]);
+    }
+
+    try {
+      const defaultData = await getDefaultPrompts();
+      // Filter out CustomRAGGenerationPrompt — it's a base class for
+      // user-created prompts, not a selectable template. It has no
+      // templates in its metadata and inherits a generic description
+      // ("Base class for RAG prompts.") from Prompt.
+      setDefaultPrompts(
+        (defaultData || []).filter(
+          (dp) => dp.name !== "CustomRAGGenerationPrompt",
+        ),
+      );
+    } catch (error) {
+      console.error("Error loading default RAG prompts:", error);
+      setDefaultPrompts([]);
+    }
+  }, []);
+
+  useEffect(() => {
+    const load = async () => {
+      await loadPrompts();
+      setLoading(false);
+    };
+    load();
+  }, [loadPrompts]);
+
+  const mergedOptions = useMemo(() => {
+    const defaults = defaultPrompts.map((dp) => ({
+      ...dp,
+      id: DEFAULT_IDS[dp.name] || dp.name,
+      _isDefault: true,
+      class_name: dp.name,
+    }));
+    const customs = customPrompts.map((cp) => ({
+      ...cp,
+      _isDefault: false,
+    }));
+    return [
+      ...defaults,
+      ...customs,
+      {
+        id: CREATE_NEW_ID,
+        name: t("generative:rag.prompt.createNewPrompt"),
+        _isCreateNew: true,
+        _isDefault: false,
+        class_name: "",
+      },
+    ];
+  }, [defaultPrompts, customPrompts, t]);
+
+  const currentTemplate = useMemo(() => {
+    if (!selectedPrompt) return "";
+    if (selectedPrompt._isDefault) {
+      return selectedPrompt.metadata?.templates?.[selectedLanguage] || "";
+    }
+    if (selectedPrompt.parameters?.templates) {
+      return selectedPrompt.parameters.templates[selectedLanguage] || "";
+    }
+    return selectedPrompt.parameters?.template || "";
+  }, [selectedPrompt, selectedLanguage]);
+
+  const isDefault = selectedPrompt?._isDefault;
+
+  useEffect(() => {
+    if (!selectedPrompt) {
+      if (onTokenCountChange) onTokenCountChange(0);
+      return;
+    }
+
+    if (!isInitializedRef.current && promptModel?.component === (selectedPrompt.class_name || selectedPrompt.name)) {
+      isInitializedRef.current = true;
+      return;
+    }
+    isInitializedRef.current = true;
+
+    setPromptModel({
+      component: selectedPrompt.class_name || selectedPrompt.name,
+      params: {
+        template: currentTemplate,
+        language: selectedLanguage,
+        ...(selectedPrompt._isDefault || selectedPrompt.parameters?.templates
+          ? { templates: selectedPrompt.parameters?.templates }
+          : {}),
+      },
+    });
+    if (onTokenCountChange) {
+      const tokenCount = Math.ceil(currentTemplate.length / 4);
+      onTokenCountChange(tokenCount);
+    }
+  }, [selectedPrompt, selectedLanguage]);
+
+  useEffect(() => {
+    const selectable = mergedOptions.filter((o) => !o._isCreateNew);
+    if (!selectable.length) return;
+
+    if (promptModel?.component) {
+      const found = selectable.find((p) => {
+        if (p._isDefault) {
+          return p.class_name === promptModel.component;
+        }
+        return (
+          p.class_name === promptModel.component &&
+          p.parameters?.template === promptModel.params?.template
+        );
+      });
+      if (found?.id !== selectedPrompt?.id) {
+        setSelectedPrompt(found || null);
+        prevSelectedRef.current = found || null;
+        if (found?._isDefault) {
+          setSelectedLanguage(promptModel.params?.language || platformLang);
+        }
+        isInitializedRef.current = false;
+      }
+      return;
+    }
+
+    if (!selectedPrompt) {
+      const firstDefault =
+        selectable.find((p) => p._isDefault) || selectable[0];
+      if (firstDefault) {
+        setSelectedPrompt(firstDefault);
+        setSelectedLanguage(platformLang);
+        prevSelectedRef.current = firstDefault;
+      }
+    }
+  }, [mergedOptions, promptModel]);
+
+  const handlePromptChange = (_event, newValue) => {
+    if (newValue?._isCreateNew) {
+      setNewPromptModalOpen(true);
+      setSelectedPrompt(prevSelectedRef.current);
+      return;
+    }
+    prevSelectedRef.current = newValue;
+    setSelectedPrompt(newValue);
+    if (newValue?._isDefault) {
+      setSelectedLanguage(platformLang);
+    }
+  };
+
+  const handleLanguageChange = (event) => {
+    setSelectedLanguage(event.target.value);
+  };
+
+  const handlePromptCreated = useCallback(
+    async (newPromptId) => {
+      const updatedPrompts = await getRAGPrompts();
+      // Apply same filter as loadPrompts: exclude system defaults and augmentation
+      setCustomPrompts(
+        (updatedPrompts || []).filter(
+          (p) =>
+            !p.class_name.startsWith("Default") &&
+            isGenerationPromptClass(p.class_name),
+        ),
+      );
+
+      const newPrompt = (updatedPrompts || []).find(
+        (p) => p.id === newPromptId,
+      );
+      if (newPrompt) {
+        const wrapped = { ...newPrompt, _isDefault: false };
+        setSelectedPrompt(wrapped);
+        prevSelectedRef.current = wrapped;
+        // Sync language from the newly created prompt's saved language
+        // so local selectedLanguage stays aligned with the persisted value.
+        if (newPrompt.parameters?.language) {
+          setSelectedLanguage(newPrompt.parameters.language);
+        }
+        // Sync parent promptModel immediately to prevent infinite
+        // useEffect 1 <-> useEffect 2 ping-pong when the new prompt's
+        // class_name differs from the previously selected prompt.
+        setPromptModel({
+          component: newPrompt.class_name,
+          params: {
+            template: newPrompt.parameters?.template || "",
+            language: newPrompt.parameters?.language || "",
+          },
+        });
+        const template = newPrompt.parameters?.template || "";
+        const tokenCount = Math.ceil(template.length / 4);
+        onTokenCountChange?.(tokenCount);
+      }
+
+      setNewPromptModalOpen(false);
+    },
+    [onTokenCountChange, setPromptModel],
+  );
+
+  if (loading) {
+    return null;
+  }
+
+  return (
+    <RAGSectionColumn>
+      <Typography variant="body2" color="textSecondary">
+        {t("generative:rag.prompt.description")}
+      </Typography>
+
+            <Button
+        variant="contained"
+        color="primary"
+        size="small"
+        startIcon={<AddIcon />}
+        onClick={() => setNewPromptModalOpen(true)}
+        sx={{
+          alignSelf: "flex-start", 
+          width: "fit-content",
+          border: "1px solid",
+          borderColor: theme.palette.primary.main,
+          backgroundColor: theme.palette.action.selected,
+          color: theme.palette.text.primary,
+        }}
+      >
+        {t("generative:rag.prompt.newPromptButton")}
+      </Button>
+
+      <Box>
+        <Autocomplete
+          options={mergedOptions}
+          value={selectedPrompt}
+          onChange={handlePromptChange}
+          getOptionLabel={(option) => getOptionLabel(option, t)}
+          isOptionEqualToValue={(option, value) => option.id === value.id}
+          renderOption={(props, option) => (
+            <li {...props}>{getOptionLabel(option, t)}</li>
+          )}
+          renderInput={(params) => (
+            <TextField
+              {...params}
+              label={t("generative:rag.prompt.selectTemplate")}
+              placeholder={t(
+                "generative:rag.prompt.selectTemplatePlaceholder",
+              )}
+            />
+          )}
+          sx={{}}
+        />
+      </Box>
+
+      {(isDefault || selectedPrompt?.parameters?.templates) && (
+        <TextField
+          select
+          fullWidth
+          label={t("generative:rag.prompt.language")}
+          value={selectedLanguage}
+          onChange={handleLanguageChange}
+          size="small"
+        >
+          {LANGUAGE_CODES.map((code) => (
+            <MenuItem key={code} value={code}>
+              {t(`generative:rag.prompt.languages.${code}`)}
+            </MenuItem>
+          ))}
+        </TextField>
+      )}
+
+      {selectedPrompt && (
+        <RAGSectionColumn>
+          <Typography variant="subtitle2" >
+            {t("generative:rag.prompt.selectedTemplate")}
+          </Typography>
+
+          <Box
+            sx={{
+              p: 2,
+              backgroundColor: "background.paper",
+              border: "1px solid",
+              borderColor: "divider",
+              borderRadius: 1,
+              fontFamily: theme.typography.code.fontFamily,
+              lineHeight: 1.6,
+              whiteSpace: "pre-wrap",
+              wordBreak: "break-word",
+            }}
+          >
+            {renderTemplateWithHighlights(currentTemplate, placeholderColors, theme.typography.code.fontFamily)}
+          </Box>
+
+          {getDescription(selectedPrompt.description, i18n) && (
+            <Typography
+              variant="caption"
+              display="block"
+              sx={{ color: "text.secondary" }}
+            >
+              {getDescription(selectedPrompt.description, i18n)}
+            </Typography>
+          )}
+        </RAGSectionColumn>
+      )}
+
+      <NewPromptModal
+        open={newPromptModalOpen}
+        handleClose={() => setNewPromptModalOpen(false)}
+        onPromptCreated={handlePromptCreated}
+        existingPrompts={customPrompts}
+      />
+    </RAGSectionColumn>
+  );
+}
+
+PromptSection.propTypes = {
+  promptModel: PropTypes.shape({
+    component: PropTypes.string,
+    params: PropTypes.shape({
+      template: PropTypes.string,
+      language: PropTypes.string,
+    }),
+  }),
+  setPromptModel: PropTypes.func.isRequired,
+  onTokenCountChange: PropTypes.func,
+};
diff --git a/DashAI/front/src/pages/generative/RAGSession/sections/RetrieverSection.jsx b/DashAI/front/src/pages/generative/RAGSession/sections/RetrieverSection.jsx
new file mode 100644
index 000000000..543745e78
--- /dev/null
+++ b/DashAI/front/src/pages/generative/RAGSession/sections/RetrieverSection.jsx
@@ -0,0 +1,416 @@
+import { useState, useEffect, useMemo, useCallback } from "react";
+import {
+  Box,
+  Typography,
+  Button,
+  CircularProgress,
+  useTheme,
+} from "@mui/material";
+import PropTypes from "prop-types";
+import { useTranslation } from "react-i18next";
+import { getRetrieverComponents } from "../../../../api/rag";
+import { resolveDefaults } from "../../../../utils/schema";
+import RetrieverAdvancedModal from "../advanced/RetrieverAdvancedModal";
+import AdvancedConfigCard from "../components/AdvancedConfigCard";
+import PresetCard from "../components/PresetCard";
+import RAGSectionColumn from "../components/RAGSectionColumn";
+
+const TOP_K_OPTIONS = [3, 5, 10, 15, 20];
+
+/**
+ * Deep equality comparison for objects, arrays, and primitives.
+ * Handles nested structures recursively. Returns false for
+ * different types, null vs object, or arrays vs objects.
+ */
+function deepEqual(a, b) {
+  if (a === b) return true;
+  if (typeof a !== typeof b) return false;
+  if (a === null || b === null) return false;
+  if (typeof a !== "object") return false;
+  if (Array.isArray(a) !== Array.isArray(b)) return false;
+  const keysA = Object.keys(a);
+  const keysB = Object.keys(b);
+  if (keysA.length !== keysB.length) return false;
+  return keysA.every((k) => deepEqual(a[k], b[k]));
+}
+
+/**
+ * Omits `top_k` key from an object for purpose of comparing
+ * retriever configurations while ignoring the top-K value.
+ */
+const omitTopK = (obj) =>
+  Object.fromEntries(
+    Object.entries(obj || {}).filter(([k]) => k !== "top_k"),
+  );
+
+function getEffectiveTopK(model) {
+  if (!model?.params) return null;
+  if (model.component === "ParallelRetriever") {
+    const children = model.params.children || [];
+    if (children.length === 0) return null;
+    return children.reduce((sum, child) => sum + (child.params?.top_k || 0), 0);
+  }
+  if (model.component === "SequentialRetriever") {
+    const children = model.params.children || [];
+    if (children.length > 0) {
+      return children[children.length - 1].params?.top_k || null;
+    }
+    return null;
+  }
+  if (model.component === "MMRRerankerRetriever") {
+    return typeof model.params.top_k === "number" ? model.params.top_k : null;
+  }
+  return typeof model.params.top_k === "number" ? model.params.top_k : null;
+}
+
+function buildKeywordModel(bm25Defaults, topK) {
+  return {
+    component: "BM25Retriever",
+    params: { ...bm25Defaults, top_k: topK },
+  };
+}
+
+function buildSemanticModel(embeddingDefaults, retrieverDefaults, topK) {
+  return {
+    component: "DenseEmbeddingRetriever",
+    params: {
+      embedding_model: {
+        component: "SentenceTransformerEmbedding",
+        params: { ...embeddingDefaults, model_name: "microsoft/harrier-oss-v1-0.6b" },
+      },
+      similarity_metric: retrieverDefaults.similarity_metric || "cosine",
+      top_k: topK,
+    },
+  };
+}
+
+function buildHybridModel(bm25Defaults, embeddingDefaults, retrieverDefaults, topK) {
+  const kwTopK = Math.ceil(topK / 2);
+  const seTopK = Math.floor(topK / 2);
+  return {
+    component: "ParallelRetriever",
+    params: {
+      merge_strategy: "round_robin",
+      children: [
+        {
+          component: "BM25Retriever",
+          params: { ...bm25Defaults, top_k: kwTopK },
+        },
+        {
+          component: "DenseEmbeddingRetriever",
+          params: {
+            embedding_model: {
+              component: "SentenceTransformerEmbedding",
+              params: { ...embeddingDefaults, model_name: "microsoft/harrier-oss-v1-0.6b" },
+            },
+            similarity_metric: retrieverDefaults.similarity_metric || "cosine",
+            top_k: seTopK,
+          },
+        },
+      ],
+    },
+  };
+}
+
+export default function RetrieverSection({
+  retrieverModel,
+  setRetrieverModel,
+}) {
+  const theme = useTheme();
+  const { t } = useTranslation(["generative"]);
+
+  const [selectedGroup, setSelectedGroup] = useState(null);
+  const [topK, setTopK] = useState(10);
+  const [showAdvanced, setShowAdvanced] = useState(false);
+  const [loading, setLoading] = useState(true);
+  const [concreteRetrievers, setConcreteRetrievers] = useState([]);
+  const [bm25Defaults, setBm25Defaults] = useState({});
+  const [embeddingDefaults, setEmbeddingDefaults] = useState({});
+  const [denseRetrieverDefaults, setDenseRetrieverDefaults] = useState({});
+
+  const effectiveTopK = getEffectiveTopK(retrieverModel);
+
+  const keywordModel = useMemo(
+    () => buildKeywordModel(bm25Defaults, effectiveTopK || topK),
+    [bm25Defaults, effectiveTopK, topK],
+  );
+
+  const semanticModel = useMemo(
+    () => buildSemanticModel(embeddingDefaults, denseRetrieverDefaults, effectiveTopK || topK),
+    [embeddingDefaults, denseRetrieverDefaults, effectiveTopK, topK],
+  );
+
+  const hybridModel = useMemo(
+    () => buildHybridModel(bm25Defaults, embeddingDefaults, denseRetrieverDefaults, effectiveTopK || topK),
+    [bm25Defaults, embeddingDefaults, denseRetrieverDefaults, effectiveTopK, topK],
+  );
+
+  const isAdvanced = useMemo(() => {
+    if (!retrieverModel?.component || !retrieverModel?.params) return false;
+
+    const modelNoTopK = {
+      component: retrieverModel.component,
+      params: omitTopK(retrieverModel.params),
+    };
+
+    const kwNoTopK = {
+      component: keywordModel.component,
+      params: omitTopK(keywordModel.params),
+    };
+
+    const seNoTopK = {
+      component: semanticModel.component,
+      params: omitTopK(semanticModel.params),
+    };
+
+    const hyNoTopK = {
+      component: hybridModel.component,
+      params: omitTopK(hybridModel.params),
+    };
+
+    return !deepEqual(modelNoTopK, kwNoTopK)
+      && !deepEqual(modelNoTopK, seNoTopK)
+      && !deepEqual(modelNoTopK, hyNoTopK);
+  }, [retrieverModel, keywordModel, semanticModel, hybridModel]);
+
+  const detectSelectedGroup = useCallback((model) => {
+    if (!model?.component) return null;
+
+    const modelNoTopK = {
+      component: model.component,
+      params: omitTopK(model.params),
+    };
+
+    if (deepEqual(modelNoTopK, {
+      component: keywordModel.component,
+      params: omitTopK(keywordModel.params),
+    })) {
+      return "keyword";
+    }
+    if (deepEqual(modelNoTopK, {
+      component: semanticModel.component,
+      params: omitTopK(semanticModel.params),
+    })) {
+      return "semantic";
+    }
+    if (deepEqual(modelNoTopK, {
+      component: hybridModel.component,
+      params: omitTopK(hybridModel.params),
+    })) {
+      return "hybrid";
+    }
+    return null;
+  }, [keywordModel, semanticModel, hybridModel]);
+
+  useEffect(() => {
+    const tk = getEffectiveTopK(retrieverModel);
+    if (tk != null) {
+      setTopK(tk);
+    }
+  }, [retrieverModel]);
+
+  useEffect(() => {
+    if (loading) return;
+    const group = detectSelectedGroup(retrieverModel);
+    setSelectedGroup(group);
+  }, [retrieverModel, loading, detectSelectedGroup]);
+
+  useEffect(() => {
+    const load = async () => {
+      try {
+        const allRetrieversRaw = await getRetrieverComponents("RetrieverModel");
+        const retrievers = allRetrieversRaw.filter(
+          (c) => !(c.flags || []).includes("abstract"),
+        );
+        setConcreteRetrievers(retrievers);
+
+        const [bm25, embeddingDefaultsRes, denseRetriever] = await Promise.all([
+          resolveDefaults("BM25Retriever").catch(() => ({})),
+          resolveDefaults("SentenceTransformerEmbedding").catch(() => ({})),
+          resolveDefaults("DenseEmbeddingRetriever").catch(() => ({})),
+        ]);
+
+        setBm25Defaults(bm25);
+        setEmbeddingDefaults(embeddingDefaultsRes);
+        setDenseRetrieverDefaults(denseRetriever);
+      } catch (error) {
+        console.error("Error loading retrievers:", error);
+      } finally {
+        setLoading(false);
+      }
+    };
+    load();
+  }, []);
+
+  const selectPreset = useCallback((groupKey) => {
+    const alreadySelected = !isAdvanced && selectedGroup === groupKey;
+    if (alreadySelected) {
+      setShowAdvanced(true);
+      return;
+    }
+
+    setShowAdvanced(false);
+    setSelectedGroup(groupKey);
+
+    if (groupKey === "keyword") {
+      setRetrieverModel(keywordModel);
+    } else if (groupKey === "semantic") {
+      setRetrieverModel(semanticModel);
+    } else if (groupKey === "hybrid") {
+      setRetrieverModel(hybridModel);
+    }
+  }, [isAdvanced, selectedGroup, keywordModel, semanticModel, hybridModel, setRetrieverModel]);
+
+  const handleTopKChange = useCallback((newValue) => {
+    const value = parseInt(newValue);
+    if (isNaN(value) || value <= 0 || isAdvanced) return;
+    setTopK(value);
+
+    if (selectedGroup === "keyword") {
+      setRetrieverModel(buildKeywordModel(bm25Defaults, value));
+    } else if (selectedGroup === "semantic") {
+      setRetrieverModel(buildSemanticModel(embeddingDefaults, denseRetrieverDefaults, value));
+    } else if (selectedGroup === "hybrid") {
+      setRetrieverModel(buildHybridModel(bm25Defaults, embeddingDefaults, denseRetrieverDefaults, value));
+    }
+  }, [
+    isAdvanced,
+    selectedGroup,
+    bm25Defaults,
+    embeddingDefaults,
+    denseRetrieverDefaults,
+    setRetrieverModel,
+  ]);
+
+  if (loading) {
+    return (
+      <Box display="flex" justifyContent="center" alignItems="center" sx={{ minHeight: 120 }}>
+        <CircularProgress />
+      </Box>
+    );
+  }
+
+  return (
+    <RAGSectionColumn>
+      <Typography variant="body2" color="textSecondary">
+        {t("generative:rag.retriever.description")}
+      </Typography>
+
+      <Box sx={{ display: "flex", gap: 1, alignItems: "stretch", flexWrap: "wrap" }}>
+        <PresetCard
+          key="keyword"
+          selected={!isAdvanced && selectedGroup === "keyword"}
+          onClick={() => selectPreset("keyword")}
+          label={t("generative:rag.retriever.keywordLabel")}
+          description="BM25"
+          sx={{ minWidth: 180}}
+        />
+        <PresetCard
+          key="semantic"
+          selected={!isAdvanced && selectedGroup === "semantic"}
+          onClick={() => selectPreset("semantic")}
+          label={t("generative:rag.retriever.semanticLabel")}
+          description="Harrier OSS v1 0.6B"
+          sx={{ minWidth: 180}}
+        />
+        <PresetCard
+          key="hybrid"
+          selected={!isAdvanced && selectedGroup === "hybrid"}
+          onClick={() => selectPreset("hybrid")}
+          label={t("generative:rag.retriever.hybridLabel")}
+          description={t("generative:rag.retriever.hybridDescription")}
+          sx={{ minWidth: 180}}
+        />
+        {isAdvanced && retrieverModel?.component && (
+          <Box sx={{ flex: 1, minWidth: 180, display: "flex", flexDirection: "column", gap: 1 }}>
+            <AdvancedConfigCard
+              modelName={retrieverModel.component}
+              onClick={() => setShowAdvanced(true)}
+            />
+            {effectiveTopK != null && (
+              <Box
+                sx={{
+                  border: "1px solid",
+                  borderColor: theme.palette.ui.border,
+                  borderRadius: 1,
+                  backgroundColor: theme.palette.action.selected,
+                  display: "flex",
+                  alignItems: "center",
+                  justifyContent: "center",
+                  py: 1.5,
+                  px: 1,
+                }}
+              >
+                <Typography
+                  variant="body2"
+                  sx={{ color: theme.palette.text.primary }}
+                >
+                  {t("generative:rag.retriever.advancedTopK", {
+                    topK: effectiveTopK,
+                  })}
+                </Typography>
+              </Box>
+            )}
+          </Box>
+        )}
+      </Box>
+
+      {selectedGroup && (
+        <Box>
+          {!isAdvanced && (
+            <>
+              <Typography variant="subtitle2" sx={{ mb: 1 }}>
+                {t("generative:rag.retriever.topKLabel")}
+              </Typography>
+              <Box sx={{ display: "flex", gap: 1, alignItems: "stretch", flexWrap: "wrap", width: "100%" }}>
+                {TOP_K_OPTIONS.map((k) => (
+                  <PresetCard
+                    key={k}
+                    selected={topK === k}
+                    onClick={() => handleTopKChange(String(k))}
+                    label={String(k)}
+                    description={""}
+                    sx={{ flex: "1 1 0", minWidth: 0, justifyContent: "center", alignItems: "center" }}
+                  />
+                ))}
+              </Box>
+            </>
+          )}
+        </Box>
+      )}
+
+      <Button
+        variant="contained"
+        color="primary"
+        size="small"
+        onClick={() => setShowAdvanced(true)
+        }
+        sx=
+        {{
+          alignSelf: "flex-start", 
+          width: "fit-content",
+          border: "1px solid",
+          borderColor: theme.palette.primary.main,
+          backgroundColor: theme.palette.action.selected,
+          color: theme.palette.text.primary,
+        }}
+      >
+        ↗ {t("generative:rag.retriever.advancedButton")}
+      </Button>
+
+      <RetrieverAdvancedModal
+        open={showAdvanced}
+        onClose={() => setShowAdvanced(false)}
+        selectedParadigm={null}
+        allParadigms={concreteRetrievers}
+        retrieverModel={retrieverModel}
+        setRetrieverModel={setRetrieverModel}
+      />
+    </RAGSectionColumn>
+  );
+}
+
+RetrieverSection.propTypes = {
+  retrieverModel: PropTypes.object,
+  setRetrieverModel: PropTypes.func.isRequired,
+};
diff --git a/DashAI/front/src/pages/generative/SessionRouter.jsx b/DashAI/front/src/pages/generative/SessionRouter.jsx
new file mode 100644
index 000000000..ae8bf12bd
--- /dev/null
+++ b/DashAI/front/src/pages/generative/SessionRouter.jsx
@@ -0,0 +1,77 @@
+import { useState, useEffect, useRef } from "react";
+import { useParams } from "react-router-dom";
+import { Box, CircularProgress, Typography } from "@mui/material";
+import { getGenerativeSession } from "../../api/generativeTask";
+import Generative from "./Generative";
+import { GenerativeProvider } from "../../components/generative/GenerativeContext";
+import RAGSessionPage from "./RAGSession/RAGSessionPage";
+
+export default function SessionRouter() {
+  const { id } = useParams();
+  const [taskName, setTaskName] = useState(null);
+  const [loading, setLoading] = useState(false);
+  const prevViewRef = useRef({ taskName: null, loaded: false });
+
+  useEffect(() => {
+    const sid = Number(id);
+    if (!Number.isFinite(sid) || sid <= 0) {
+      setTaskName(null);
+      setLoading(false);
+      return;
+    }
+
+    let cancelled = false;
+    setLoading(true);
+
+    getGenerativeSession(sid)
+      .then((session) => {
+        if (!cancelled) {
+          const tn = session?.task_name || null;
+          setTaskName(tn);
+          prevViewRef.current = { taskName: tn, loaded: true };
+        }
+      })
+      .catch(() => {
+        if (!cancelled) {
+          setTaskName(null);
+          prevViewRef.current = { taskName: null, loaded: true };
+        }
+      })
+      .finally(() => {
+        if (!cancelled) setLoading(false);
+      });
+
+    return () => { cancelled = true; };
+  }, [id]);
+
+  const effectiveView = loading ? prevViewRef.current.taskName : taskName;
+  const isFirstLoad = !prevViewRef.current.loaded && loading;
+
+  if (isFirstLoad) {
+    return (
+      <Box display="flex" alignItems="center" justifyContent="center" height="100vh">
+        <CircularProgress />
+      </Box>
+    );
+  }
+
+  if (!effectiveView) {
+    return (
+      <Box display="flex" alignItems="center" justifyContent="center" height="100vh">
+        <Typography variant="h6" color="text.secondary">
+          Session not found
+        </Typography>
+      </Box>
+    );
+  }
+
+  if (effectiveView === "RAGTask") {
+    return (
+      <GenerativeProvider key={id}>
+        <RAGSessionPage />
+      </GenerativeProvider>
+    );
+  }
+
+  return <Generative key={id} />;
+}
diff --git a/DashAI/front/src/styles/theme.js b/DashAI/front/src/styles/theme.js
index 10502a175..f70dd83b9 100644
--- a/DashAI/front/src/styles/theme.js
+++ b/DashAI/front/src/styles/theme.js
@@ -94,6 +94,20 @@ const getTheme = (mode) => ({
       light: "#A7C7FF",
     },
 
+    // Placeholder highlight for prompt templates (chunks, input, etc.)
+    placeholder:
+      mode === "dark"
+        ? {
+            bg: "rgba(255, 183, 77, 0.15)",
+            text: "#FFB74D",
+            border: "rgba(255, 183, 77, 0.30)",
+          }
+        : {
+            bg: "rgba(249, 168, 37, 0.12)",
+            text: "#E65100",
+            border: "rgba(230, 81, 0, 0.25)",
+          },
+
     // Module accent colors — dark uses lighter variants, light uses darker variants
     accent: {
       // Datasets → orange-brown
diff --git a/DashAI/front/src/types/component.ts b/DashAI/front/src/types/component.ts
index 9bedf444b..eed427cf0 100644
--- a/DashAI/front/src/types/component.ts
+++ b/DashAI/front/src/types/component.ts
@@ -9,5 +9,6 @@ export interface IComponent {
   metadata: ITaskMetadataParameters;
   description: string;
   display_name?: string;
+  flags?: string[];
   color?: string;
 }
diff --git a/DashAI/front/src/types/documentResponse.ts b/DashAI/front/src/types/documentResponse.ts
new file mode 100644
index 000000000..10e4089e6
--- /dev/null
+++ b/DashAI/front/src/types/documentResponse.ts
@@ -0,0 +1,8 @@
+export interface IDocumentResponse {
+  id: number;
+  file_name: string;
+  file_type: string;
+  created: Date;
+  optional_metadata: Record<string, any> | null;
+  file_url: string;
+}
diff --git a/DashAI/front/src/types/ragPrompt.ts b/DashAI/front/src/types/ragPrompt.ts
new file mode 100644
index 000000000..38cd4ff73
--- /dev/null
+++ b/DashAI/front/src/types/ragPrompt.ts
@@ -0,0 +1,12 @@
+export interface IRAGPrompt {
+  id: number;
+  class_name: string;
+  name: string;
+  created?: string;
+  last_modified?: string;
+  parameters: {
+    template?: string;
+    templates?: Record<string, string>;
+    language?: string;
+  };
+}
diff --git a/DashAI/front/src/utils/i18n/locales/de/generative.json b/DashAI/front/src/utils/i18n/locales/de/generative.json
index 2446ff33d..4519a18ab 100644
--- a/DashAI/front/src/utils/i18n/locales/de/generative.json
+++ b/DashAI/front/src/utils/i18n/locales/de/generative.json
@@ -57,11 +57,377 @@
     "attachMedia": "Medien anhängen",
     "attachMediaToContinue": "Medien anhängen, um fortzufahren",
     "noInputAvailable": "Keine Eingabe für diese Aufgabe verfügbar",
-    "selectModelCrumb": "Modell auswählen"
+    "selectModelCrumb": "Modell auswählen",
+    "ragTask": "RAG - Retrieval Augmented Generation",
+    "ragTaskDescription": "Laden Sie Dokumente hoch und chatten Sie mit ihnen durch Retrieval-Augmented Generation."
   },
   "message": {
     "sessionCreatedSuccess": "Sitzung erfolgreich erstellt.",
     "sessionDeleted": "Sitzung erfolgreich gelöscht",
     "sessionUpdated": "Sitzung erfolgreich aktualisiert"
+  },
+  "rag": {
+    "setup": {
+      "title": "RAG-Sitzung erstellen",
+      "subtitle": "Geben Sie grundlegende Informationen an, wählen Sie Dokumente aus und konfigurieren Sie die Komponenten für Ihre RAG-Sitzung.",
+      "sessionDetails": "Sitzungsdetails",
+      "sessionName": "Sitzungsname *",
+      "sessionNamePlaceholder": "z. B. Produktdokumentation RAG",
+      "sessionNameDuplicate": "Dieser Sitzungsname existiert bereits",
+      "sessionNameDuplicateAlert": "Eine Sitzung mit diesem Namen existiert bereits. Bitte verwenden Sie einen anderen Namen.",
+      "description": "Beschreibung (Optional)",
+      "descriptionPlaceholder": "Beschreiben Sie den Zweck dieser RAG-Sitzung...",
+      "selectDocuments": "Dokumente auswählen",
+      "selectDocumentsDescription": "Laden Sie neue Dokumente hoch oder wählen Sie aus vorhandenen, um sie für RAG zu verwenden.",
+      "chunkingStrategy": "Chunking-Strategie",
+      "retrieverModel": "Retriever-Modell",
+      "promptTemplate": "Prompt-Vorlage",
+      "languageModel": "Sprachmodell (LLM)",
+      "cancel": "Abbrechen",
+      "saveSession": "Sitzung speichern",
+      "openDocumentsLibrary": "Dokumentbibliothek öffnen"
+    },
+    "validation": {
+      "nameRequired": "Sitzungsname ist erforderlich",
+      "nameUnique": "Sitzungsname muss eindeutig sein",
+      "documentsRequired": "Mindestens ein Dokument muss ausgewählt werden",
+      "chunkingRequired": "Chunking-Modell muss konfiguriert werden",
+      "retrieverRequired": "Retriever-Modell muss konfiguriert werden",
+      "generatorRequired": "Generator-Modell muss konfiguriert werden",
+      "generatorInvalid": "Generatorkonfiguration ist ungültig (Kontextfenster prüfen)",
+      "promptRequired": "Prompt-Vorlage muss ausgewählt werden"
+    },
+    "messages": {
+      "success": "RAG-Sitzung erfolgreich erstellt"
+    },
+    "chunking": {
+      "description": "Wählen Sie, wie Ihre Dokumente in Chunks für die Abfrage aufgeteilt werden.",
+      "presets": {
+        "small": {
+          "label": "Kleine Chunks"
+        },
+        "paragraph": {
+          "label": "Absatzlänge"
+        },
+        "page": {
+          "label": "Seiten-Chunk"
+        },
+        "large": {
+          "label": "Große Abschnitte"
+        },
+        "chunkSizeFormat": "{{chars}} Zeichen ≈ {{tokens}} Tokens",
+        "custom": {
+          "label": "Benutzerdefiniertes Chunking-Modell",
+          "description": "Erweiterte Konfiguration angewendet"
+        }
+      },
+      "advancedButton": "Erweiterte Konfiguration öffnen"
+    },
+    "retriever": {
+      "description": "Wählen Sie Ihr Abrufparadigma. Der Retriever durchsucht Ihre Dokumente und gibt die relevantesten Chunks zurück.",
+      "paradigmLabel": "Abrufparadigma",
+      "keywordLabel": "Stichwort",
+      "semanticLabel": "Semantisch",
+      "hybridLabel": "Hybrid",
+      "hybridDescription": "BM25 + Harrier 0.6B",
+      "advancedApplied": "Erweiterte Konfiguration angewendet",
+      "topKLabel": "Anzahl der Chunks (Top K)",
+      "topKCardDescription": "Chunks",
+      "advancedTopK": "Erweitertes Top K: {{topK}}",
+      "advancedButton": "Erweiterte Konfiguration öffnen",
+      "modifyFromAdvanced": "Top K muss über die erweiterte Konfigurationsansicht geändert werden",
+      "explanations": {
+        "SparseRetriever": "Am besten für bestimmte Stichwort- und ID-Suche. Nützlich zum Finden exakter Übereinstimmungen in Dokumenten.",
+        "DenseRetriever": "Kontextbezogene Suche, die Bedeutung versteht. Ideal zum Finden semantisch ähnlicher Inhalte, kann aber bestimmte Stichwörter oder IDs übersehen.",
+        "HybridRetriever": "Kombiniert Stichwort- und kontextbezogene Suche für ausgewogene Leistung bei verschiedenen Abfragetypen."
+      }
+    },
+    "composite": {
+      "addChild": "Retriever-Knoten hinzufügen",
+      "removeNode": "Knoten entfernen",
+      "configureNode": "Knoten konfigurieren",
+      "selectModel": "Retriever-Modell auswählen",
+      "parameters": "Parameter",
+      "simpleGroup": "Einfache Retriever",
+      "keywordGroup": "Stichwort-Retriever",
+      "denseGroup": "Dense-Retriever",
+      "embeddingGroup": "Embedding-Retriever",
+      "hybridGroup": "Hybrid",
+      "hybridDescription": "BM25 + HuggingFace multilinguale Embeddings",
+      "compositeGroup": "Zusammengesetzte Retriever",
+      "nestedCompositeNote": "Dies ist ein zusammengesetzter Retriever. Seine Kindknoten können im unteren Baum konfiguriert werden.",
+      "allChunks": "Alle Chunks",
+      "selectedChunks": "Ausgewählte Chunks",
+      "retrieverDescription": "Wählen Sie einen Retriever, um seine Parameter zu konfigurieren. Stichwort-Retriever (BM25, TF-IDF) finden Dokumente nach Termhäufigkeit. Embedding-Retriever erfassen semantische Bedeutung mit dichten Vektoren. Zusammengesetzte Retriever verketten mehrere Retriever, sequenziell oder parallel.",
+      "compositeInstructions": "Klicken Sie auf die +-Schaltfläche, um diesem Composite einen Kind-Retriever hinzuzufügen. Klicken Sie auf eine Retriever-Karte, um dessen Parameter zu bearbeiten oder ein anderes Modell zu wählen."
+    },
+    "generator": {
+      "description": "Wählen Sie das Sprachmodell, das Antworten basierend auf dem abgerufenen Kontext und Ihren Prompts generiert.",
+      "modelLabel": "Generator-Modell",
+      "advancedApplied": "Erweiterte Konfiguration angewendet",
+      "selectModel": "Sprachmodell auswählen",
+      "selectModelPlaceholder": "z. B. GPT-4, Claude, Llama",
+      "modelInfo": "Modell:",
+      "advancedButton": "Erweiterte Konfiguration öffnen"
+    },
+    "prompt": {
+      "description": "Wählen Sie eine Prompt-Vorlage, die definiert, wie der abgerufene Kontext und die Chat-Nachrichten kombiniert werden, um Antworten zu generieren.",
+      "availablePrompts": "Verfügbare Prompts",
+      "openPrompts": "Prompt-Bibliothek öffnen",
+      "promptLabel": "Prompt",
+      "selectTemplate": "Prompt-Vorlage auswählen",
+      "selectTemplatePlaceholder": "z. B. Standard-Prompt, Benutzerdefinierte Anweisung",
+      "selectedTemplate": "Ausgewählte Prompt-Vorlage:",
+      "newPromptButton": "Neuer Prompt",
+      "defaultGenerationPrompt": "Standard-Generierungsprompt",
+      "defaultQAGenerationPrompt": "Standard-Q&A-Generierungsprompt",
+      "createNewPrompt": "Neuen Prompt erstellen",
+      "language": "Sprache",
+      "collapse": "Einklappen",
+      "expand": "Ausklappen",
+      "descriptionToggle": "Beschreibung",
+      "languages": {
+        "en": "English",
+        "es": "Español",
+        "pt": "Português"
+      }
+    },
+    "promptView": {
+      "type": "Typ",
+      "language": "Sprache",
+      "languageNotAvailable": "Sprache nicht verfügbar",
+      "languageNone": "Keine Sprache",
+      "noContent": "Kein Vorlageninhalt",
+      "close": "Schließen",
+      "templateContent": "Vorlageninhalt",
+      "untitledPrompt": "Prompt",
+      "table": {
+        "id": "ID",
+        "name": "Name",
+        "type": "Typ",
+        "language": "Sprache",
+        "created": "Erstellt",
+        "edited": "Bearbeitet",
+        "actions": "Aktionen",
+        "viewPrompt": "Prompt anzeigen",
+        "currentPrompts": "Aktuelle Prompts",
+        "choosePrompt": "Wählen oder passen Sie Ihren Prompt an, um das Modellverhalten zu definieren.",
+        "newPrompt": "Neuer Prompt"
+      }
+    },
+    "documents": {
+      "uploadButton": "Dokumente hochladen",
+      "emptyUploadText": "Dokument(e) hochladen",
+      "table": {
+        "id": "ID",
+        "name": "Name",
+        "type": "Typ",
+        "created": "Erstellt",
+        "lastModified": "Zuletzt geändert",
+        "actions": "Aktionen",
+        "preview": "Vorschau",
+        "delete": "Löschen",
+        "addDocument": "Dokument hinzufügen",
+        "currentDocuments": "Aktuelle Dokumente",
+        "noDocumentsAvailable": "Keine Dokumente verfügbar.",
+        "emptyUploadText": "Dokument(e) hochladen",
+        "unknownType": "Unbekannt"
+      }
+    },
+    "summary": {
+      "title": "RAG-Sitzung",
+      "descriptionPlaceholder": "Keine Beschreibung verfügbar",
+      "createdLabel": "Erstellt",
+      "openChatButton": "RAG-Sitzungschat öffnen",
+      "detailedConfig": "Detaillierte Konfiguration",
+      "chunkingModel": "Chunking-Modell",
+      "retrieverModel": "Retriever-Modell",
+      "generationModel": "Generierungsmodell",
+      "promptSelected": "Prompt ausgewählt",
+      "promptName": "Name:",
+      "promptTemplate": "Vorlage:",
+      "noPrompt": "Kein Prompt konfiguriert",
+      "sessionNotFound": "Sitzung nicht gefunden",
+      "loading": "Sitzungsdaten werden geladen...",
+      "apiKeyCopied": "API-Schlüssel in die Zwischenablage kopiert",
+      "failedToCopyApiKey": "API-Schlüssel konnte nicht kopiert werden",
+      "noParameters": "Keine Parameter",
+      "viewDetails": "Details anzeigen",
+      "parameterDetails": "Parameterdetails",
+      "completeConfig": "Vollständige Parameterkonfiguration:",
+      "close": "Schließen",
+      "sessionName": "Sitzungsname",
+      "sessionDescription": "Sitzungsbeschreibung",
+      "parameterDetailTitle": "{{component}} - {{param}}",
+      "edit": "Bearbeiten",
+      "save": "Speichern",
+      "cancel": "Abbrechen",
+      "sessionUpdated": "Sitzung erfolgreich aktualisiert",
+      "failedToUpdateSession": "Sitzung konnte nicht aktualisiert werden"
+    },
+    "newPrompt": {
+      "title": "Neuen Prompt erstellen",
+      "description": "Die Prompt-Vorlage definiert, wie die Chunks (Dokumentteile) und Chat-Nachrichten integriert werden, um Antworten zu generieren. Passen Sie den Prompt an, um das Verhalten Ihrer RAG-Sitzungen anzupassen.",
+      "placeholdersInfo": "Verwenden Sie {chunks} für die Stelle, an der die abgerufenen Dokument-Chunks eingefügt werden, und {input} für die Benutzernachricht.",
+      "nameLabel": "Prompt-Name",
+      "nameRequired": "Prompt-Name ist erforderlich",
+      "promptLabel": "Prompt",
+      "promptPlaceholder": "Hier können Sie den Prompt ändern, zum Beispiel:\nJede Benutzernachricht wird als {input} hinzugefügt\nDie Quellen werden als {chunks} hinzugefügt",
+      "cancel": "Abbrechen",
+      "save": "Speichern",
+      "unsavedChanges": "Sie haben nicht gespeicherte Änderungen. Möchten Sie wirklich abbrechen?",
+      "success": "Prompt erfolgreich erstellt!",
+      "error": "Prompt konnte nicht erstellt werden",
+      "languageLabel": "Sprache (optional)"
+    },
+    "advanced": {
+      "chunkingTitle": "Erweiterte Chunking-Konfiguration",
+      "retrieverTitle": "Erweiterte Retriever-Konfiguration",
+      "generatorTitle": "Erweiterte Generator-Konfiguration",
+      "promptTitle": "Erweiterte Prompt-Konfiguration",
+      "close": "Schließen",
+      "done": "Fertig",
+      "noDescription": "Keine Beschreibung verfügbar",
+      "promptTemplate": "Prompt-Vorlage",
+      "promptInfo": "Die erweiterte Konfiguration für Prompts wird normalerweise über die Prompt-Vorlagen-Oberfläche verwaltet. Dort können Sie benutzerdefinierte Prompts erstellen oder bestehende ändern."
+    },
+    "rightPanel": {
+      "title": "RAG-Grundlagen",
+      "subtitle": "Retrieval-Augmented Generation (RAG) hilft dem Modell, mit Ihren Dokumenten zu antworten.",
+      "howItWorksTitle": "So funktioniert es",
+      "howItWorks": {
+        "step1": "Wählen Sie die Dokumente aus, die der Assistent verwenden soll.",
+        "step2": "Dokumente werden in kleinere Chunks aufgeteilt (Chunking), um die Abfrage effizient zu machen.",
+        "step3": "Ein Retriever findet die relevantesten Chunks für Ihre Frage.",
+        "step4": "Das LLM generiert eine Antwort mit diesen Chunks und der ausgewählten Prompt-Vorlage."
+      },
+      "settingsTitle": "Was jede Einstellung bewirkt",
+      "settings": {
+        "documents": {
+          "title": "Dokumente",
+          "point1": "Ihre Dokumente sind die Wissensbasis: Der Assistent kann nur das verwenden, was er daraus abrufen kann.",
+          "point2": "Laden Sie klare, gut strukturierte Dateien (PDF, TXT, etc.) für beste Ergebnisse hoch.",
+          "point3": "Derzeit werden nur textbasierte Dokumente unterstützt. Bilder, Audio und Video werden noch nicht unterstützt."
+        },
+        "chunking": {
+          "title": "Chunking",
+          "point1": "Chunking teilt Dokumente in kleinere durchsuchbare Teile auf.",
+          "point2": "Zu kleine Chunks: Die Abfrage kann Kontext verpassen und Antworten können fragmentiert wirken.",
+          "point3": "Zu große Chunks: Die Abfrage kann zusätzliches Rauschen bringen und Antworten können langsamer sein (mehr Tokens).",
+          "point4": "Überlappung hilft, Kontext zwischen Chunks zu erhalten, erhöht aber Wiederholungen."
+        },
+        "retriever": {
+          "title": "Retriever (Sparse vs Dense)",
+          "point1": "Der Retriever entscheidet, welche Chunks an das LLM gesendet werden.",
+          "point2": "Sparse (stichwortbasiert): ideal für exakte Begriffe wie IDs, Produktnamen oder Code; normalerweise schnell.",
+          "point3": "Sparse-Einschränkung: hat Schwierigkeiten mit Synonymen und stark paraphrasierten Fragen.",
+          "point4": "Dense (semantisch): besser für natürliche Sprache und Paraphrasen; sucht nach Bedeutung.",
+          "point5": "Dense-Einschränkung: kann 'ähnliche', aber nicht exakte Übereinstimmungen liefern und kann langsamer sein."
+        },
+        "llm": {
+          "title": "Sprachmodell (LLM)",
+          "point1": "Das LLM schreibt die endgültige Antwort mit den abgerufenen Chunks.",
+          "point2": "Größere Modelle antworten normalerweise besser, können aber langsamer und teurer sein.",
+          "point3": "Wenn Sie Geschwindigkeit benötigen, beginnen Sie mit einem kleineren Modell und iterieren Sie."
+        },
+        "prompt": {
+          "title": "Prompt-Vorlage",
+          "point1": "Prompts steuern den Ton, das Format und worauf sich das Modell konzentrieren soll.",
+          "point2": "Wenn Antworten zu lang/kurz oder unstrukturiert sind, versuchen Sie eine andere Vorlage oder verfeinern Sie den Prompt."
+        }
+      },
+      "tipsTitle": "Praktische Tipps",
+      "tips": {
+        "tip1": "Wenn Antworten Kontext vermissen, versuchen Sie größere Chunks oder mehr Überlappung.",
+        "tip2": "Wenn Antworten zu allgemein wirken, versuchen Sie kleinere Chunks oder passen Sie das Retriever-Modell an.",
+        "tip3": "Wählen Sie eine Prompt-Vorlage, die zu Ihrer Domäne und Ihrem Ton passt."
+      }
+    },
+    "common": {
+      "loading": "Wird geladen..."
+    },
+    "chunkingConfig": {
+      "configureTitle": "Chunking-Modell konfigurieren",
+      "modelLabel": "Chunking-Modell"
+    },
+    "generatorConfig": {
+      "configureTitle": "Generator-Modell konfigurieren (LLM)",
+      "modelLabel": "Generator-Modell"
+    },
+    "retrieverConfig": {
+      "modelLabel": "Retriever-Modell"
+    },
+    "breadcrumbs": {
+      "generative": "Generativ",
+      "rag": "RAG",
+      "documents": "Dokumente",
+      "prompts": "Prompts",
+      "goBack": "Zurück",
+      "sessionSuffix": "Sitzung"
+    },
+    "placeholders": {
+      "title": "Erforderliche Platzhalter",
+      "insert": "{{placeholder}} einfügen"
+    },
+    "documentPreview": {
+      "title": "Dokumentvorschau",
+      "noPreview": "Keine Vorschau für diesen Dateityp verfügbar.",
+      "pdfPreview": "PDF-Vorschau"
+    },
+    "paramsPanel": {
+      "title": "RAG Parameter",
+      "save": "Speichern",
+      "failedToLoad": "RAG-Sitzung konnte nicht geladen werden",
+      "updated": "RAG-Parameter aktualisiert",
+      "failedToUpdate": "RAG-Parameter konnten nicht aktualisiert werden"
+    }
+  },
+  "validation": {
+    "contextSpace": "Mit Ihrer aktuellen Konfiguration haben Sie Platz für Ihre Quellen (Chunks) und {{availableChars}} Zeichen für Ihre Nachricht.",
+    "insufficientContextTitle": "Unzureichendes Kontextfenster",
+    "insufficientContextDescription": "Das ausgewählte Modell hat nicht genug Kontext, um die aktuellen Chunk-Größen- und Top-K-Einstellungen zu verarbeiten. Bitte reduzieren Sie die Chunk-Größe, die Anzahl der Ergebnisse (Top K) oder wählen Sie ein Modell mit einem größeren Kontextfenster.",
+    "apiKeyMissingTitle": "API-Schlüssel erforderlich",
+    "apiKeyMissingDescription": "Dieses Modell benötigt einen API-Schlüssel, um zu funktionieren. Öffnen Sie die erweiterte Konfiguration, um Ihren API-Schlüssel hinzuzufügen."
+  },
+  "documentReferences": {
+    "modalTitle": "Dokumentreferenzen",
+    "fallbackTitle": "Dokument {{id}}",
+    "chunksCount_one": "Dieses Dokument lieferte {{count}} Chunk für diese Antwort:",
+    "chunksCount_other": "Dieses Dokument lieferte {{count}} Chunks für diese Antwort:",
+    "chunkLabel": "Chunk {{position}}",
+    "close": "Schließen"
+  },
+  "sourcesDisplay": {
+    "viewSources": "Nachrichtenquellen anzeigen",
+    "provided": "bereitgestellt",
+    "chunk_one": "Chunk",
+    "chunk_other": "Chunks",
+    "viewChunks": "Chunks anzeigen"
+  },
+  "documentsBar": {
+    "title": "Dokumente",
+    "documentCount_one": "{{count}} Dokument",
+    "documentCount_other": "{{count}} Dokumente",
+    "inCurrentSession": " in aktueller Sitzung",
+    "available": " verfügbar",
+    "addDocuments": "Dokumente hinzufügen",
+    "searchPlaceholder": "Dokumente suchen",
+    "detailedView": "Detailansicht anzeigen",
+    "noDocumentsFound": "Keine Dokumente gefunden",
+    "noDocumentsInSession": "Keine Dokumente in dieser Sitzung",
+    "noDocumentsAvailable": "Keine Dokumente verfügbar",
+    "uploadDocuments": "Dokument(e) hochladen",
+    "failedFetch": "Dokumente konnten nicht abgerufen werden",
+    "successUpload_one": "{{count}} Dokument erfolgreich hochgeladen",
+    "successUpload_other": "{{count}} Dokumente erfolgreich hochgeladen",
+    "failedUpload": "Dokument(e) konnten nicht hochgeladen werden"
+  },
+  "ragPromptsPage": {
+    "title": "RAG-Prompts",
+    "description": "Verwalten Sie Prompts für Ihre RAG-Sitzungen: Zeigen Sie alle verfügbaren Prompts an und erstellen Sie neue, um Ihre KI-Interaktionen zu verbessern."
+  },
+  "ragDocumentsPage": {
+    "title": "RAG-Dokumente",
+    "description": "Verwalten Sie Dokumente für Ihre RAG-Sitzungen: Zeigen Sie vorhandene Dokumente an und laden Sie neue hoch, um Ihre KI-Wissensdatenbank zu erweitern."
   }
 }
diff --git a/DashAI/front/src/utils/i18n/locales/en/generative.json b/DashAI/front/src/utils/i18n/locales/en/generative.json
index 654597724..76a0cf911 100644
--- a/DashAI/front/src/utils/i18n/locales/en/generative.json
+++ b/DashAI/front/src/utils/i18n/locales/en/generative.json
@@ -26,7 +26,7 @@
     "showLess": "Show less",
     "readMore": "Read more",
     "generativeModule": "Generative Module",
-    "createNewSession": "Create a New Session",
+    "createNewSession": "Create a new session",
     "createNewSessionDescription": "Pick a model, configure its parameters, and start chatting.",
     "configureSession": "Configure session",
     "modelHasNoParameters": "This model has no configurable parameters.",
@@ -39,7 +39,7 @@
     "noSessionsFound": "No sessions found",
     "parameterChangeEvent": "Parameters updated: <1></1>",
     "parameterChangeHistory": "Parameter change history for the current session",
-    "searchSessions": "Search Sessions",
+    "searchSessions": "Search sessions",
     "selectGenerativeTask": "Select a generative task to start a new session",
     "selectModel": "Select a model",
     "selectModelAndConfigureParameters": "Select a model and configure its parameters.",
@@ -57,11 +57,377 @@
     "attachMedia": "Attach media",
     "attachMediaToContinue": "Attach media to continue",
     "noInputAvailable": "No input available for this task",
-    "selectModelCrumb": "Select Model"
+    "selectModelCrumb": "Select Model",
+    "ragTask": "RAG - Retrieval Augmented Generation",
+    "ragTaskDescription": "Upload documents and chat with them using retrieval-augmented generation."
   },
   "message": {
     "sessionCreatedSuccess": "Session successfully created.",
     "sessionDeleted": "Session deleted successfully",
     "sessionUpdated": "Session updated successfully"
+  },
+  "rag": {
+    "setup": {
+      "title": "Create RAG session",
+      "subtitle": "Provide basic information, select documents, and configure the components for your RAG session.",
+      "sessionDetails": "Session Details",
+      "sessionName": "Session Name *",
+      "sessionNamePlaceholder": "e.g., Product Documentation RAG",
+      "sessionNameDuplicate": "This session name already exists",
+      "sessionNameDuplicateAlert": "A session with this name already exists. Please use a different name.",
+      "description": "Description (Optional)",
+      "descriptionPlaceholder": "Describe the purpose of this RAG session...",
+      "selectDocuments": "Select documents",
+      "selectDocumentsDescription": "Upload new documents or select from existing ones to be used for RAG.",
+      "chunkingStrategy": "Chunking strategy",
+      "retrieverModel": "Retriever Model",
+      "promptTemplate": "Prompt Template",
+      "languageModel": "Language Model (LLM)",
+      "cancel": "Cancel",
+      "saveSession": "Save session",
+      "openDocumentsLibrary": "Open documents library"
+    },
+    "validation": {
+      "nameRequired": "Session name is required",
+      "nameUnique": "Session name must be unique",
+      "documentsRequired": "At least one document must be selected",
+      "chunkingRequired": "Chunking model must be configured",
+      "retrieverRequired": "Retriever model must be configured",
+      "generatorRequired": "Generator model must be configured",
+      "generatorInvalid": "Generator configuration is invalid (check context window)",
+      "promptRequired": "Prompt template must be selected"
+    },
+    "messages": {
+      "success": "RAG Session created successfully"
+    },
+    "chunking": {
+      "description": "Select how your documents will be split into chunks for retrieval.",
+      "presets": {
+        "small": {
+          "label": "Small chunks"
+        },
+        "paragraph": {
+          "label": "Paragraph length"
+        },
+        "page": {
+          "label": "Page chunk"
+        },
+        "large": {
+          "label": "Large sections"
+        },
+        "chunkSizeFormat": "{{chars}} characters ≈ {{tokens}} tokens",
+        "custom": {
+          "label": "Custom chunking model",
+          "description": "Advanced configuration applied"
+        }
+      },
+      "advancedButton": "Open advanced configuration"
+    },
+    "retriever": {
+      "description": "Select your retrieval paradigm. The retriever will search through your documents and return the most relevant chunks.",
+      "paradigmLabel": "Retrieval paradigm",
+      "keywordLabel": "Keyword",
+      "semanticLabel": "Semantic",
+      "hybridLabel": "Hybrid",
+      "hybridDescription": "BM25 + Harrier 0.6B",
+      "advancedApplied": "Advanced configuration applied",
+      "topKLabel": "Number of chunks (Top K)",
+      "topKCardDescription": "Chunks",
+      "advancedTopK": "Advanced Top K: {{topK}}",
+      "advancedButton": "Open advanced configuration",
+      "modifyFromAdvanced": "Top K must be modified from the advanced configuration view",
+      "explanations": {
+        "SparseRetriever": "Best for specific keyword and ID search. Useful for finding exact matches in documents.",
+        "DenseRetriever": "Contextual search that understands meaning. Great for finding semantically similar content but may miss specific keywords or IDs.",
+        "HybridRetriever": "Combines keyword and contextual search for balanced performance across different query types."
+      }
+    },
+    "composite": {
+      "addChild": "Add retriever node",
+      "removeNode": "Remove node",
+      "configureNode": "Configure node",
+      "selectModel": "Select retriever model",
+      "parameters": "Parameters",
+      "simpleGroup": "Simple retrievers",
+      "keywordGroup": "Keyword retrievers",
+      "denseGroup": "Dense retrievers",
+      "embeddingGroup": "Embedding retrievers",
+      "hybridGroup": "Hybrid",
+      "hybridDescription": "BM25 + HuggingFace multilingual embeddings",
+      "compositeGroup": "Composite retrievers",
+      "nestedCompositeNote": "This is a composite retriever. Its children can be configured in the tree below.",
+      "allChunks": "All chunks",
+      "selectedChunks": "Selected chunks",
+      "retrieverDescription": "Choose a retriever to configure its parameters. Keyword retrievers (BM25, TF-IDF) find documents by term frequency. Embedding retrievers capture semantic meaning with dense vectors. Composite retrievers chain multiple retrievers together, sequentially or in parallel.",
+      "compositeInstructions": "Click the + button to add a child retriever to this composite. Click on a retriever card to edit its parameters or choose another model."
+    },
+    "generator": {
+      "description": "Select the language model that will generate responses based on the retrieved context and your prompts.",
+      "modelLabel": "Generator model",
+      "advancedApplied": "Advanced configuration applied",
+      "selectModel": "Select language model",
+      "selectModelPlaceholder": "e.g., GPT-4, Claude, Llama",
+      "modelInfo": "Model:",
+      "advancedButton": "Open advanced configuration"
+    },
+    "prompt": {
+      "description": "Select a prompt template that defines how the retrieved context and chat messages are combined to generate responses.",
+      "availablePrompts": "Available prompts",
+      "openPrompts": "Open prompt library",
+      "promptLabel": "Prompt",
+      "selectTemplate": "Select prompt template",
+      "selectTemplatePlaceholder": "e.g., Default Prompt, Custom Instruction",
+      "selectedTemplate": "Selected prompt template:",
+      "newPromptButton": "New prompt",
+      "defaultGenerationPrompt": "Default Generation Prompt",
+      "defaultQAGenerationPrompt": "Default Q&A Generation Prompt",
+      "createNewPrompt": "Create new prompt",
+      "language": "Language",
+      "collapse": "Collapse",
+      "expand": "Expand",
+      "descriptionToggle": "Description",
+      "languages": {
+        "en": "English",
+        "es": "Español",
+        "pt": "Português"
+      }
+    },
+    "promptView": {
+      "type": "Type",
+      "language": "Language",
+      "languageNotAvailable": "Language not available",
+      "languageNone": "No language",
+      "noContent": "No template content",
+      "close": "Close",
+      "templateContent": "Template content",
+      "untitledPrompt": "Prompt",
+      "table": {
+        "id": "ID",
+        "name": "Name",
+        "type": "Type",
+        "language": "Language",
+        "created": "Created",
+        "edited": "Edited",
+        "actions": "Actions",
+        "viewPrompt": "View prompt",
+        "currentPrompts": "Current prompts",
+        "choosePrompt": "Choose or customize your prompt to define the model's behavior.",
+        "newPrompt": "New Prompt"
+      }
+    },
+    "documents": {
+      "uploadButton": "Upload documents",
+      "emptyUploadText": "Upload your document(s)",
+      "table": {
+        "id": "ID",
+        "name": "Name",
+        "type": "Type",
+        "created": "Created",
+        "lastModified": "Last Modified",
+        "actions": "Actions",
+        "preview": "Preview",
+        "delete": "Delete",
+        "addDocument": "Add new document",
+        "currentDocuments": "Current documents",
+        "noDocumentsAvailable": "No documents available.",
+        "emptyUploadText": "Upload your document(s)",
+        "unknownType": "Unknown"
+      }
+    },
+    "summary": {
+      "title": "RAG session",
+      "descriptionPlaceholder": "No description available",
+      "createdLabel": "Created",
+      "openChatButton": "Open RAG session chat",
+      "detailedConfig": "Detailed configuration",
+      "chunkingModel": "Chunking model",
+      "retrieverModel": "Retriever Model",
+      "generationModel": "Generation model",
+      "promptSelected": "Prompt selected",
+      "promptName": "Name:",
+      "promptTemplate": "Template:",
+      "noPrompt": "No prompt configured",
+      "sessionNotFound": "Session not found",
+      "loading": "Loading session data...",
+      "apiKeyCopied": "API key copied to clipboard",
+      "failedToCopyApiKey": "Failed to copy API key",
+      "noParameters": "No parameters",
+      "viewDetails": "View details",
+      "parameterDetails": "Parameter details",
+      "completeConfig": "Complete parameter configuration:",
+      "close": "Close",
+      "sessionName": "Session Name",
+      "sessionDescription": "Session Description",
+      "parameterDetailTitle": "{{component}} - {{param}}",
+      "edit": "Edit",
+      "save": "Save",
+      "cancel": "Cancel",
+      "sessionUpdated": "Session updated successfully",
+      "failedToUpdateSession": "Failed to update session"
+    },
+    "newPrompt": {
+      "title": "Create a new prompt",
+      "description": "Prompt template defines how the chunks (pieces of documents) and chat messages are integrated to generate responses. Customize the prompt to tailor the behavior of your RAG sessions.",
+      "placeholdersInfo": "Use {chunks} to represent where the retrieved document chunks will be inserted, and {input} for the user message.",
+      "nameLabel": "Prompt Name",
+      "nameRequired": "Prompt name is required",
+      "promptLabel": "Prompt",
+      "promptPlaceholder": "Here you can modify the prompt, for example:\nEach user message is added as {input}\nThe sources are added as {chunks}",
+      "cancel": "Cancel",
+      "save": "Save",
+      "unsavedChanges": "You have unsaved changes. Are you sure you want to cancel?",
+      "success": "Prompt created successfully!",
+      "error": "Failed to create prompt",
+      "languageLabel": "Language (optional)"
+    },
+    "advanced": {
+      "chunkingTitle": "Advanced chunking configuration",
+      "retrieverTitle": "Advanced retriever configuration",
+      "generatorTitle": "Advanced generator configuration",
+      "promptTitle": "Advanced prompt configuration",
+      "close": "Close",
+      "done": "Done",
+      "noDescription": "No description available",
+      "promptTemplate": "Prompt Template",
+      "promptInfo": "Advanced configuration for prompts is typically managed through the prompt templates interface. You can create custom prompts or modify existing ones there."
+    },
+    "rightPanel": {
+      "title": "RAG basics",
+      "subtitle": "Retrieval-Augmented Generation (RAG) helps the model answer using your documents.",
+      "howItWorksTitle": "How it works",
+      "howItWorks": {
+        "step1": "Select the documents you want the assistant to use.",
+        "step2": "Documents are split into smaller chunks (chunking) to make retrieval efficient.",
+        "step3": "A retriever finds the most relevant chunks for your question.",
+        "step4": "The LLM generates an answer using those chunks and the selected prompt template."
+      },
+      "settingsTitle": "What each setting does",
+      "settings": {
+        "documents": {
+          "title": "Documents",
+          "point1": "Your documents are the knowledge base: the assistant can only use what it can retrieve from them.",
+          "point2": "Upload clear, well-structured files (PDF, TXT, etc.) for best results.",
+          "point3": "Currently, only text-based documents are supported. Images, audio, and video are not yet supported."
+        },
+        "chunking": {
+          "title": "Chunking",
+          "point1": "Chunking splits documents into smaller pieces that can be searched.",
+          "point2": "Too small chunks: retrieval can miss context and answers may feel fragmented.",
+          "point3": "Too large chunks: retrieval can bring extra noise and answers may be slower (more tokens).",
+          "point4": "Overlap helps keep context between chunks, but increases repetition."
+        },
+        "retriever": {
+          "title": "Retriever (Sparse vs Dense)",
+          "point1": "The retriever decides which chunks are sent to the LLM.",
+          "point2": "Sparse (keyword-based): great for exact terms like IDs, product names, or code; usually fast.",
+          "point3": "Sparse limitation: struggles with synonyms and heavily paraphrased questions.",
+          "point4": "Dense (semantic): better for natural language and paraphrases; searches by meaning.",
+          "point5": "Dense limitation: can return ‘similar’ but not exact matches, and may be slower."
+        },
+        "llm": {
+          "title": "Language Model (LLM)",
+          "point1": "The LLM writes the final answer using the retrieved chunks.",
+          "point2": "Larger models usually answer better, but can be slower and costlier.",
+          "point3": "If you need speed, start with a smaller model and iterate."
+        },
+        "prompt": {
+          "title": "Prompt template",
+          "point1": "Prompts control the tone, format, and what the model should focus on.",
+          "point2": "If answers are too long/short or lack structure, try another template or refine the prompt."
+        }
+      },
+      "tipsTitle": "Practical tips",
+      "tips": {
+        "tip1": "If answers miss context, try larger chunks or more overlap.",
+        "tip2": "If answers feel too generic, try smaller chunks or adjust the retriever model.",
+        "tip3": "Choose a prompt template that matches your domain and tone."
+      }
+    },
+    "common": {
+      "loading": "Loading..."
+    },
+    "chunkingConfig": {
+      "configureTitle": "Configure Chunking Model",
+      "modelLabel": "Chunking Model"
+    },
+    "generatorConfig": {
+      "configureTitle": "Configure Generator Model (LLM)",
+      "modelLabel": "Generator Model"
+    },
+    "retrieverConfig": {
+      "modelLabel": "Retriever model"
+    },
+    "breadcrumbs": {
+      "generative": "Generative",
+      "rag": "RAG",
+      "documents": "Documents",
+      "prompts": "Prompts",
+      "goBack": "Go back",
+      "sessionSuffix": "session"
+    },
+    "placeholders": {
+      "title": "Required Placeholders",
+      "insert": "Insert {{placeholder}}"
+    },
+    "documentPreview": {
+      "title": "Preview Document",
+      "noPreview": "No preview available for this file type.",
+      "pdfPreview": "PDF Preview"
+    },
+    "paramsPanel": {
+      "title": "RAG Parameters",
+      "save": "Save",
+      "failedToLoad": "Failed to load RAG session",
+      "updated": "RAG parameters updated",
+      "failedToUpdate": "Failed to update RAG parameters"
+    }
+  },
+  "validation": {
+    "contextSpace": "With your current configuration you have space for your sources (chunks) and {{availableChars}} characters for your message.",
+    "insufficientContextTitle": "Insufficient context window",
+    "insufficientContextDescription": "The selected model doesn't have enough context to handle the current Chunk Size and Top K settings. Please reduce the Chunk Size, the number of results (Top K), or select a model with a larger context window.",
+    "apiKeyMissingTitle": "API key required",
+    "apiKeyMissingDescription": "This model requires an API key to function. Open Advanced Configuration to add your API key."
+  },
+  "documentReferences": {
+    "modalTitle": "Document references",
+    "fallbackTitle": "Document {{id}}",
+    "chunksCount_one": "This document provided {{count}} chunk for this response:",
+    "chunksCount_other": "This document provided {{count}} chunks for this response:",
+    "chunkLabel": "Chunk {{position}}",
+    "close": "Close"
+  },
+  "sourcesDisplay": {
+    "viewSources": "View message sources",
+    "provided": "provided",
+    "chunk_one": "chunk",
+    "chunk_other": "chunks",
+    "viewChunks": "View chunks"
+  },
+  "documentsBar": {
+    "title": "Documents",
+    "documentCount_one": "{{count}} document",
+    "documentCount_other": "{{count}} documents",
+    "inCurrentSession": " in current session",
+    "available": " available",
+    "addDocuments": "Add documents",
+    "searchPlaceholder": "Search documents",
+    "detailedView": "See detailed view",
+    "noDocumentsFound": "No documents found",
+    "noDocumentsInSession": "No documents in this session",
+    "noDocumentsAvailable": "No documents available",
+    "uploadDocuments": "Upload your document(s)",
+    "failedFetch": "Failed to fetch documents",
+    "successUpload_one": "Successfully uploaded {{count}} document",
+    "successUpload_other": "Successfully uploaded {{count}} documents",
+    "failedUpload": "Failed to upload document(s)"
+  },
+  "ragPromptsPage": {
+    "title": "RAG prompts",
+    "description": "Manage prompts for your RAG sessions: view all available prompts and create new ones to improve your AI interactions."
+  },
+  "ragDocumentsPage": {
+    "title": "RAG documents",
+    "description": "Manage documents for your RAG sessions: view existing documents and upload new ones to enhance your AI knowledge base."
   }
 }
diff --git a/DashAI/front/src/utils/i18n/locales/es/generative.json b/DashAI/front/src/utils/i18n/locales/es/generative.json
index 97f059735..588551864 100644
--- a/DashAI/front/src/utils/i18n/locales/es/generative.json
+++ b/DashAI/front/src/utils/i18n/locales/es/generative.json
@@ -1,6 +1,6 @@
 {
   "button": {
-    "backToTaskSelection": "Volver a Selección de Tarea",
+    "backToTaskSelection": "Volver a selección de tarea",
     "createSession": "Crear una sesión",
     "generativeHub": "Centro Generativo"
   },
@@ -21,47 +21,413 @@
   },
   "label": {
     "confirmDeleteSession": "¿Seguro que quieres eliminar la sesión \"{{name}}\"? Esta acción no se puede deshacer.",
-    "deleteSessionConfirmation": "¿Está seguro de que desea eliminar esta sesión? Esta acción no se puede deshacer.",
+    "deleteSessionConfirmation": "¿Seguro que quieres eliminar esta sesión? Esta acción no se puede deshacer.",
     "noDescriptionAvailable": "No hay descripción disponible",
     "showLess": "Mostrar menos",
     "readMore": "Leer más",
     "generativeModule": "Módulo Generativo",
-    "createNewSession": "Crear una Nueva Sesión",
+    "createNewSession": "Crear una nueva sesión",
     "createNewSessionDescription": "Elige un modelo, configura sus parámetros y empieza a conversar.",
     "configureSession": "Configurar sesión",
     "modelHasNoParameters": "Este modelo no tiene parámetros configurables.",
     "nameAndDescribeYourSession": "Nombra tu sesión. Ajusta los parámetros a la derecha. Puedes cambiarlos en cualquier momento durante la sesión.",
-    "nameYourSession": "Nombre su sesión",
-    "newSession": "Nueva Sesión",
+    "nameYourSession": "Nombra tu sesión",
+    "newSession": "Nueva sesión",
     "pickAModelGroupedByTask": "Elige un modelo. Los modelos están agrupados por tarea.",
     "searchModels": "Buscar modelos...",
     "startBySelectingATask": "Comienza seleccionando una tarea generativa.",
     "noSessionsFound": "No se encontraron sesiones",
     "parameterChangeEvent": "Parámetros actualizados: <1></1>",
     "parameterChangeHistory": "Historial de cambios de parámetros para la sesión actual",
-    "searchSessions": "Buscar Sesiones",
+    "searchSessions": "Buscar sesiones",
     "selectGenerativeTask": "Seleccione una tarea generativa para comenzar una nueva sesión",
     "selectModel": "Seleccione un modelo",
     "selectModelAndConfigureParameters": "Seleccione un modelo y configure sus parámetros.",
     "selectSessionToViewParameters": "Seleccione una sesión para ver y editar sus parámetros.",
-    "sessionDescription": "Descripción de la Sesión",
-    "sessionInformation": "Información de la Sesión",
-    "sessionName": "Nombre de la Sesión",
+    "sessionDescription": "Descripción de la sesión",
+    "sessionInformation": "Información de la sesión",
+    "sessionName": "Nombre de la sesión",
     "sessions": "Sesiones",
     "tourDisabledMessage": "Regresa al inicio para comenzar el tour",
     "typeYourMessage": "Escribe un mensaje",
-    "untitledSession": "Sesión Sin Título",
+    "untitledSession": "Sesión sin título",
     "attachImage": "Adjuntar imagen",
     "attachAudio": "Adjuntar audio",
     "attachVideo": "Adjuntar video",
     "attachMedia": "Adjuntar medio",
     "attachMediaToContinue": "Adjunta medios para continuar",
     "noInputAvailable": "No hay entrada disponible para esta tarea",
-    "selectModelCrumb": "Select Model"
+    "selectModelCrumb": "Select Model",
+    "ragTask": "RAG - Generación aumentada por recuperación",
+    "ragTaskDescription": "Sube documentos y conversa con ellos usando generación aumentada por recuperación."
   },
   "message": {
     "sessionCreatedSuccess": "Sesión creada exitosamente.",
     "sessionDeleted": "Sesión eliminada exitosamente",
     "sessionUpdated": "Sesión actualizada exitosamente"
+  },
+  "rag": {
+    "setup": {
+      "title": "Crear sesión RAG",
+      "subtitle": "Proporciona información básica, selecciona documentos y configura los componentes para tu sesión RAG.",
+      "sessionDetails": "Detalles de la sesión",
+      "sessionName": "Nombre de la sesión *",
+      "sessionNamePlaceholder": "p. ej., Documentación de producto RAG",
+      "sessionNameDuplicate": "Este nombre de sesión ya existe",
+      "sessionNameDuplicateAlert": "Ya existe una sesión con este nombre. Por favor, usa un nombre diferente.",
+      "description": "Descripción (Opcional)",
+      "descriptionPlaceholder": "Describe el propósito de esta sesión RAG...",
+      "selectDocuments": "Seleccionar documentos",
+      "selectDocumentsDescription": "Sube nuevos documentos o selecciona de los existentes para utilizarlos en RAG.",
+      "chunkingStrategy": "Estrategia de chunking",
+      "retrieverModel": "Modelo de recuperación",
+      "promptTemplate": "Plantilla de prompt",
+      "languageModel": "Modelo de lenguaje (LLM)",
+      "cancel": "Cancelar",
+      "saveSession": "Guardar sesión",
+      "openDocumentsLibrary": "Abrir biblioteca de documentos"
+    },
+    "validation": {
+      "nameRequired": "El nombre de la sesión es obligatorio",
+      "nameUnique": "El nombre de la sesión debe ser único",
+      "documentsRequired": "Se debe seleccionar al menos un documento",
+      "chunkingRequired": "Se debe configurar el modelo de chunking",
+      "retrieverRequired": "Se debe configurar el modelo de recuperación",
+      "generatorRequired": "Se debe configurar el modelo de generación",
+      "generatorInvalid": "La configuración del generador es inválida (revisa la ventana de contexto)",
+      "promptRequired": "Se debe seleccionar una plantilla de prompt"
+    },
+    "messages": {
+      "success": "Sesión RAG creada exitosamente"
+    },
+    "chunking": {
+      "description": "Selecciona cómo se dividirán tus documentos en fragmentos para la recuperación.",
+      "presets": {
+        "small": {
+          "label": "Fragmentos pequeños"
+        },
+        "paragraph": {
+          "label": "Longitud de párrafo"
+        },
+        "page": {
+          "label": "Fragmento de página"
+        },
+        "large": {
+          "label": "Secciones grandes"
+        },
+        "chunkSizeFormat": "{{chars}} caracteres ≈ {{tokens}} tokens",
+        "custom": {
+          "label": "Modelo de chunking personalizado",
+          "description": "Configuración avanzada aplicada"
+        }
+      },
+      "advancedButton": "Abrir configuración avanzada"
+    },
+    "retriever": {
+      "description": "Selecciona tu paradigma de recuperación. El recuperador buscará en tus documentos y devolverá los fragmentos más relevantes.",
+      "paradigmLabel": "Paradigma de recuperación",
+      "keywordLabel": "Palabras clave",
+      "semanticLabel": "Semántico",
+      "hybridLabel": "Híbrido",
+      "hybridDescription": "BM25 + Harrier 0.6B",
+      "advancedApplied": "Configuración avanzada aplicada",
+      "topKLabel": "Número de fragmentos (Top K)",
+      "topKCardDescription": "Fragmentos",
+      "advancedTopK": "Top K avanzado: {{topK}}",
+      "advancedButton": "Abrir configuración avanzada",
+      "modifyFromAdvanced": "El Top K debe modificarse desde la vista de configuración avanzada",
+      "explanations": {
+        "SparseRetriever": "Ideal para búsquedas por palabras clave e IDs específicos. Útil para encontrar coincidencias exactas en los documentos.",
+        "DenseRetriever": "Búsqueda contextual que entiende el significado. Ideal para encontrar contenido semánticamente similar pero puede omitir palabras clave o IDs específicos.",
+        "HybridRetriever": "Combina la búsqueda por palabras clave y contextual para un rendimiento equilibrado en diferentes tipos de consultas."
+      }
+    },
+    "composite": {
+      "addChild": "Añadir nodo recuperador",
+      "removeNode": "Eliminar nodo",
+      "configureNode": "Configurar nodo",
+      "selectModel": "Seleccionar modelo recuperador",
+      "parameters": "Parámetros",
+      "simpleGroup": "Recuperadores simples",
+      "keywordGroup": "Recuperadores por palabras clave",
+      "denseGroup": "Recuperadores densos",
+      "embeddingGroup": "Recuperadores por Embeddings",
+      "hybridGroup": "Híbrido",
+      "hybridDescription": "BM25 + embeddings multilingües HuggingFace",
+      "compositeGroup": "Recuperadores compuestos",
+      "nestedCompositeNote": "Este es un recuperador compuesto. Sus hijos pueden configurarse en el árbol inferior.",
+      "allChunks": "Todos los fragmentos",
+      "selectedChunks": "Fragmentos seleccionados",
+      "retrieverDescription": "Elige un recuperador para configurar sus parámetros. Los recuperadores por palabras clave (BM25, TF-IDF) encuentran documentos por frecuencia de términos. Los recuperadores por embeddings capturan el significado semántico con vectores densos. Los recuperadores compuestos encadenan varios recuperadores, de forma secuencial o en paralelo.",
+      "compositeInstructions": "Haz clic en el botón + para añadir un recuperador hijo a este compuesto. Haz clic sobre una tarjeta para editar sus parámetros o elegir otro modelo."
+    },
+    "generator": {
+      "description": "Selecciona el modelo de lenguaje que generará las respuestas basadas en el contexto recuperado y tus prompts.",
+      "modelLabel": "Modelo de generación",
+      "advancedApplied": "Configuración avanzada aplicada",
+      "selectModel": "Seleccionar modelo de lenguaje",
+      "selectModelPlaceholder": "ej. GPT-4, Claude, Llama",
+      "modelInfo": "Modelo:",
+      "advancedButton": "Abrir configuración avanzada"
+    },
+    "prompt": {
+      "description": "Selecciona una plantilla de prompt que define cómo se combinan el contexto recuperado y los mensajes del chat para generar respuestas.",
+      "availablePrompts": "Plantillas disponibles",
+      "openPrompts": "Abrir biblioteca de prompts",
+      "promptLabel": "Prompt",
+      "selectTemplate": "Seleccionar plantilla de prompt",
+      "selectTemplatePlaceholder": "p. ej., Prompt por defecto, Instrucción personalizada",
+      "selectedTemplate": "Plantilla de prompt seleccionada:",
+      "newPromptButton": "Nuevo prompt",
+      "defaultGenerationPrompt": "Prompt de generación predeterminado",
+      "defaultQAGenerationPrompt": "Prompt predeterminado para preguntas y respuestas",
+      "createNewPrompt": "Crear nuevo prompt",
+      "language": "Idioma",
+      "collapse": "Contraer",
+      "expand": "Expandir",
+      "descriptionToggle": "Descripción",
+      "languages": {
+        "en": "English",
+        "es": "Español",
+        "pt": "Português"
+      }
+    },
+    "promptView": {
+      "type": "Tipo",
+      "language": "Idioma",
+      "languageNotAvailable": "Idioma no disponible",
+      "languageNone": "Sin idioma",
+      "noContent": "Sin contenido de plantilla",
+      "close": "Cerrar",
+      "templateContent": "Contenido de la plantilla",
+      "untitledPrompt": "Prompt",
+      "table": {
+        "id": "ID",
+        "name": "Nombre",
+        "type": "Tipo",
+        "language": "Idioma",
+        "created": "Creado",
+        "edited": "Editado",
+        "actions": "Acciones",
+        "viewPrompt": "Ver prompt",
+        "currentPrompts": "Prompts actuales",
+        "choosePrompt": "Elige o personaliza tu prompt para definir el comportamiento del modelo.",
+        "newPrompt": "Nuevo prompt"
+      }
+    },
+    "documents": {
+      "uploadButton": "Subir documentos",
+      "emptyUploadText": "Sube tu(s) documento(s)",
+      "table": {
+        "id": "ID",
+        "name": "Nombre",
+        "type": "Tipo",
+        "created": "Creado",
+        "lastModified": "Última modificación",
+        "actions": "Acciones",
+        "preview": "Vista previa",
+        "delete": "Eliminar",
+        "addDocument": "Añadir documento",
+        "currentDocuments": "Documentos actuales",
+        "noDocumentsAvailable": "No hay documentos disponibles.",
+        "emptyUploadText": "Sube tu(s) documento(s)",
+        "unknownType": "Desconocido"
+      }
+    },
+    "summary": {
+      "title": "Sesión RAG",
+      "descriptionPlaceholder": "No hay descripción disponible",
+      "createdLabel": "Creado",
+      "openChatButton": "Abrir chat de sesión RAG",
+      "detailedConfig": "Configuración detallada",
+      "chunkingModel": "Modelo de chunking",
+      "retrieverModel": "Modelo de recuperación",
+      "generationModel": "Modelo de generación",
+      "promptSelected": "Prompt seleccionado",
+      "promptName": "Nombre:",
+      "promptTemplate": "Plantilla:",
+      "noPrompt": "No hay prompt configurado",
+      "sessionNotFound": "Sesión no encontrada",
+      "loading": "Cargando datos de la sesión...",
+      "apiKeyCopied": "Clave API copiada al portapapeles",
+      "failedToCopyApiKey": "Error al copiar la clave API",
+      "noParameters": "Sin parámetros",
+      "viewDetails": "Ver detalles",
+      "parameterDetails": "Detalles del parámetro",
+      "completeConfig": "Configuración completa del parámetro:",
+      "close": "Cerrar",
+      "sessionName": "Nombre de la sesión",
+      "sessionDescription": "Descripción de la sesión",
+      "parameterDetailTitle": "{{component}} - {{param}}",
+      "edit": "Editar",
+      "save": "Guardar",
+      "cancel": "Cancelar",
+      "sessionUpdated": "Sesión actualizada correctamente",
+      "failedToUpdateSession": "Error al actualizar la sesión"
+    },
+    "newPrompt": {
+      "title": "Crear una nueva plantilla",
+      "description": "La plantilla de prompt define cómo se integran los fragmentos (piezas de documentos) y los mensajes del chat para generar respuestas. Personaliza el prompt para adaptar el comportamiento de tus sesiones RAG.",
+      "placeholdersInfo": "Usa {chunks} para representar dónde se insertarán los fragmentos del documento recuperado, e {input} para el mensaje del usuario.",
+      "nameLabel": "Nombre del prompt",
+      "nameRequired": "El nombre del prompt es obligatorio",
+      "promptLabel": "Prompt",
+      "promptPlaceholder": "Aquí puedes modificar el prompt, por ejemplo:\nCada mensaje del usuario se añade como {input}\nLas fuentes se añaden como {chunks}",
+      "cancel": "Cancelar",
+      "save": "Guardar",
+      "unsavedChanges": "Tienes cambios sin guardar. ¿Estás seguro de que quieres cancelar?",
+      "success": "¡Prompt creado exitosamente!",
+      "error": "Error al crear el prompt",
+      "languageLabel": "Idioma (opcional)"
+    },
+    "advanced": {
+      "chunkingTitle": "Configuración avanzada de chunking",
+      "retrieverTitle": "Configuración avanzada de recuperación",
+      "generatorTitle": "Configuración avanzada de generación",
+      "promptTitle": "Configuración avanzada de prompt",
+      "close": "Cerrar",
+      "done": "Listo",
+      "noDescription": "No hay descripción disponible",
+      "promptTemplate": "Plantilla de prompt",
+      "promptInfo": "La configuración avanzada para los prompts se gestiona normalmente a través de la interfaz de plantillas de prompt. Allí puedes crear prompts personalizados o modificar los existentes."
+    },
+    "rightPanel": {
+      "title": "Conceptos básicos de RAG",
+      "subtitle": "La Generación Aumentada por Recuperación (RAG) ayuda al modelo a responder usando tus documentos.",
+      "howItWorksTitle": "Cómo funciona",
+      "howItWorks": {
+        "step1": "Selecciona los documentos que quieres que use el asistente.",
+        "step2": "Los documentos se dividen en fragmentos (chunking) para que la búsqueda sea eficiente.",
+        "step3": "Un recuperador (retriever) encuentra los fragmentos más relevantes para tu pregunta.",
+        "step4": "El LLM genera la respuesta usando esos fragmentos y la plantilla de prompt seleccionada."
+      },
+      "settingsTitle": "Qué hace cada configuración",
+      "settings": {
+        "documents": {
+          "title": "Documentos",
+          "point1": "Tus documentos son la base de conocimiento: el asistente sólo puede usar lo que logre recuperar de ellos.",
+          "point2": "Sube archivos claros y bien estructurados (PDF, TXT, etc.) para mejores resultados.",
+          "point3": "Actualmente sólo documentos de texto son soportados. Imágenes, audio y video aún no son soportados."
+        },
+        "chunking": {
+          "title": "Chunking",
+          "point1": "El chunking divide los documentos en fragmentos más pequeños que se pueden buscar.",
+          "point2": "Chunks muy pequeños: se puede perder contexto y la respuesta puede quedar fragmentada.",
+          "point3": "Chunks muy grandes: puede entrar ruido y la respuesta puede ser más lenta (más tokens).",
+          "point4": "Más solapamiento (overlap) conserva contexto, pero aumenta la repetición."
+        },
+        "retriever": {
+          "title": "Retriever (Sparse vs Dense)",
+          "point1": "El retriever decide qué fragmentos se envían al LLM.",
+          "point2": "Sparse (por palabras clave): ideal para términos exactos como IDs, nombres de producto o código; suele ser rápido.",
+          "point3": "Limitación de sparse: le cuesta con sinónimos o preguntas muy parafraseadas.",
+          "point4": "Dense (semántico): mejor para lenguaje natural y parafraseos; busca por significado.",
+          "point5": "Limitación de dense: puede traer fragmentos ‘parecidos’ pero no exactos, y puede ser más lento."
+        },
+        "llm": {
+          "title": "Modelo de lenguaje (LLM)",
+          "point1": "El LLM redacta la respuesta final usando los fragmentos recuperados.",
+          "point2": "Modelos más grandes suelen responder mejor, pero pueden ser más lentos y costosos.",
+          "point3": "Si priorizas velocidad, empieza con uno más pequeño e itera."
+        },
+        "prompt": {
+          "title": "Plantilla de prompt",
+          "point1": "Los prompts controlan el tono, el formato y en qué debe fijarse el modelo.",
+          "point2": "Si las respuestas son muy largas/cortas o poco estructuradas, prueba otra plantilla o ajusta el prompt."
+        }
+      },
+      "tipsTitle": "Consejos prácticos",
+      "tips": {
+        "tip1": "Si a las respuestas les falta contexto, prueba fragmentos más grandes o más solapamiento.",
+        "tip2": "Si las respuestas son muy genéricas, prueba fragmentos más pequeños o ajusta el modelo recuperador.",
+        "tip3": "Elige una plantilla de prompt que encaje con tu dominio y tono."
+      }
+    },
+    "common": {
+      "loading": "Cargando..."
+    },
+    "chunkingConfig": {
+      "configureTitle": "Configurar modelo de chunking",
+      "modelLabel": "Modelo de chunking"
+    },
+    "generatorConfig": {
+      "configureTitle": "Configurar modelo generador (LLM)",
+      "modelLabel": "Modelo generador"
+    },
+    "retrieverConfig": {
+      "modelLabel": "Modelo de recuperación"
+    },
+    "breadcrumbs": {
+      "generative": "Generativo",
+      "rag": "RAG",
+      "documents": "Documentos",
+      "prompts": "Prompts",
+      "goBack": "Volver",
+      "sessionSuffix": "sesión"
+    },
+    "placeholders": {
+      "title": "Marcadores obligatorios",
+      "insert": "Insertar {{placeholder}}"
+    },
+    "documentPreview": {
+      "title": "Vista previa del documento",
+      "noPreview": "No hay vista previa disponible para este tipo de archivo.",
+      "pdfPreview": "Vista previa de PDF"
+    },
+    "paramsPanel": {
+      "title": "Parámetros RAG",
+      "save": "Guardar",
+      "failedToLoad": "Error al cargar la sesión RAG",
+      "updated": "Parámetros RAG actualizados",
+      "failedToUpdate": "Error al actualizar los parámetros RAG"
+    }
+  },
+  "validation": {
+    "contextSpace": "Con tu configuración actual tienes espacio para tus fuentes (chunks) y {{availableChars}} caracteres para tu mensaje.",
+    "insufficientContextTitle": "Ventana de contexto insuficiente",
+    "insufficientContextDescription": "El modelo seleccionado no tiene suficiente contexto para manejar la configuración actual de tamaño de fragmento y Top K. Por favor, reduce el tamaño de fragmento, el número de resultados (Top K), o selecciona un modelo con una ventana de contexto más grande.",
+    "apiKeyMissingTitle": "API key requerida",
+    "apiKeyMissingDescription": "Este modelo requiere una API key para funcionar. Abre la configuración avanzada para añadir tu API key."
+  },
+  "documentReferences": {
+    "modalTitle": "Referencias del documento",
+    "fallbackTitle": "Documento {{id}}",
+    "chunksCount_one": "Este documento proporcionó {{count}} fragmento para esta respuesta:",
+    "chunksCount_other": "Este documento proporcionó {{count}} fragmentos para esta respuesta:",
+    "chunkLabel": "Fragmento {{position}}",
+    "close": "Cerrar"
+  },
+  "sourcesDisplay": {
+    "viewSources": "Ver fuentes del mensaje",
+    "provided": "proporcionó",
+    "chunk_one": "fragmento",
+    "chunk_other": "fragmentos",
+    "viewChunks": "Ver fragmentos"
+  },
+  "documentsBar": {
+    "title": "Documentos",
+    "documentCount_one": "{{count}} documento",
+    "documentCount_other": "{{count}} documentos",
+    "inCurrentSession": " en la sesión actual",
+    "available": " disponibles",
+    "addDocuments": "Agregar documentos",
+    "searchPlaceholder": "Buscar documentos",
+    "detailedView": "Ver vista detallada",
+    "noDocumentsFound": "No se encontraron documentos",
+    "noDocumentsInSession": "No hay documentos en esta sesión",
+    "noDocumentsAvailable": "No hay documentos disponibles",
+    "uploadDocuments": "Sube tu(s) documento(s)",
+    "failedFetch": "Error al obtener documentos",
+    "successUpload_one": "{{count}} documento subido exitosamente",
+    "successUpload_other": "{{count}} documentos subidos exitosamente",
+    "failedUpload": "Error al subir documento(s)"
+  },
+  "ragPromptsPage": {
+    "title": "Prompts RAG",
+    "description": "Gestiona los prompts para tus sesiones RAG: ve todos los prompts disponibles y crea nuevos para mejorar tus interacciones de IA."
+  },
+  "ragDocumentsPage": {
+    "title": "Documentos RAG",
+    "description": "Gestiona los documentos para tus sesiones RAG: ve los documentos existentes y sube nuevos para mejorar tu base de conocimiento de IA."
   }
 }
diff --git a/DashAI/front/src/utils/i18n/locales/pt/generative.json b/DashAI/front/src/utils/i18n/locales/pt/generative.json
index c2b513a51..077afa1c7 100644
--- a/DashAI/front/src/utils/i18n/locales/pt/generative.json
+++ b/DashAI/front/src/utils/i18n/locales/pt/generative.json
@@ -57,11 +57,376 @@
     "attachMedia": "Anexar mídia",
     "attachMediaToContinue": "Anexe uma mídia para continuar",
     "noInputAvailable": "Não há entrada disponível para esta tarefa",
-    "selectModelCrumb": "Selecionar Modelo"
+    "selectModelCrumb": "Selecionar Modelo",
+    "ragTask": "RAG - Geração Aumentada por Recuperação",
+    "ragTaskDescription": "Faça upload de documentos e converse com eles usando geração aumentada por recuperação."
   },
   "message": {
     "sessionCreatedSuccess": "Sessão criada com sucesso.",
     "sessionDeleted": "Sessão excluída com sucesso",
     "sessionUpdated": "Sessão atualizada com sucesso"
+  },
+  "rag": {
+    "setup": {
+      "title": "Criar sessão RAG",
+      "subtitle": "Forneça informações básicas, selecione documentos e configure os componentes para sua sessão RAG.",
+      "sessionDetails": "Detalhes da sessão",
+      "sessionName": "Nome da sessão *",
+      "sessionNamePlaceholder": "ex. Documentação de Produto RAG",
+      "sessionNameDuplicate": "Este nome de sessão já existe",
+      "sessionNameDuplicateAlert": "Já existe uma sessão com este nome. Por favor, use um nome diferente.",
+      "description": "Descrição (Opcional)",
+      "descriptionPlaceholder": "Descreva o propósito desta sessão RAG...",
+      "selectDocuments": "Selecionar documentos",
+      "selectDocumentsDescription": "Faça upload de novos documentos ou selecione entre os existentes para usar no RAG.",
+      "chunkingStrategy": "Estratégia de chunking",
+      "retrieverModel": "Modelo de recuperação",
+      "promptTemplate": "Modelo de Prompt",
+      "languageModel": "Modelo de linguagem (LLM)",
+      "cancel": "Cancelar",
+      "saveSession": "Salvar sessão",
+      "openDocumentsLibrary": "Abrir biblioteca de documentos"
+    },
+    "validation": {
+      "nameRequired": "O nome da sessão é obrigatório",
+      "nameUnique": "O nome da sessão deve ser único",
+      "documentsRequired": "Pelo menos um documento deve ser selecionado",
+      "chunkingRequired": "O modelo de chunking deve ser configurado",
+      "retrieverRequired": "O modelo de recuperação deve ser configurado",
+      "generatorRequired": "O modelo de geração deve ser configurado",
+      "generatorInvalid": "A configuração do gerador é inválida (verifique a janela de contexto)",
+      "promptRequired": "Um modelo de prompt deve ser selecionado"
+    },
+    "messages": {
+      "success": "Sessão RAG criada com sucesso"
+    },
+    "chunking": {
+      "description": "Selecione como seus documentos serão divididos em fragmentos para recuperação.",
+      "presets": {
+        "small": {
+          "label": "Fragmentos pequenos"
+        },
+        "paragraph": {
+          "label": "Tamanho de parágrafo"
+        },
+        "page": {
+          "label": "Fragmento de página"
+        },
+        "large": {
+          "label": "Seções grandes"
+        },
+        "chunkSizeFormat": "{{chars}} caracteres ≈ {{tokens}} tokens",
+        "custom": {
+          "label": "Modelo de chunking personalizado",
+          "description": "Configuração avançada aplicada"
+        }
+      },
+      "advancedButton": "Abrir configuração avançada"
+    },
+    "retriever": {
+      "description": "Selecione seu paradigma de recuperação. O recuperador buscará em seus documentos e retornará os fragmentos mais relevantes.",
+      "keywordLabel": "Palavra-chave",
+      "semanticLabel": "Semântico",
+      "hybridLabel": "Híbrido",
+      "hybridDescription": "BM25 + Harrier 0.6B",
+      "advancedApplied": "Configuração avançada aplicada",
+      "topKLabel": "Número de fragmentos (Top K)",
+      "topKCardDescription": "Fragmentos",
+      "advancedTopK": "Top K avançado: {{topK}}",
+      "advancedButton": "Abrir configuração avançada",
+      "modifyFromAdvanced": "O Top K deve ser modificado na visualização de configuração avançada",
+      "explanations": {
+        "SparseRetriever": "Ideal para busca por palavras-chave e IDs específicos. Útil para encontrar correspondências exatas em documentos.",
+        "DenseRetriever": "Busca contextual que entende o significado. Ótimo para encontrar conteúdo semanticamente similar, mas pode perder palavras-chave ou IDs específicos.",
+        "HybridRetriever": "Combina busca por palavras-chave e contextual para desempenho equilibrado em diferentes tipos de consulta."
+      }
+    },
+    "composite": {
+      "addChild": "Adicionar nó recuperador",
+      "removeNode": "Remover nó",
+      "configureNode": "Configurar Nó",
+      "selectModel": "Selecionar modelo recuperador",
+      "parameters": "Parâmetros",
+      "simpleGroup": "Recuperadores simples",
+      "keywordGroup": "Recuperadores por palavras-chave",
+      "denseGroup": "Recuperadores densos",
+      "embeddingGroup": "Recuperadores por Embeddings",
+      "hybridGroup": "Híbrido",
+      "hybridDescription": "BM25 + embeddings multilíngues HuggingFace",
+      "compositeGroup": "Recuperadores compostos",
+      "nestedCompositeNote": "Este é um recuperador composto. Seus filhos podem ser configurados na árvore abaixo.",
+      "allChunks": "Todos os fragmentos",
+      "selectedChunks": "Fragmentos selecionados",
+      "retrieverDescription": "Escolha um recuperador para configurar seus parâmetros. Os recuperadores por palavras-chave (BM25, TF-IDF) encontram documentos por frequência de termos. Os recuperadores por embeddings capturam o significado semântico com vetores densos. Os recuperadores compostos encadeiam vários recuperadores, de forma sequencial ou em paralelo.",
+      "compositeInstructions": "Clique no botão + para adicionar um recuperador filho a este composto. Clique em um cartão para editar seus parâmetros ou escolher outro modelo."
+    },
+    "generator": {
+      "description": "Selecione o modelo de linguagem que gerará respostas com base no contexto recuperado e seus prompts.",
+      "modelLabel": "Modelo de geração",
+      "advancedApplied": "Configuração avançada aplicada",
+      "selectModel": "Selecionar modelo de linguagem",
+      "selectModelPlaceholder": "ex. GPT-4, Claude, Llama",
+      "modelInfo": "Modelo:",
+      "advancedButton": "Abrir configuração avançada"
+    },
+    "prompt": {
+      "description": "Selecione um modelo de prompt que define como o contexto recuperado e as mensagens do chat são combinados para gerar respostas.",
+      "availablePrompts": "Prompts disponíveis",
+      "openPrompts": "Abrir biblioteca de prompts",
+      "promptLabel": "Prompt",
+      "selectTemplate": "Selecionar modelo de prompt",
+      "selectTemplatePlaceholder": "ex. Prompt Padrão, Instrução Personalizada",
+      "selectedTemplate": "Modelo de prompt selecionado:",
+      "newPromptButton": "Novo prompt",
+      "defaultGenerationPrompt": "Prompt de Geração Padrão",
+      "defaultQAGenerationPrompt": "Prompt de Geração Q&A Padrão",
+      "createNewPrompt": "Criar novo prompt",
+      "language": "Idioma",
+      "collapse": "Recolher",
+      "expand": "Expandir",
+      "descriptionToggle": "Descrição",
+      "languages": {
+        "en": "English",
+        "es": "Español",
+        "pt": "Português"
+      }
+    },
+    "promptView": {
+      "type": "Tipo",
+      "language": "Idioma",
+      "languageNotAvailable": "Idioma não disponível",
+      "languageNone": "Sem idioma",
+      "noContent": "Sem conteúdo de template",
+      "close": "Fechar",
+      "templateContent": "Conteúdo do template",
+      "untitledPrompt": "Prompt",
+      "table": {
+        "id": "ID",
+        "name": "Nome",
+        "type": "Tipo",
+        "language": "Idioma",
+        "created": "Criado",
+        "edited": "Editado",
+        "actions": "Ações",
+        "viewPrompt": "Ver prompt",
+        "currentPrompts": "Prompts atuais",
+        "choosePrompt": "Escolha ou personalize seu prompt para definir o comportamento do modelo.",
+        "newPrompt": "Novo prompt"
+      }
+    },
+    "newPrompt": {
+      "title": "Criar um novo prompt",
+      "description": "O template do prompt define como os fragmentos (partes dos documentos) e as mensagens do chat são integrados para gerar respostas. Personalize o prompt para ajustar o comportamento das suas sessões RAG.",
+      "placeholdersInfo": "Use {chunks} para representar onde os fragmentos dos documentos recuperados serão inseridos, e {input} para a mensagem do usuário.",
+      "nameLabel": "Nome do Prompt",
+      "nameRequired": "O nome do prompt é obrigatório",
+      "promptLabel": "Prompt",
+      "promptPlaceholder": "Aqui você pode modificar o prompt, por exemplo:\nCada mensagem do usuário é adicionada como {input}\nAs fontes são adicionadas como {chunks}",
+      "cancel": "Cancelar",
+      "save": "Salvar",
+      "unsavedChanges": "Você tem alterações não salvas. Tem certeza de que deseja cancelar?",
+      "success": "Prompt criado com sucesso!",
+      "error": "Falha ao criar o prompt",
+      "languageLabel": "Idioma (opcional)"
+    },
+    "advanced": {
+      "chunkingTitle": "Configuração avançada de chunking",
+      "retrieverTitle": "Configuração avançada de recuperação",
+      "generatorTitle": "Configuração avançada de geração",
+      "promptTitle": "Configuração avançada de prompt",
+      "close": "Fechar",
+      "done": "Concluído",
+      "noDescription": "Nenhuma descrição disponível",
+      "promptTemplate": "Modelo de Prompt",
+      "promptInfo": "A configuração avançada para prompts é geralmente gerenciada através da interface de modelos de prompt. Você pode criar prompts personalizados ou modificar os existentes lá."
+    },
+    "rightPanel": {
+      "title": "Conceitos básicos de RAG",
+      "subtitle": "A Geração Aumentada por Recuperação (RAG) ajuda o modelo a responder usando seus documentos.",
+      "howItWorksTitle": "Como funciona",
+      "howItWorks": {
+        "step1": "Selecione os documentos que deseja que o assistente use.",
+        "step2": "Os documentos são divididos em fragmentos (chunking) para tornar a recuperação eficiente.",
+        "step3": "Um recuperador (retriever) encontra os fragmentos mais relevantes para sua pergunta.",
+        "step4": "O LLM gera uma resposta usando esses fragmentos e o modelo de prompt selecionado."
+      },
+      "settingsTitle": "O que cada configuração faz",
+      "settings": {
+        "documents": {
+          "title": "Documentos",
+          "point1": "Seus documentos são a base de conhecimento: o assistente só pode usar o que conseguir recuperar deles.",
+          "point2": "Faça upload de arquivos claros e bem estruturados (PDF, TXT, etc.) para melhores resultados.",
+          "point3": "Atualmente, apenas documentos de texto são suportados. Imagens, áudio e vídeo ainda não são suportados."
+        },
+        "chunking": {
+          "title": "Chunking",
+          "point1": "O chunking divide os documentos em pedaços menores que podem ser pesquisados.",
+          "point2": "Fragmentos muito pequenos: a recuperação pode perder contexto e as respostas podem parecer fragmentadas.",
+          "point3": "Fragmentos muito grandes: a recuperação pode trazer ruído extra e as respostas podem ser mais lentas (mais tokens).",
+          "point4": "A sobreposição (overlap) ajuda a manter o contexto entre os fragmentos, mas aumenta a repetição."
+        },
+        "retriever": {
+          "title": "Recuperador (Sparse vs Denso)",
+          "point1": "O recuperador decide quais fragmentos são enviados ao LLM.",
+          "point2": "Sparse (por palavras-chave): ótimo para termos exatos como IDs, nomes de produtos ou código; geralmente rápido.",
+          "point3": "Limitação do sparse: tem dificuldade com sinônimos e perguntas muito parafraseadas.",
+          "point4": "Denso (semântico): melhor para linguagem natural e paráfrases; busca por significado.",
+          "point5": "Limitação do denso: pode retornar fragmentos 'semelhantes' mas não exatos, e pode ser mais lento."
+        },
+        "llm": {
+          "title": "Modelo de Linguagem (LLM)",
+          "point1": "O LLM escreve a resposta final usando os fragmentos recuperados.",
+          "point2": "Modelos maiores geralmente respondem melhor, mas podem ser mais lentos e custar mais.",
+          "point3": "Se precisar de velocidade, comece com um modelo menor e iterar."
+        },
+        "prompt": {
+          "title": "Modelo de prompt",
+          "point1": "Os prompts controlam o tom, formato e no que o modelo deve focar.",
+          "point2": "Se as respostas forem muito longas/curtas ou sem estrutura, tente outro modelo ou refine o prompt."
+        }
+      },
+      "tipsTitle": "Dicas práticas",
+      "tips": {
+        "tip1": "Se as respostas perderem contexto, tente fragmentos maiores ou mais sobreposição.",
+        "tip2": "Se as respostas parecerem muito genéricas, tente fragmentos menores ou ajuste o modelo recuperador.",
+        "tip3": "Escolha um modelo de prompt que corresponda ao seu domínio e tom."
+      }
+    },
+    "common": {
+      "loading": "Carregando..."
+    },
+    "chunkingConfig": {
+      "configureTitle": "Configurar modelo de chunking",
+      "modelLabel": "Modelo de chunking"
+    },
+    "generatorConfig": {
+      "configureTitle": "Configurar modelo gerador (LLM)",
+      "modelLabel": "Modelo gerador"
+    },
+    "retrieverConfig": {
+      "modelLabel": "Modelo de recuperação"
+    },
+    "breadcrumbs": {
+      "generative": "Generativo",
+      "rag": "RAG",
+      "documents": "Documentos",
+      "prompts": "Prompts",
+      "goBack": "Voltar",
+      "sessionSuffix": "sessão"
+    },
+    "placeholders": {
+      "title": "Marcadores obrigatórios",
+      "insert": "Inserir {{placeholder}}"
+    },
+    "documentPreview": {
+      "title": "Visualização do documento",
+      "noPreview": "Não há visualização disponível para este tipo de arquivo.",
+      "pdfPreview": "Visualização de PDF"
+    },
+    "paramsPanel": {
+      "title": "Parâmetros RAG",
+      "save": "Salvar",
+      "failedToLoad": "Erro ao carregar a sessão RAG",
+      "updated": "Parâmetros RAG atualizados",
+      "failedToUpdate": "Erro ao atualizar os parâmetros RAG"
+    },
+    "documents": {
+      "uploadButton": "Upload de documentos",
+      "emptyUploadText": "Faça upload do(s) seu(s) documento(s)",
+      "table": {
+        "id": "ID",
+        "name": "Nome",
+        "type": "Tipo",
+        "created": "Criado",
+        "lastModified": "Última modificação",
+        "actions": "Ações",
+        "preview": "Visualizar",
+        "delete": "Excluir",
+        "addDocument": "Adicionar documento",
+        "currentDocuments": "Documentos atuais",
+        "noDocumentsAvailable": "Nenhum documento disponível.",
+        "emptyUploadText": "Envie seu(s) documento(s)",
+        "unknownType": "Desconhecido"
+      }
+    },
+    "summary": {
+      "title": "Sessão RAG",
+      "descriptionPlaceholder": "Nenhuma descrição disponível",
+      "createdLabel": "Criado",
+      "openChatButton": "Abrir chat da sessão RAG",
+      "detailedConfig": "Configuração detalhada",
+      "chunkingModel": "Modelo de chunking",
+      "retrieverModel": "Modelo de recuperação",
+      "generationModel": "Modelo de geração",
+      "promptSelected": "Prompt selecionado",
+      "promptName": "Nome:",
+      "promptTemplate": "Modelo:",
+      "noPrompt": "Nenhum prompt configurado",
+      "sessionNotFound": "Sessão não encontrada",
+      "loading": "Carregando dados da sessão...",
+      "apiKeyCopied": "Chave da API copiada para a área de transferência",
+      "failedToCopyApiKey": "Falha ao copiar a chave da API",
+      "noParameters": "Sem parâmetros",
+      "viewDetails": "Ver detalhes",
+      "parameterDetails": "Detalhes do parâmetro",
+      "completeConfig": "Configuração completa do parâmetro:",
+      "close": "Fechar",
+      "sessionName": "Nome da sessão",
+      "sessionDescription": "Descrição da sessão",
+      "parameterDetailTitle": "{{component}} - {{param}}",
+      "edit": "Editar",
+      "save": "Salvar",
+      "cancel": "Cancelar",
+      "sessionUpdated": "Sessão atualizada com sucesso",
+      "failedToUpdateSession": "Falha ao atualizar a sessão"
+    }
+  },
+  "validation": {
+    "contextSpace": "Com sua configuração atual, você tem espaço para suas fontes (chunks) e {{availableChars}} caracteres para sua mensagem.",
+    "insufficientContextTitle": "Janela de Contexto Insuficiente",
+    "insufficientContextDescription": "O modelo selecionado não tem contexto suficiente para lidar com as configurações atuais de Tamanho do Fragmento e Top K. Reduza o Tamanho do Fragmento, o número de resultados (Top K) ou selecione um modelo com uma janela de contexto maior.",
+    "apiKeyMissingTitle": "Chave de API necessária",
+    "apiKeyMissingDescription": "Este modelo requer uma chave de API para funcionar. Abra a Configuração Avançada para adicionar sua chave de API."
+  },
+  "documentReferences": {
+    "modalTitle": "Referências do documento",
+    "fallbackTitle": "Documento {{id}}",
+    "chunksCount_one": "Este documento forneceu {{count}} fragmento para esta resposta:",
+    "chunksCount_other": "Este documento forneceu {{count}} fragmentos para esta resposta:",
+    "chunkLabel": "Fragmento {{position}}",
+    "close": "Fechar"
+  },
+  "sourcesDisplay": {
+    "viewSources": "Ver fontes da mensagem",
+    "provided": "forneceu",
+    "chunk_one": "fragmento",
+    "chunk_other": "fragmentos",
+    "viewChunks": "Ver fragmentos"
+  },
+  "documentsBar": {
+    "title": "Documentos",
+    "documentCount_one": "{{count}} documento",
+    "documentCount_other": "{{count}} documentos",
+    "inCurrentSession": " na sessão atual",
+    "available": " disponíveis",
+    "addDocuments": "Adicionar documentos",
+    "searchPlaceholder": "Pesquisar documentos",
+    "detailedView": "Ver visualização detalhada",
+    "noDocumentsFound": "Nenhum documento encontrado",
+    "noDocumentsInSession": "Nenhum documento nesta sessão",
+    "noDocumentsAvailable": "Nenhum documento disponível",
+    "uploadDocuments": "Faça upload do(s) seu(s) documento(s)",
+    "failedFetch": "Falha ao carregar documentos",
+    "successUpload_one": "{{count}} documento enviado com sucesso",
+    "successUpload_other": "{{count}} documentos enviados com sucesso",
+    "failedUpload": "Falha ao enviar documento(s)"
+  },
+  "ragPromptsPage": {
+    "title": "Prompts RAG",
+    "description": "Gerencie os prompts para suas sessões RAG: veja todos os prompts disponíveis e crie novos para melhorar suas interações de IA."
+  },
+  "ragDocumentsPage": {
+    "title": "Documentos RAG",
+    "description": "Gerencie os documentos para suas sessões RAG: veja os documentos existentes e faça upload de novos para melhorar sua base de conhecimento de IA."
   }
 }
diff --git a/DashAI/front/src/utils/i18n/locales/zh/generative.json b/DashAI/front/src/utils/i18n/locales/zh/generative.json
index bd86dc015..10a171892 100644
--- a/DashAI/front/src/utils/i18n/locales/zh/generative.json
+++ b/DashAI/front/src/utils/i18n/locales/zh/generative.json
@@ -57,11 +57,377 @@
     "attachMedia": "附加媒体",
     "attachMediaToContinue": "附加媒体以继续",
     "noInputAvailable": "此任务无可用输入",
-    "selectModelCrumb": "选择模型"
+    "selectModelCrumb": "选择模型",
+    "ragTask": "RAG - 检索增强生成",
+    "ragTaskDescription": "上传文档并使用检索增强生成与它们对话。"
   },
   "message": {
     "sessionCreatedSuccess": "会话创建成功。",
     "sessionDeleted": "会话删除成功",
     "sessionUpdated": "会话更新成功"
+  },
+  "rag": {
+    "setup": {
+      "title": "创建 RAG 会话",
+      "subtitle": "提供基本信息，选择文档，并为您的 RAG 会话配置组件。",
+      "sessionDetails": "会话详情",
+      "sessionName": "会话名称 *",
+      "sessionNamePlaceholder": "例如：产品文档 RAG",
+      "sessionNameDuplicate": "此会话名称已存在",
+      "sessionNameDuplicateAlert": "已存在同名会话。请使用其他名称。",
+      "description": "描述（可选）",
+      "descriptionPlaceholder": "描述此 RAG 会话的用途...",
+      "selectDocuments": "选择文档",
+      "selectDocumentsDescription": "上传新文档或从现有文档中选择以用于 RAG。",
+      "chunkingStrategy": "分块策略",
+      "retrieverModel": "检索模型",
+      "promptTemplate": "提示词模板",
+      "languageModel": "语言模型 (LLM)",
+      "cancel": "取消",
+      "saveSession": "保存会话",
+      "openDocumentsLibrary": "打开文档库"
+    },
+    "validation": {
+      "nameRequired": "会话名称为必填项",
+      "nameUnique": "会话名称必须唯一",
+      "documentsRequired": "必须至少选择一个文档",
+      "chunkingRequired": "必须配置分块模型",
+      "retrieverRequired": "必须配置检索模型",
+      "generatorRequired": "必须配置生成器模型",
+      "generatorInvalid": "生成器配置无效（请检查上下文窗口）",
+      "promptRequired": "必须选择提示词模板"
+    },
+    "messages": {
+      "success": "RAG 会话创建成功"
+    },
+    "chunking": {
+      "description": "选择您的文档如何拆分为用于检索的块。",
+      "presets": {
+        "small": {
+          "label": "小块"
+        },
+        "paragraph": {
+          "label": "段落长度"
+        },
+        "page": {
+          "label": "页面块"
+        },
+        "large": {
+          "label": "大节"
+        },
+        "chunkSizeFormat": "{{chars}} 字符 ≈ {{tokens}} 标记",
+        "custom": {
+          "label": "自定义分块模型",
+          "description": "已应用高级配置"
+        }
+      },
+      "advancedButton": "打开高级配置"
+    },
+    "retriever": {
+      "description": "选择您的检索范式。检索器将搜索您的文档并返回最相关的块。",
+      "paradigmLabel": "检索范式",
+      "keywordLabel": "关键词",
+      "semanticLabel": "语义",
+      "hybridLabel": "混合",
+      "hybridDescription": "BM25 + Harrier 0.6B",
+      "advancedApplied": "已应用高级配置",
+      "topKLabel": "块数量 (Top K)",
+      "topKCardDescription": "块",
+      "advancedTopK": "高级 Top K: {{topK}}",
+      "advancedButton": "打开高级配置",
+      "modifyFromAdvanced": "必须从高级配置视图中修改 Top K",
+      "explanations": {
+        "SparseRetriever": "最适合特定关键词和 ID 搜索。用于查找文档中的精确匹配。",
+        "DenseRetriever": "理解语义的上下文搜索。适合查找语义相似的内容，但可能遗漏特定关键词或 ID。",
+        "HybridRetriever": "结合关键词搜索和上下文搜索，在不同查询类型间提供平衡的性能。"
+      }
+    },
+    "composite": {
+      "addChild": "添加检索节点",
+      "removeNode": "删除节点",
+      "configureNode": "配置节点",
+      "selectModel": "选择检索模型",
+      "parameters": "参数",
+      "simpleGroup": "简单检索器",
+      "keywordGroup": "关键词检索器",
+      "denseGroup": "稠密检索器",
+      "embeddingGroup": "嵌入检索器",
+      "hybridGroup": "混合",
+      "hybridDescription": "BM25 + HuggingFace 多语言嵌入",
+      "compositeGroup": "组合检索器",
+      "nestedCompositeNote": "这是一个组合检索器。其子节点可以在下方的树中配置。",
+      "allChunks": "所有块",
+      "selectedChunks": "已选块",
+      "retrieverDescription": "选择一个检索器来配置其参数。关键词检索器（BM25、TF-IDF）通过词频查找文档。嵌入检索器通过稠密向量捕获语义含义。组合检索器按顺序或并行地链接多个检索器。",
+      "compositeInstructions": "点击 + 按钮向此组合添加子检索器。点击检索器卡片以编辑其参数或选择其他模型。"
+    },
+    "generator": {
+      "description": "选择将根据检索到的上下文和您的提示词生成回复的语言模型。",
+      "modelLabel": "生成器模型",
+      "advancedApplied": "已应用高级配置",
+      "selectModel": "选择语言模型",
+      "selectModelPlaceholder": "例如：GPT-4、Claude、Llama",
+      "modelInfo": "模型：",
+      "advancedButton": "打开高级配置"
+    },
+    "prompt": {
+      "description": "选择一种提示词模板，定义如何组合检索到的上下文和聊天消息以生成回复。",
+      "availablePrompts": "可用的提示词",
+      "openPrompts": "打开提示词库",
+      "promptLabel": "提示词",
+      "selectTemplate": "选择提示词模板",
+      "selectTemplatePlaceholder": "例如：默认提示词、自定义指令",
+      "selectedTemplate": "已选择的提示词模板：",
+      "newPromptButton": "新建提示词",
+      "defaultGenerationPrompt": "默认生成提示词",
+      "defaultQAGenerationPrompt": "默认问答生成提示词",
+      "createNewPrompt": "创建新提示词",
+      "language": "语言",
+      "collapse": "收起",
+      "expand": "展开",
+      "descriptionToggle": "描述",
+      "languages": {
+        "en": "English",
+        "es": "Español",
+        "pt": "Português"
+      }
+    },
+    "promptView": {
+      "type": "类型",
+      "language": "语言",
+      "languageNotAvailable": "语言不可用",
+      "languageNone": "无语言",
+      "noContent": "无模板内容",
+      "close": "关闭",
+      "templateContent": "模板内容",
+      "untitledPrompt": "提示词",
+      "table": {
+        "id": "ID",
+        "name": "名称",
+        "type": "类型",
+        "language": "语言",
+        "created": "创建时间",
+        "edited": "编辑时间",
+        "actions": "操作",
+        "viewPrompt": "查看提示词",
+        "currentPrompts": "当前提示词",
+        "choosePrompt": "选择或自定义您的提示词以定义模型的行为。",
+        "newPrompt": "新建提示词"
+      }
+    },
+    "documents": {
+      "uploadButton": "上传文档",
+      "emptyUploadText": "上传您的文档",
+      "table": {
+        "id": "ID",
+        "name": "名称",
+        "type": "类型",
+        "created": "创建时间",
+        "lastModified": "最后修改",
+        "actions": "操作",
+        "preview": "预览",
+        "delete": "删除",
+        "addDocument": "添加文档",
+        "currentDocuments": "当前文档",
+        "noDocumentsAvailable": "没有可用的文档。",
+        "emptyUploadText": "上传您的文档",
+        "unknownType": "未知"
+      }
+    },
+    "summary": {
+      "title": "RAG 会话",
+      "descriptionPlaceholder": "暂无描述",
+      "createdLabel": "创建时间",
+      "openChatButton": "打开 RAG 会话聊天",
+      "detailedConfig": "详细配置",
+      "chunkingModel": "分块模型",
+      "retrieverModel": "检索模型",
+      "generationModel": "生成模型",
+      "promptSelected": "已选提示词",
+      "promptName": "名称：",
+      "promptTemplate": "模板：",
+      "noPrompt": "未配置提示词",
+      "sessionNotFound": "未找到会话",
+      "loading": "正在加载会话数据...",
+      "apiKeyCopied": "API 密钥已复制到剪贴板",
+      "failedToCopyApiKey": "复制 API 密钥失败",
+      "noParameters": "无参数",
+      "viewDetails": "查看详情",
+      "parameterDetails": "参数详情",
+      "completeConfig": "完整参数配置：",
+      "close": "关闭",
+      "sessionName": "会话名称",
+      "sessionDescription": "会话描述",
+      "parameterDetailTitle": "{{component}} - {{param}}",
+      "edit": "编辑",
+      "save": "保存",
+      "cancel": "取消",
+      "sessionUpdated": "会话更新成功",
+      "failedToUpdateSession": "更新会话失败"
+    },
+    "newPrompt": {
+      "title": "创建新提示词",
+      "description": "提示词模板定义了如何集成文档块和聊天消息以生成回复。自定义提示词以调整您的 RAG 会话行为。",
+      "placeholdersInfo": "使用 {chunks} 表示检索到的文档块将被插入的位置，使用 {input} 表示用户消息。",
+      "nameLabel": "提示词名称",
+      "nameRequired": "提示词名称为必填项",
+      "promptLabel": "提示词",
+      "promptPlaceholder": "您可以在此修改提示词，例如：\n每条用户消息添加为 {input}\n来源添加为 {chunks}",
+      "cancel": "取消",
+      "save": "保存",
+      "unsavedChanges": "您有未保存的更改。确定要取消吗？",
+      "success": "提示词创建成功！",
+      "error": "创建提示词失败",
+      "languageLabel": "语言（可选）"
+    },
+    "advanced": {
+      "chunkingTitle": "高级分块配置",
+      "retrieverTitle": "高级检索配置",
+      "generatorTitle": "高级生成器配置",
+      "promptTitle": "高级提示词配置",
+      "close": "关闭",
+      "done": "完成",
+      "noDescription": "暂无描述",
+      "promptTemplate": "提示词模板",
+      "promptInfo": "提示词的高级配置通常通过提示词模板界面管理。您可以在那里创建自定义提示词或修改现有模板。"
+    },
+    "rightPanel": {
+      "title": "RAG 基础",
+      "subtitle": "检索增强生成 (RAG) 帮助模型使用您的文档回答问题。",
+      "howItWorksTitle": "工作原理",
+      "howItWorks": {
+        "step1": "选择您希望助手使用的文档。",
+        "step2": "文档被拆分为较小的块（分块）以使检索高效。",
+        "step3": "检索器找到与您问题最相关的块。",
+        "step4": "LLM 使用这些块和选择的提示词模板生成答案。"
+      },
+      "settingsTitle": "各设置的作用",
+      "settings": {
+        "documents": {
+          "title": "文档",
+          "point1": "您的文档是知识库：助手只能使用其能够从中检索到的内容。",
+          "point2": "上传清晰、结构良好的文件（PDF、TXT 等）以获得最佳效果。",
+          "point3": "目前仅支持基于文本的文档。图像、音频和视频尚不支持。"
+        },
+        "chunking": {
+          "title": "分块",
+          "point1": "分块将文档拆分为可搜索的较小片段。",
+          "point2": "块太小：检索可能会丢失上下文，回答可能显得支离破碎。",
+          "point3": "块太大：检索可能带来额外噪音，回答可能更慢（更多标记）。",
+          "point4": "重叠有助于保持块之间的上下文，但会增加重复。"
+        },
+        "retriever": {
+          "title": "检索器（稀疏 vs 稠密）",
+          "point1": "检索器决定将哪些块发送给 LLM。",
+          "point2": "稀疏（关键词）：适合精确术语如 ID、产品名称或代码；通常快速。",
+          "point3": "稀疏限制：难以处理同义词和高度改述的问题。",
+          "point4": "稠密（语义）：更适合自然语言和改述；通过含义搜索。",
+          "point5": "稠密限制：可能返回「相似」但不精确的匹配，且可能较慢。"
+        },
+        "llm": {
+          "title": "语言模型 (LLM)",
+          "point1": "LLM 使用检索到的块编写最终答案。",
+          "point2": "较大的模型通常回答更好，但可能更慢且更昂贵。",
+          "point3": "如果您需要速度，从较小的模型开始并迭代。"
+        },
+        "prompt": {
+          "title": "提示词模板",
+          "point1": "提示词控制语气、格式以及模型应关注的内容。",
+          "point2": "如果回答太长/太短或缺乏结构，请尝试其他模板或优化提示词。"
+        }
+      },
+      "tipsTitle": "实用技巧",
+      "tips": {
+        "tip1": "如果回答缺少上下文，请尝试更大的块或更多重叠。",
+        "tip2": "如果回答感觉太泛泛，请尝试更小的块或调整检索模型。",
+        "tip3": "选择与您领域和语气匹配的提示词模板。"
+      }
+    },
+    "common": {
+      "loading": "加载中..."
+    },
+    "chunkingConfig": {
+      "configureTitle": "配置分块模型",
+      "modelLabel": "分块模型"
+    },
+    "generatorConfig": {
+      "configureTitle": "配置生成器模型 (LLM)",
+      "modelLabel": "生成器模型"
+    },
+    "retrieverConfig": {
+      "modelLabel": "检索模型"
+    },
+    "breadcrumbs": {
+      "generative": "生成式",
+      "rag": "RAG",
+      "documents": "文档",
+      "prompts": "提示词",
+      "goBack": "返回",
+      "sessionSuffix": "会话"
+    },
+    "placeholders": {
+      "title": "必填占位符",
+      "insert": "插入 {{placeholder}}"
+    },
+    "documentPreview": {
+      "title": "文档预览",
+      "noPreview": "此文件类型没有可用的预览。",
+      "pdfPreview": "PDF 预览"
+    },
+    "paramsPanel": {
+      "title": "RAG 参数",
+      "save": "保存",
+      "failedToLoad": "加载 RAG 会话失败",
+      "updated": "RAG 参数已更新",
+      "failedToUpdate": "更新 RAG 参数失败"
+    }
+  },
+  "validation": {
+    "contextSpace": "使用当前配置，您有空间容纳源（块）和 {{availableChars}} 个字符用于您的消息。",
+    "insufficientContextTitle": "上下文窗口不足",
+    "insufficientContextDescription": "所选模型没有足够的上下文来处理当前的块大小和 Top K 设置。请减小块大小、减少结果数量 (Top K)，或选择具有更大上下文窗口的模型。",
+    "apiKeyMissingTitle": "需要 API 密钥",
+    "apiKeyMissingDescription": "此模型需要 API 密钥才能运行。打开高级配置以添加您的 API 密钥。"
+  },
+  "documentReferences": {
+    "modalTitle": "文档引用",
+    "fallbackTitle": "文档 {{id}}",
+    "chunksCount_one": "此文档为此回复提供了 {{count}} 个块：",
+    "chunksCount_other": "此文档为此回复提供了 {{count}} 个块：",
+    "chunkLabel": "第 {{position}} 块",
+    "close": "关闭"
+  },
+  "sourcesDisplay": {
+    "viewSources": "查看消息来源",
+    "provided": "提供了",
+    "chunk_one": "个块",
+    "chunk_other": "个块",
+    "viewChunks": "查看块"
+  },
+  "documentsBar": {
+    "title": "文档",
+    "documentCount_one": "{{count}} 个文档",
+    "documentCount_other": "{{count}} 个文档",
+    "inCurrentSession": " 在当前会话中",
+    "available": " 可用",
+    "addDocuments": "添加文档",
+    "searchPlaceholder": "搜索文档",
+    "detailedView": "查看详细视图",
+    "noDocumentsFound": "未找到文档",
+    "noDocumentsInSession": "此会话中没有文档",
+    "noDocumentsAvailable": "没有可用的文档",
+    "uploadDocuments": "上传您的文档",
+    "failedFetch": "获取文档失败",
+    "successUpload_one": "成功上传 {{count}} 个文档",
+    "successUpload_other": "成功上传 {{count}} 个文档",
+    "failedUpload": "上传文档失败"
+  },
+  "ragPromptsPage": {
+    "title": "RAG 提示词",
+    "description": "管理 RAG 会话的提示词：查看所有可用的提示词并创建新的，以提升您的 AI 交互体验。"
+  },
+  "ragDocumentsPage": {
+    "title": "RAG 文档",
+    "description": "管理 RAG 会话的文档：查看现有文档并上传新的，以增强您的 AI 知识库。"
   }
 }
diff --git a/DashAI/front/src/utils/schema.js b/DashAI/front/src/utils/schema.js
index 5f82c0b83..e220e5429 100644
--- a/DashAI/front/src/utils/schema.js
+++ b/DashAI/front/src/utils/schema.js
@@ -1,6 +1,20 @@
 import * as Yup from "yup";
 import { getComponents } from "../api/component";
 
+export async function resolveDefaults(modelName) {
+  try {
+    const result = await getComponents({ model: modelName });
+    const info = Array.isArray(result) ? result[0] : result;
+    if (!info?.schema) return {};
+    const formatted = await formattedModel(info.schema);
+    const { initialValues } = generateYupSchema(formatted);
+    return initialValues;
+  } catch (e) {
+    console.warn(`[resolveDefaults] Failed for ${modelName}:`, e);
+    return {};
+  }
+}
+
 // Generate a Yup schema from a JSON schema object based on the JSON schema specification from the api, it also generates the initial values of the form
 export const generateYupSchema = (schemaObj) => {
   const schema = {};
@@ -17,7 +31,7 @@ export const generateYupSchema = (schemaObj) => {
   return { schema: Yup.object().shape(schema), initialValues };
 };
 
-const generateInitialValues = (subSchema) => {
+export const generateInitialValues = (subSchema) => {
   let initialValues = {};
 
   // Special case for optimizable fields
@@ -348,18 +362,24 @@ export const getParamsFromSubform = (subform) => {
   if (!subform) {
     return null;
   }
-  if (subform.properties.params.comp) {
+  if (subform.properties?.params?.comp?.params) {
     return subform.properties.params.comp.params;
   }
-  return subform.properties.params;
+  if (subform.params !== undefined) {
+    return subform.params;
+  }
+  return subform.properties?.params ?? null;
 };
 
 export const getModelFromSubform = (subform) => {
   if (!subform) {
     return null;
   }
-  if (subform.properties.params.comp) {
+  if (subform.component !== undefined) {
+    return subform.component;
+  }
+  if (subform.properties?.params?.comp?.component) {
     return subform.properties.params.comp.component;
   }
-  return subform.properties.component;
+  return subform.properties?.component ?? null;
 };
diff --git a/DashAI/front/src/utils/urlUtils.js b/DashAI/front/src/utils/urlUtils.js
new file mode 100644
index 000000000..62292e0d7
--- /dev/null
+++ b/DashAI/front/src/utils/urlUtils.js
@@ -0,0 +1,14 @@
+const API_BASE = process.env.REACT_APP_API_URL || "";
+const API_ORIGIN = (() => {
+  try {
+    return new URL(API_BASE).origin;
+  } catch {
+    return API_BASE;
+  }
+})();
+
+export const normalizeUrl = (url) => {
+  if (!url) return url;
+  if (url.startsWith("http://") || url.startsWith("https://")) return url;
+  return `${API_ORIGIN}${url}`;
+};
diff --git a/DashAI/front/yarn.lock b/DashAI/front/yarn.lock
index 132aafbbe..43a22549d 100644
--- a/DashAI/front/yarn.lock
+++ b/DashAI/front/yarn.lock
@@ -3148,6 +3148,33 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@mui/x-data-grid@npm:^8.27.2":
+  version: 8.28.2
+  resolution: "@mui/x-data-grid@npm:8.28.2"
+  dependencies:
+    "@babel/runtime": ^7.28.4
+    "@mui/utils": ^7.3.5
+    "@mui/x-internals": 8.26.0
+    "@mui/x-virtualizer": 0.3.4
+    clsx: ^2.1.1
+    prop-types: ^15.8.1
+    use-sync-external-store: ^1.6.0
+  peerDependencies:
+    "@emotion/react": ^11.9.0
+    "@emotion/styled": ^11.8.1
+    "@mui/material": ^5.15.14 || ^6.0.0 || ^7.0.0
+    "@mui/system": ^5.15.14 || ^6.0.0 || ^7.0.0
+    react: ^17.0.0 || ^18.0.0 || ^19.0.0
+    react-dom: ^17.0.0 || ^18.0.0 || ^19.0.0
+  peerDependenciesMeta:
+    "@emotion/react":
+      optional: true
+    "@emotion/styled":
+      optional: true
+  checksum: 27e65f40281e94602fc4c3a04df5e6a81c3b82fa7a72453a70b5526d9b986470a56815c4dadcdb77bfc7d80bee4fbcc4bfa515920e22cc443918458ccdc63019
+  languageName: node
+  linkType: hard
+
 "@mui/x-date-pickers@npm:^8.27.2":
   version: 8.27.2
   resolution: "@mui/x-date-pickers@npm:8.27.2"
@@ -3210,6 +3237,20 @@ __metadata:
   languageName: node
   linkType: hard
 
+"@mui/x-virtualizer@npm:0.3.4":
+  version: 0.3.4
+  resolution: "@mui/x-virtualizer@npm:0.3.4"
+  dependencies:
+    "@babel/runtime": ^7.28.4
+    "@mui/utils": ^7.3.5
+    "@mui/x-internals": 8.26.0
+  peerDependencies:
+    react: ^17.0.0 || ^18.0.0 || ^19.0.0
+    react-dom: ^17.0.0 || ^18.0.0 || ^19.0.0
+  checksum: a94ae0d8bdaaec90396b985d7514d44f257b5b94da17ffb1204b516bd689dd99e960aa458279b47711859bdd26c908ed8748684481e7224cae3346f77cd991f0
+  languageName: node
+  linkType: hard
+
 "@nicolo-ribaudo/eslint-scope-5-internals@npm:5.1.1-v1":
   version: 5.1.1-v1
   resolution: "@nicolo-ribaudo/eslint-scope-5-internals@npm:5.1.1-v1"
@@ -6304,11 +6345,11 @@ __metadata:
   linkType: hard
 
 "baseline-browser-mapping@npm:^2.8.9":
-  version: 2.8.14
-  resolution: "baseline-browser-mapping@npm:2.8.14"
+  version: 2.10.27
+  resolution: "baseline-browser-mapping@npm:2.10.27"
   bin:
-    baseline-browser-mapping: dist/cli.js
-  checksum: 422a3c25169ef6ffb89d2fab297f92c72496e0e87bcff6c7af3fbe917a9ee4ca3092ea8bd0ca128d915b2c1b2a0c7921edacdefb701e347d87158f2fa5b2bb1a
+    baseline-browser-mapping: dist/cli.cjs
+  checksum: 7b89ecfcf34132b11a8f07249e21ede07363f02136f61b9750d953da22e376002a0a8f01163dfeb57ec7cee9eb32569f24113659f692b194125d89c64ea8156d
   languageName: node
   linkType: hard
 
@@ -6622,9 +6663,9 @@ __metadata:
   linkType: hard
 
 "caniuse-lite@npm:^1.0.0, caniuse-lite@npm:^1.0.30001702, caniuse-lite@npm:^1.0.30001746":
-  version: 1.0.30001749
-  resolution: "caniuse-lite@npm:1.0.30001749"
-  checksum: 0a2692a7d51e4f4cecd2e8714e1d3d9982479fb59fa2fc8d6a462844bb7f5243ffe0bc94b25a1ff944f63bb2372ff5f6d01ef422729ca3c262975f1b91d78c07
+  version: 1.0.30001791
+  resolution: "caniuse-lite@npm:1.0.30001791"
+  checksum: 9b2f55d51b85abbb270a0d58c28b8e799ccb8fd5ef017179db3d328c8fead4f1738d95a354e75169af31c0424ffb1b691533722522ff280d4aab05d0fe4eddbd
   languageName: node
   linkType: hard
 
@@ -7999,6 +8040,7 @@ __metadata:
     "@mui/material": ^7
     "@mui/system": ^7
     "@mui/utils": ^7
+    "@mui/x-data-grid": ^8.27.2
     "@mui/x-date-pickers": ^8.27.2
     "@stylistic/eslint-plugin-js": ^1.7.2
     "@tanstack/react-table": ^8.21.3
diff --git a/docs/rag/01-overview.md b/docs/rag/01-overview.md
new file mode 100644
index 000000000..a549e0fbc
--- /dev/null
+++ b/docs/rag/01-overview.md
@@ -0,0 +1,96 @@
+# RAG Module Overview
+
+## What is RAG in DashAI
+
+Retrieval-Augmented Generation (RAG) lets users upload documents and have
+conversations with them. The pipeline retrieves relevant document chunks for
+each user query and feeds them to an LLM as context, producing grounded
+answers that cite the source material.
+
+## Pipeline Stages
+
+```
+User query
+     │
+     ▼
+┌──────────────────────────────────────────────────────────┐
+│                  1. Document Loading                      │
+│   Load Document rows from the DB and hydrate in-memory   │
+│   BaseDocument objects (txt, pdf, md, csv, rst, tex).   │
+└──────────────────────────────────────────────────────────┘
+     │
+     ▼
+┌──────────────────────────────────────────────────────────┐
+│                  2. Chunking                              │
+│   Split each document into smaller pieces using one of:  │
+│     • CharacterChunkModel    — fixed-size char windows   │
+│     • TokenChunkModel        — token-aware splitting     │
+│     • RecursiveCharacterChunk — separator-priority split │
+│   Chunks are persisted to the DB and cached by SHA-256   │
+│   signature for reuse across sessions.                   │
+└──────────────────────────────────────────────────────────┘
+     │
+     ▼
+┌──────────────────────────────────────────────────────────┐
+│                  3. Retrieval                             │
+│   Find the top-k most relevant chunks for the query via: │
+│     • Sparse retrievers    — TF-IDF, BM25                │
+│     • Dense retrievers     — embedding-based similarity  │
+│     • Composite retrievers — Sequential, Parallel, MMR   │
+│   (Composite pattern: leaf + composite nodes)            │
+└──────────────────────────────────────────────────────────┘
+     │
+     ▼
+┌──────────────────────────────────────────────────────────┐
+│                  4. Generation                            │
+│   Inject chunks into a prompt template and generate a    │
+│   response via an LLM. Supports multi-language prompts   │
+│   (en/es/pt) and custom templates.                      │
+└──────────────────────────────────────────────────────────┘
+     │
+     ▼
+   Response + cited chunks (returned as JSON)
+```
+
+## Code Location
+
+| Layer     | Path                                          |
+|-----------|-----------------------------------------------|
+| Pipeline  | `DashAI/back/models/RAG/RAG_pipeline.py`      |
+| Factories | `DashAI/back/models/RAG/rag_models_factory.py` |
+| Retrievers| `DashAI/back/models/RAG/retrievers/`          |
+| Chunking  | `DashAI/back/models/RAG/chunking_models/`     |
+| Prompts   | `DashAI/back/models/RAG/prompts/`             |
+| Documents | `DashAI/back/models/RAG/documents/`           |
+| Embeddings| `DashAI/back/models/RAG/embeddings/`          |
+| Jobs      | `DashAI/back/job/rag_job.py`                 |
+| Task      | `DashAI/back/tasks/RAG_task.py`              |
+| Frontend  | `DashAI/front/src/pages/generative/RAGSession/` |
+
+## Quick Architecture
+
+```
+Browser / PyWebView
+  → React (port 3000)
+    → POST /api/v1/generative-session/   (create session)
+    → POST /api/v1/generative-process/   (create process)
+    → POST /api/v1/job/                  (enqueue job)
+      → Huey Job Queue
+        → GenerativeJob.run()
+          → RAGJob.run() (if RAGTask detected)
+            → RAGPipeline.generate()
+              → DocumentLoader → ChunkingModel → Retriever → LLM
+            → RAGTask.process_output() (serialize)
+          → Frontend polls GET /api/v1/jobs/{id}
+```
+
+## Component Types
+
+- **Chunking models** — Character, Token, Recursive character (all implement
+  `BaseChunkingModel`)
+- **Retrievers** — Sparse (TF-IDF, BM25), Dense (via `DenseEmbedding`
+  subclasses), Composite (Sequential, Parallel, MMR Reranker)
+- **Prompts** — Generation prompts (Default, Custom, QA) in en/es/pt +
+  augmentation prompts (Default, Custom) for future query expansion
+- **LLMs** — Any `TextToTextGenerationTaskModel` registered in the component
+  registry (OpenAI-compatible APIs, HuggingFace models)
diff --git a/docs/rag/02-backend-architecture.md b/docs/rag/02-backend-architecture.md
new file mode 100644
index 000000000..7cfb9a31f
--- /dev/null
+++ b/docs/rag/02-backend-architecture.md
@@ -0,0 +1,232 @@
+# Backend Architecture
+
+## Overview
+
+The RAG backend follows a **layered architecture** with clear separation of
+concerns:
+
+```
+RAGJob (orchestration)
+  │
+  ├── RAGPipelineConfig     (validated parameter object)
+  ├── RAGModelsFactory      (component creation)
+  │     ├── PromptFactory
+  │     ├── ChunkingModelFactory
+  │     ├── RetrieverFactory
+  │     └── LLMFactory
+  ├── PipelineRepository    (DB persistence)
+  └── DocumentLoader        (document hydration)
+       │
+       ▼
+  RAGPipeline (orchestrator)
+       │
+       ├── single_interaction()  → retrieve chunks
+       ├── _build_chunk_references() → ChunkReference[]
+       └── generate()            → RAGGenerationOutput
+```
+
+## RAGPipeline (`RAG_pipeline.py`)
+
+Central orchestrator that ties together all sub-systems. Receives its
+dependencies injected as constructor arguments — it does not construct
+factories, repositories, or loaders itself.
+
+### Initialisation Flow
+
+1. **`PipelineRepository.ensure_db_record()`** — Creates or finds the
+   `rag_pipeline` DB record for the session.
+2. **`DocumentLoader.load()`** — Hydrates `BaseDocument` objects from the
+   document table.
+3. **`get_or_create_chunk_set()`** — Computes a SHA-256 hash of
+   `(document_ids + chunking_config)`. If a chunk set with this hash exists,
+   reuses it; otherwise creates a new one.
+4. **`RAGModelsFactory`** — Facade that delegates to four sub-factories:
+   - `PromptFactory` — Creates or looks up the prompt component.
+   - `ChunkingModelFactory` — Creates the chunking model, chunks documents,
+     persists chunks to the DB.
+   - `RetrieverFactory` — Creates the retriever (sparse/dense/composite),
+     computes embeddings if needed, persists.
+   - `LLMFactory` — Creates or looks up the text generation model.
+5. **`PipelineRepository.update_db_record()`** — Patches the pipeline record
+   with the real component IDs.
+
+### Generation Flow
+
+`generate(input_data)`:
+
+1. Extracts the current message from `input_data[-1]` and history from
+   `input_data[:-1]`.
+2. Calls `single_interaction(query)` → `retriever.retrieve(query)` returns
+   top-k `Chunk` objects.
+3. Calls `_build_chunk_references(chunks)` → formats chunks into text +
+   `Dict[str, ChunkReference]` metadata.
+4. Formats the prompt using `prompt_model.format(input, chunks)`.
+5. Calls `llm_model.generate()` with the full prompt + history.
+6. Returns a typed `RAGGenerationOutput(message, chunks)`.
+
+## RAGPipelineConfig
+
+Dataclass that parses and validates the raw kwargs dict once, providing typed
+field access throughout the pipeline. Eliminates magic strings in `__init__`.
+
+- Validates that all required keys exist (`session_id`, `db`,
+  `component_registry`, `env_rag_path`, `documents`, plus four `ModelRef`
+  entries).
+- Validates that each model reference is a `dict` with `component` and `params`
+  keys.
+- Rejects unknown keys via set subtraction.
+
+## RAGModelsFactory (`rag_models_factory.py`)
+
+Abstract Factory (GoF) with four `create_*` methods, each delegating to a
+specialised sub-factory. Shared dependencies (`db`, `registry`, `env_rag_path`)
+are injected once into the factory.
+
+### Sub-Factory Pattern (Lookup-or-Create)
+
+All four sub-factories follow the same pattern:
+
+1. Sort the parameters dictionary by keys.
+2. Query the database for an existing record matching
+   `(class_name, sorted_parameters)`.
+3. If found → re-instantiate the model from the stored parameters.
+4. If not found → create a new DB record, persist it, instantiate the model.
+5. Return a typed `*FactoryResult` dataclass with `db_record_id` and `model`.
+
+**Critical:** The `sorted_params` dict used for DB lookup and save must be
+identical. Factory methods that need to add infrastructure keys do so *after*
+computing `sorted_params`.
+
+### ChunkingModelFactory
+
+Wraps the full chunking lifecycle:
+1. Resolves or creates the DB record.
+2. Instantiates the chunking model (injects `documents` kwarg).
+3. Runs `model.chunk_documents()` — splits text into chunks.
+4. Persists new chunks to the `Chunk` DB table (avoids duplicates via
+   `_persist_chunks`).
+5. Updates in-memory chunk IDs from the persisted records.
+
+### RetrieverFactory
+
+Handles three retriever categories:
+
+- **Sparse** (TF-IDF, BM25) — Creates or loads from `rag_sparse_retriever`
+  table, manages on-disk persistence via `SparsePersistence`.
+- **Dense** (`DenseEmbeddingRetriever`) — Creates or loads from
+  `rag_dense_retriever` table, manages embedding computation and similarity
+  matrix persistence via `DensePersistence`.
+- **Composite** (Sequential, Parallel, MMR) — Recursively creates child
+  retrievers, manages the bridge record in `rag_retriever` and child links in
+  `rag_retriever_child`.
+
+### LLMFactory
+
+Creates or looks up `TextToTextGenerationTaskModel` instances. Persists
+records in `rag_generation_model` table.
+
+## Document Loader (`document_loader.py`)
+
+Resolves document IDs to hydrated `BaseDocument` subclass instances.
+Supported file types and their document classes:
+
+| Extension | Class       |
+|-----------|-------------|
+| `.txt`    | `TxtDocument` |
+| `.pdf`    | `PDFDocument` |
+| `.md`     | `TxtDocument` |
+| `.rst`    | `TxtDocument` |
+| `.tex`    | `TxtDocument` |
+| `.csv`    | `TxtDocument` |
+
+## Retriever Architecture
+
+### Class Hierarchy
+
+```
+RetrieverModel (ABC, DashAI Component)
+  ├── UnitRetriever (ABC, leaf node)
+  │     ├── SparseRetriever (ABC)
+  │     │     ├── TFIDFRetriever
+  │     │     └── BM25Retriever
+  │     └── DenseRetriever (ABC)
+  │           └── DenseEmbeddingRetriever
+  │                 └── HuggingFaceDenseRetriever
+  └── CompositeRetriever (ABC, composite node)
+        ├── SequentialRetriever
+        ├── ParallelRetriever
+        └── MMRRerankerRetriever
+```
+
+- **Unit retrievers** are leaf nodes — they perform actual retrieval against
+  their index.
+- **Composite retrievers** wrap one or more child retrievers and combine their
+  results via a strategy (sequential cascade, parallel merge, MMR re-ranking).
+
+### Dense Retriever Two-Layer Design
+
+The dense retriever separates concerns into two layers:
+
+1. **Embedding layer** (`DenseEmbedding` subclasses) — Handle model loading,
+   tokenization, pooling, and vector generation. These are DashAI Components
+   registered in the component registry. Available: `BERTEmbedding`,
+   `DistilBERTEmbedding`, `E5Embedding`, `GemmaEmbedding`,
+   `HuggingFaceEmbedding`, `InstructorEmbedding`, `LaBSEmbedding`,
+   `OpenAIEmbedding`, `RoBERTaEmbedding`, `SentenceTransformerEmbedding`.
+2. **Retriever layer** (shared `DenseEmbeddingRetriever` class) — Accepts any
+   `DenseEmbedding` via a `component_field` schema parameter. The embedding
+   family is selected at configuration time.
+
+### Retriever Persistence
+
+All retriever SQL is isolated in `RetrieverRepository`
+(`retriever_repository.py`). It owns every query, INSERT, and DB-model
+construction related to retrievers.
+
+The `rag_retriever` table acts as a **bridge** — every retriever (unit,
+composite, dense, sparse) gets one canonical identity record here. Sub-tables
+(`rag_sparse_retriever`, `rag_dense_retriever`) hold type-specific details.
+Composite child links are stored in `rag_retriever_child`.
+
+## Chunk Set Caching (`chunk_set_utils.py`)
+
+`get_or_create_chunk_set()` computes a deterministic SHA-256 hash of:
+- Sorted document IDs
+- Sorted chunking configuration (e.g., chunking model component + params)
+
+If a chunk set with this signature exists, it is reused. This avoids
+re-chunking the same documents with the same parameters.
+
+**Note:** Uses SELECT-then-INSERT without a lock. Safe for single-user
+operation but could create duplicate chunk sets under concurrent requests.
+
+## Job Layer (`rag_job.py`, `generative_job.py`)
+
+- **`GenerativeJob`** is the generic job for text generation. At runtime, it
+  detects whether the session uses `RAGTask` and delegates to `RAGJob`.
+- **`RAGJob`** builds the `RAGPipeline` with all dependencies and runs
+  generate/save/output in a background Huey task. RAGJob duplicates some
+  status-update methods from GenerativeJob by design — see
+  [known limitations](./05-known-limitations.md#code-duplication) for the
+  rationale.
+
+## Task Layer (`RAG_task.py`)
+
+`RAGTask` extends `BaseGenerativeTask` and:
+- Prepares input by folding chat history into the message list.
+- Processes output by serializing the response and chunk references into
+  DB-suitable format.
+- Declares `USE_HISTORY = True` to enable history folding in the job layer.
+
+## Typed Return Values
+
+The pipeline uses typed dataclasses instead of raw dicts or lists:
+
+| Dataclass              | Purpose                                      |
+|------------------------|----------------------------------------------|
+| `RAGPipelineConfig`    | Validated pipeline parameters                |
+| `ChunkReference`       | A single retrieved chunk with document info  |
+| `RAGGenerationOutput`  | Typed pipeline output (message + chunks)     |
+| `*FactoryResult`       | Each factory returns a typed result          |
+| `SparsePersistence`    | On-disk reference for sparse retrievers      |
+| `DensePersistence`     | Embedding matrix references for dense retrievers |
diff --git a/docs/rag/03-frontend-architecture.md b/docs/rag/03-frontend-architecture.md
new file mode 100644
index 000000000..0458e6b52
--- /dev/null
+++ b/docs/rag/03-frontend-architecture.md
@@ -0,0 +1,67 @@
+# Frontend Architecture
+
+## Routes
+
+| Path                              | Component              | Purpose                           |
+|-----------------------------------|------------------------|-----------------------------------|
+| `/app/generative`                 | `SessionRouter`        | Routes to RAG or non-RAG view     |
+| `/app/generative/sessions/:id`    | `RAGSessionPage`       | RAG session detail + chat         |
+
+## Main RAG Page Flow
+
+All files under `pages/generative/RAGSession/`:
+
+1. **`RAGSessionSetup.jsx`** — Session creation form with accordion sections:
+   - Document selection, chunking config, retriever config (3-preset card
+     system), prompt template selection, generator (LLM) config.
+
+2. **`RAGSessionPage.jsx`** — 3-panel orchestrator:
+   - Left: session list + document manager
+   - Center: setup form / summary view / chat view
+   - Right: info bar (educational) / params panel
+
+3. **`GenerativeChat`** — Active chat view (shared with non-RAG sessions).
+
+### Per-Stage Config Sections
+
+Each pipeline stage has a section component in `sections/`:
+- `ChunkingSection.jsx`
+- `RetrieverSection.jsx`
+- `GeneratorSection.jsx`
+- `PromptSection.jsx`
+
+### Advanced Configuration Modals
+
+In `advanced/`:
+- `CompositeRetrieverBuilder.jsx`
+- `ChunkingConfigurationStep.jsx`
+- `RetrieverConfigurationStep.jsx`
+
+### Supporting Components
+
+In `components/generative/RAG/`:
+- `RAGInfoBar.jsx` — Educational right panel content
+- `RAGSessionSummary.jsx` — Session overview
+- `RAGParamsPanel.jsx` — Parameter editing
+
+## Key Features
+
+- **Retriever Presets** — 3-card system: Keyword (BM25), Semantic (Dense),
+  Hybrid (Sequential BM25 + Dense).
+- **Context Window Validation** — Validates that
+  `chunk_size * top_k + prompt_tokens <= context_window`.
+- **Multi-Language Prompts** — Templates in en/es/pt, selected via dropdown.
+- **Translation Keys** — All RAG translations use the `generative:rag.*`
+  namespace.
+
+## API Layer
+
+All RAG API calls use standard DashAI endpoints:
+
+| Endpoint                          | Purpose           |
+|-----------------------------------|-------------------|
+| `/api/v1/generative-session/`     | Session CRUD      |
+| `/api/v1/generative-process/`     | Process CRUD      |
+| `/api/v1/job/`                    | Job dispatch      |
+| `/api/v1/document/`               | Document management |
+| `/api/v1/prompt/`                 | Prompt management |
diff --git a/docs/rag/04-execution-flow.md b/docs/rag/04-execution-flow.md
new file mode 100644
index 000000000..52bd876cb
--- /dev/null
+++ b/docs/rag/04-execution-flow.md
@@ -0,0 +1,153 @@
+# End-to-End Execution Flow
+
+## Step-by-Step
+
+### 1. Session Configuration
+
+The user fills the `RAGSessionSetup` form:
+- Selects documents from the repository.
+- Configures the chunking model (type, chunk size, overlap).
+- Chooses a retriever (preset or custom composite).
+- Selects a prompt template (language, optional custom template).
+- Picks an LLM (model name, parameters like temperature, context window).
+
+### 2. Session Creation
+
+Frontend calls `POST /api/v1/generative-session/` with the full parameter
+payload. The endpoint validates parameters against `RAGPipeline.SCHEMA` and
+persists a `GenerativeSession` record in the database. The session stores:
+- `task_name` — Set to `"RAGTask"` for RAG sessions.
+- `model_name` — Set to `"RAGPipeline"`.
+- `parameters` — The full configuration dict (documents, chunking, retriever,
+  prompt, generation model).
+
+### 3. Process Creation
+
+When the user sends a message, the frontend calls
+`POST /api/v1/generative-process/` with the input text. This creates a
+`GenerativeProcess` record linked to the session, with `status="PENDING"`.
+
+### 4. Job Dispatch
+
+The frontend calls `POST /api/v1/job/` with `job_type="GenerativeJob"` and the
+`generative_process_id`. The job is enqueued in the Huey async task queue.
+
+### 5. Job Execution
+
+`GenerativeJob.run()` performs an initial DB lookup:
+```
+Query session → check task_name → if issubclass(RAGTask):
+    delegate to RAGJob.run()
+    return
+```
+
+If the task is **not** a RAG task, `GenerativeJob` proceeds with the standard
+generation flow.
+
+### 6. Pipeline Construction (in RAGJob)
+
+```
+RAGJob.run():
+  │
+  ├── 1. Query GenerativeProcess + GenerativeSession from DB
+  │
+  ├── 2. Build RAGPipelineConfig.from_kwargs()
+  │       Uses session parameters + injected DB + registry + env path
+  │
+  ├── 3. Create RAGModelsFactory(db, registry, rag_path)
+  │
+  ├── 4. Create PipelineRepository(db)
+  │
+  ├── 5. Create DocumentLoader(db)
+  │
+  └── 6. Assemble RAGPipeline(config, models, repo, doc_loader)
+```
+
+### 7. Pipeline Initialisation
+
+During `RAGPipeline.__init__()`:
+
+```
+  │
+  ├── a. PipelineRepository.ensure_db_record(session_id)
+  │       Creates or finds the rag_pipeline table row.
+  │
+  ├── b. DocumentLoader.load(document_ids)
+  │       Fetches documents from DB, hydrates BaseDocument objects.
+  │
+  ├── c. get_or_create_chunk_set(db, doc_ids, chunking_config)
+  │       SHA-256 signature → reuse or create new chunk set.
+  │
+  ├── d. ModelsFactory.create_prompt(component, params)
+  │       Lookup-or-create prompt record in rag_prompt table.
+  │
+  ├── e. ModelsFactory.create_chunking_model(docs, chunk_set, component, params)
+  │       Creates chunker → chunks docs → persists chunks → returns model.
+  │
+  ├── f. ModelsFactory.create_retriever(pipeline_id, chunks, chunk_set, component, params)
+  │       Creates retriever → computes/trains → persists → returns model.
+  │
+  ├── g. ModelsFactory.create_llm(component, params)
+  │       Lookup-or-create LLM record in rag_generation_model table.
+  │
+  └── h. PipelineRepository.update_db_record(...)
+        Patches the pipeline record with real component FK IDs.
+```
+
+### 8. History Preparation
+
+The job queries all previous `GenerativeProcess` records for the session with
+`status="FINISHED"` and builds a list of `(input, output)` tuples. The
+`RAGTask.prepare_for_task()` method folds this history into a message list:
+
+```
+[
+  {"role": "user", "content": "Previous input"},
+  {"role": "assistant", "content": "Previous output"},
+  {"role": "user", "content": "Current input"}
+]
+```
+
+### 9. Generation
+
+`RAGPipeline.generate(input_data)`:
+
+```
+  │
+  ├── a. Extract input message + history from input_data
+  │
+  ├── b. retriever.retrieve(query)
+  │       Returns top-k Chunk objects relevant to the query.
+  │
+  ├── c. _build_chunk_references(chunks)
+  │       Formats chunks into text block + ChunkReference dict.
+  │
+  ├── d. prompt_model.format(input=query, chunks=chunk_text)
+  │       Renders the prompt template with query + chunk context.
+  │
+  ├── e. llm_model.generate(history + prompt)
+  │       LLM generates a response text.
+  │
+  └── f. Return RAGGenerationOutput(message, chunks)
+```
+
+### 10. Output Processing
+
+`RAGTask.process_output(output)`:
+
+1. Unpacks `output.message` (LLM response text) and `output.chunks` (dict of
+   `ChunkReference` objects with document metadata).
+2. Returns a list of `(data, type)` tuples:
+   - `(response_text, "str")`
+   - `(json_chunk_references, "Dict")`
+
+### 11. Persistence
+
+`RAGJob` creates `ProcessData` records for each output tuple, links them to the
+`GenerativeProcess`, and sets the process status to `"FINISHED"`.
+
+### 12. Frontend Polling
+
+The frontend polls `GET /api/v1/jobs/{job_id}` for job status. When the job
+completes, it reads the process output and renders the chat response with
+cited chunk references.
diff --git a/docs/rag/05-known-limitations.md b/docs/rag/05-known-limitations.md
new file mode 100644
index 000000000..8f4c7a36b
--- /dev/null
+++ b/docs/rag/05-known-limitations.md
@@ -0,0 +1,76 @@
+# Known Limitations and Operational Notes
+
+## Performance
+
+- **Chunk similarity matrices are held in RAM.** Works for low hundreds of
+  documents but will not scale to millions. No out-of-core or approximate
+  indexing.
+
+- **No streaming.** The frontend waits for the full LLM response before
+  displaying it. Streaming support is not implemented.
+
+- **No FAISS or HNSW indexing.** Dense retrieval performs O(n*dim) brute-force
+  similarity search per query. Large document collections will be slow.
+
+- **DB session held open during LLM inference.** The same database transaction
+  remains open while the LLM generates a response, which risks connection
+  timeout on slow LLM calls.
+
+## Concurrency
+
+- **`get_or_create_chunk_set()`** uses a SELECT-then-INSERT pattern without a
+  lock or upsert. Safe for single-user usage but could create duplicate chunk
+  sets under concurrent requests.
+
+- **Huey consumer runs in-process.** In dev mode it spawns as a subprocess; in
+  PyInstaller bundles it runs as a daemon thread. Both models limit
+  parallelism.
+
+## Unused Components
+
+- **`DefaultAugmentationPrompt`** is registered in the component registry but
+  **commented out** in `get_initial_components()`. The augmentation prompt
+  family (query expansion / HyDE-style) is not yet wired into the pipeline.
+
+- **`CustomAugmentationPrompt`** is registered but unused by the pipeline.
+
+## Type System Caveats
+
+- **Prompt `chunks` type:** The pipeline passes a formatted `str` to
+  `prompt.format()`, but the abstract `RAGGenerationPrompt.format()` and some
+  subclasses declare `chunks: List[str]`. Only `str` works at runtime. The
+  abstract base signature should match the concrete usage.
+
+- **`score_chunks()` contract:** The abstract method declares return type
+  `List[float]`, but all implementations return `List[Tuple[int, float]]`
+  (chunk ID, distance). Consumers destructure tuples, so the abstract
+  signature is wrong.
+
+## Maintenance
+
+- **After RAG model schema changes**, delete `sqlite.db` and the
+  `~/.DashAI/rag/` directory to rebuild the database and chunk cache from
+  scratch.
+
+- **Tokenizer downloads on first use.** `TokenChunkModel` downloads the
+  tokenizer from HuggingFace Hub lazily (on first `chunk_text()` call), but
+  this still requires network access. Pre-download tokenizers for offline use.
+
+- **Embedding models downloaded on first use.** Each `DenseEmbedding` subclass
+  downloads its model from HuggingFace Hub on first instantiation. Pre-download
+  for offline environments.
+
+## Code Duplication
+
+- **`GenerativeJob` and `RAGJob` share identical `set_status_as_delivered()`,
+  `set_status_as_error()`, and `get_job_name()` methods.** The code is
+  duplicated rather than extracted into a shared helper. This is
+  **intentional** — `GenerativeJob` lives in the main `develop` branch while
+  `RAGJob` lives in the `RAG` branch; extracting shared code would require
+  either a shared module that crosses branch boundaries or continuous
+  backporting. The duplication is accepted technical debt to keep each job
+  self-contained and avoid cross-branch coupling.
+
+- **`GenerativeJob` must not be modified.** Any improvements to the shared
+  status-update logic should be applied to `RAGJob` independently, or
+  coordinated when the branches are merged.
diff --git a/docs/rag/06-future-work.md b/docs/rag/06-future-work.md
new file mode 100644
index 000000000..6afd4e074
--- /dev/null
+++ b/docs/rag/06-future-work.md
@@ -0,0 +1,55 @@
+# Future Work
+
+This document outlines planned improvements and known gaps in the current RAG implementation. These are not limitations per se, but rather opportunities for expansion.
+
+## Retrieval Paradigms
+
+The current pipeline follows a straightforward sequential flow: chunk → retrieve → generate. It does not support more complex retrieval strategies such as:
+
+- **Iterative retrieval** — alternating between retrieval and generation steps, using intermediate LLM outputs to refine subsequent queries.
+- **Generate-then-retrieve** — generating a preliminary answer and using it to retrieve supporting evidence.
+- **Corrective RAG** — evaluating retrieved chunks for relevance and triggering re-retrieval when no chunk is relevant.
+- **HyDE (Hypothetical Document Embeddings)** — generating a hypothetical ideal document from the query and retrieving by its embedding rather than the raw query.
+
+## Query Transformation
+
+The pipeline currently passes the user's raw query directly to the retriever. No query transformation phase exists. Future work could add:
+
+- **Query expansion** — expanding short queries with synonyms or related terms.
+- **Query decomposition** — splitting a complex question into sub-queries routed to different retrievers.
+- **Rewriting / compression** — rephrasing ambiguous queries or compressing long chat histories into a single standalone query.
+
+## PDF Parsing
+
+The current `PDFDocument` class supports two backends: **textract** (default) and **PyPDF2** (via the `parser` parameter). Future improvements include:
+
+- Implementing OCR for PDFs.
+- Making the PDF parser configurable per-session or per-document.
+- Adding support for more parsers such as pdfplumber or marker (for complex layouts).
+- Implementing an LLM-based preprocessing stage for structured extraction from PDFs (tables, headers, figures).
+
+## Documents library
+
+Currently, the `Documents` library is a simple table of documents, with a column for metadata that is not used, future work could include:
+
+- Metadata-based filtering: allowing users to filter documents by metadata fields (e.g., author, date, tags).
+- Document versioning: supporting multiple versions of the same document and allowing retrieval from a specific version.
+- Document similarity search: enabling users to find documents similar to a given document based on content or metadata.
+- Import/export functionality: allowing users to import/export documents and their metadata in various formats (e.g., JSON, CSV, XML), including support for bibtex and Mendely references.
+
+## Tree-Based Retrieval
+
+Only flat (sparse + dense) and composite (sequential / parallel / MMR) retrievers are implemented. There is no tree-based indexing or retrieval.
+
+## Document Subset Filtering at Message Level
+
+When chatting, the user might want a given response to use only a specific subset of their uploaded documents (e.g., "use only the documents I referenced in my previous message"). Currently the retriever always searches across all documents assigned to the session. A document-level filter per message would enable more fine-grained control.
+
+## Multi-Modal Document Support
+
+Currently only text-based documents are supported (txt, pdf, md, rst, tex, csv). A major challenge is extending the pipeline to handle other modalities:
+
+- **Graph documents** — knowledge graphs, property graphs, RDF data.
+- **Datasets** — structured tabular data as found in the rest of DashAI (e.g., CSV datasets used for ML training).
+- **Images** — both extracting text from images (OCR) and, more ambitiously, retrieving chunks based on visual similarity.
+- **Audio / video** — transcription and retrieval from spoken content.
diff --git a/docs/rag/README.md b/docs/rag/README.md
new file mode 100644
index 000000000..894a0103e
--- /dev/null
+++ b/docs/rag/README.md
@@ -0,0 +1,24 @@
+# RAG Module Documentation
+
+DashAI includes a **Retrieval-Augmented Generation (RAG)** module for chatting
+with your documents. The pipeline has 4 stages: Document Loading, Chunking,
+Retrieval, and Generation.
+
+## Documentation Index
+
+- **[01-overview.md](./01-overview.md)** — High-level overview, architecture
+  diagram, pipeline stages, component types, and code location.
+- **[02-backend-architecture.md](./02-backend-architecture.md)** — Detailed
+  backend architecture: pipeline orchestration, factory pattern, retriever
+  hierarchy, persistence, typed return values.
+- **[03-frontend-architecture.md](./03-frontend-architecture.md)** — Frontend
+  routes, component tree, API endpoints, and key features.
+- **[04-execution-flow.md](./04-execution-flow.md)** — End-to-end execution
+  flow: session creation → job dispatch → pipeline construction → generation
+  → frontend polling.
+- **[05-known-limitations.md](./05-known-limitations.md)** — Current
+  constraints, performance notes, concurrency caveats, and maintenance
+  guidance.
+- **[06-future-work.md](./06-future-work.md)** — Planned improvements:
+  retrieval paradigms, query transformation, PDF parsing, tree-based
+  retrieval, message-level document filtering, and multi-modal support.
diff --git a/requirements-plugins.txt b/requirements-plugins.txt
index 356960f3a..b7435b5b6 100644
--- a/requirements-plugins.txt
+++ b/requirements-plugins.txt
@@ -1,2 +1,2 @@
-torch==1.13.0 #+cu116
+torch#==1.13.0 #+cu116
 --find-links https://download.pytorch.org/whl/torch_stable.html
diff --git a/requirements.txt b/requirements.txt
index 165d4afe6..25a3e6f41 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -39,12 +39,23 @@ wordcloud
 opencv-python
 protobuf
 imblearn
+llama-cpp-python
+llvmlite
 numba
 llvmlite
 huey
 ijson
 greenery==3.2
 xlrd
+
+
+fasttext-wheel
+pypdf2
+textract
+openai
+sentence-transformers
+InstructorEmbedding
+
 filetype
 torchmetrics
 pywebview
diff --git a/run_pipeline_benchmark.py b/run_pipeline_benchmark.py
new file mode 100644
index 000000000..2f4a61d0c
--- /dev/null
+++ b/run_pipeline_benchmark.py
@@ -0,0 +1,27 @@
+"""Thin wrapper that delegates to RAG_benchmark/run_pipeline_benchmark.py.
+
+The actual implementation has moved to ``RAG_benchmark/run_pipeline_benchmark.py``.
+Please use that going forward, or run:
+    python -m RAG_benchmark.benchmarks.cli [OPTIONS] COMMAND [ARGS]
+"""
+
+from __future__ import annotations
+
+import sys
+from pathlib import Path
+
+_runner = Path(__file__).parent / "RAG_benchmark" / "run_pipeline_benchmark.py"
+if not _runner.exists():
+    print(
+        "Error: RAG_benchmark/run_pipeline_benchmark.py not found.\n"
+        "Please run: python -m RAG_benchmark.benchmarks.cli --help",
+        file=sys.stderr,
+    )
+    sys.exit(1)
+
+# Forward to the moved script
+sys.path.insert(0, str(_runner.parent.parent))
+from RAG_benchmark.run_pipeline_benchmark import main
+
+if __name__ == "__main__":
+    main()
diff --git a/tests/back/RAG/__init__.py b/tests/back/RAG/__init__.py
new file mode 100644
index 000000000..e69de29bb
diff --git a/tests/back/RAG/conftest.py b/tests/back/RAG/conftest.py
new file mode 100644
index 000000000..667e90502
--- /dev/null
+++ b/tests/back/RAG/conftest.py
@@ -0,0 +1,49 @@
+"""Fixtures for RAG API integration tests.
+
+Shares the same client fixture as the api/ tests so that generative
+session / prompt endpoints are available.
+"""
+
+import shutil
+import time
+from pathlib import Path
+
+import pytest
+from fastapi.testclient import TestClient
+
+from DashAI.back.app import create_app
+from DashAI.back.dependencies.job_queues.huey_job_queue import HueyJobQueue
+from DashAI.back.job.dataset_job import DatasetJob
+
+
+def remove_dir_with_retry(directory, max_attempts=5, sleep_seconds=1):
+    for attempt in range(max_attempts):
+        try:
+            shutil.rmtree(directory)
+            break
+        except PermissionError as e:
+            print(f"Attempt {attempt + 1} failed with error: {e}")
+            time.sleep(sleep_seconds)
+    else:
+        print(f"Failed to remove directory after {max_attempts} attempts.")
+
+
+@pytest.fixture(scope="module")
+def client(test_path: Path):
+    app = create_app(
+        local_path=test_path,
+        logging_level="ERROR",
+    )
+
+    job_queue = app.container._services.get("job_queue")
+    if job_queue and isinstance(job_queue, HueyJobQueue):
+        job_queue.set_test_mode(True)
+
+    test_client = TestClient(app)
+    yield test_client
+
+    if job_queue and isinstance(job_queue, HueyJobQueue):
+        job_queue.set_test_mode(False)
+
+    app.container._services["engine"].dispose()
+    remove_dir_with_retry(app.container._services["config"]["LOCAL_PATH"])
diff --git a/tests/back/RAG/test_rag_component_api_configs.py b/tests/back/RAG/test_rag_component_api_configs.py
new file mode 100644
index 000000000..d18751324
--- /dev/null
+++ b/tests/back/RAG/test_rag_component_api_configs.py
@@ -0,0 +1,796 @@
+"""Integration tests validating every RAG component configuration dimension.
+
+Each test creates a RAG session through the API with a specific parameter
+combination and verifies:
+
+- Session creation returns 201 Created
+- All parameters can be retrieved via GET and match the expected configuration
+
+All models are ≤ 8B parameters. Larger variants (Llama-2-13B, Qwen-14B,
+Mixtral, Harrier-27B, Qwen3-Embedding-8B, Meta-Llama-3.1-8B) are excluded.
+
+Note: Session creation validates only the top-level RAGPipeline schema
+({component, params} structure). Sub-component schemas are validated at
+pipeline runtime, not during session creation.
+"""
+
+import pytest
+from fastapi.testclient import TestClient
+
+from DashAI.back.dependencies.database.models import Document
+
+
+def _create_test_document(client: TestClient, suffix: str = "") -> int:
+    """Create a minimal test document in the DB and return its ID."""
+    session_factory = client.app.container["session_factory"]
+    with session_factory() as db:
+        doc = Document(
+            file_name=f"test_doc{suffix}.txt",
+            file_type="txt",
+            file_path=f"/tmp/test_doc{suffix}.txt",
+            file_hash=f"test_hash_123_{suffix}" if suffix else "test_hash_123",
+        )
+        db.add(doc)
+        db.commit()
+        db.refresh(doc)
+        return doc.id
+
+
+@pytest.fixture(scope="module")
+def test_doc_id(client: TestClient) -> int:
+    """Module-scoped test document ID shared across all component config tests."""
+    return _create_test_document(client, suffix="_component_configs")
+
+
+# ---------------------------------------------------------------------------
+# helpers
+# ---------------------------------------------------------------------------
+
+ST_MINI_LM = "sentence-transformers/all-MiniLM-L6-v2"
+LLAMA_1B = "bartowski/Llama-3.2-1B-Instruct-GGUF"
+LLAMA_3B = "bartowski/Llama-3.2-3B-Instruct-GGUF"
+
+
+def _base_params(test_doc_id: int) -> dict:
+    """Return the minimal default RAG session payload."""
+    return {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "CharacterChunkModel",
+                "params": {"chunk_size": 400, "chunk_overlap": 40},
+            },
+            "retriever_model": {
+                "component": "BM25Retriever",
+                "params": {
+                    "BM25Vectorizer": {
+                        "component": "BM25VectorizerModel",
+                        "params": {},
+                    },
+                    "k1": 1.5,
+                    "b": 0.75,
+                    "delta": 0.0,
+                    "similarity_function": "cosine",
+                    "top_k": 5,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": LLAMA_1B,
+                    "max_tokens": 100,
+                    "temperature": 0.7,
+                    "frequency_penalty": 0.1,
+                    "context_window": 512,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Config Test",
+        "description": None,
+    }
+
+
+def _dense_st_retriever(
+    similarity_metric: str = "cosine",
+    top_k: int = 10,
+    model_name: str = ST_MINI_LM,
+) -> dict:
+    """Build a DenseEmbeddingRetriever + SentenceTransformerEmbedding config."""
+    return {
+        "component": "DenseEmbeddingRetriever",
+        "params": {
+            "embedding_model": {
+                "component": "SentenceTransformerEmbedding",
+                "params": {
+                    "model_name": model_name,
+                    "normalize": True,
+                    "device": "CPU",
+                },
+            },
+            "similarity_metric": similarity_metric,
+            "top_k": top_k,
+        },
+    }
+
+
+def _dense_openai_retriever(top_k: int = 10) -> dict:
+    """Build a DenseEmbeddingRetriever + OpenAIEmbedding config."""
+    return {
+        "component": "DenseEmbeddingRetriever",
+        "params": {
+            "embedding_model": {
+                "component": "OpenAIEmbedding",
+                "params": {
+                    "model_name": "text-embedding-3-small",
+                    "api_key": "test-key",
+                },
+            },
+            "similarity_metric": "cosine",
+            "top_k": top_k,
+        },
+    }
+
+
+def _post_and_get(client: TestClient, params: dict):
+    """POST a session, assert 201 + metadata, then GET and return stored JSON.
+
+    NOTE: The GET response returns the raw SQLAlchemy model (serialized by
+    FastAPI), which may include extra DB columns (e.g. ``user_id``) not
+    present in the POST response.  This is intentional — the test verifies
+    that the *stored* parameters match what was sent.
+    """
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == params["name"]
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    return get_resp.json()
+
+
+# ===================================================================
+# Category 1: Encoding Models  (6 tests)
+# ===================================================================
+
+
+class TestEncodingModels:
+    """6 configurations: sparse and dense encoding models with hyperparameter variations."""
+
+    def test_encoding_bm25_default(self, client: TestClient, test_doc_id: int):
+        """BM25Retriever with all default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc BM25 Default"
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "BM25Retriever"
+        assert ret["params"]["k1"] == 1.5
+        assert ret["params"]["b"] == 0.75
+        assert ret["params"]["delta"] == 0.0
+        assert ret["params"]["similarity_function"] == "cosine"
+        assert ret["params"]["top_k"] == 5
+
+    def test_encoding_bm25_custom_hyperparams(self, client: TestClient, test_doc_id: int):
+        """BM25Retriever with custom hyperparams k1=2.0, b=0.5, delta=0.5, euclidean, top_k=7."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc BM25 Custom"
+        params["parameters"]["retriever_model"] = {
+            "component": "BM25Retriever",
+            "params": {
+                "BM25Vectorizer": {
+                    "component": "BM25VectorizerModel",
+                    "params": {},
+                },
+                "k1": 2.0,
+                "b": 0.5,
+                "delta": 0.5,
+                "similarity_function": "euclidean",
+                "top_k": 7,
+            },
+        }
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "BM25Retriever"
+        assert ret["params"]["k1"] == 2.0
+        assert ret["params"]["b"] == 0.5
+        assert ret["params"]["delta"] == 0.5
+        assert ret["params"]["similarity_function"] == "euclidean"
+        assert ret["params"]["top_k"] == 7
+
+    def test_encoding_tfidf_default(self, client: TestClient, test_doc_id: int):
+        """TFIDFRetriever with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc TFIDF Default"
+        params["parameters"]["retriever_model"] = {
+            "component": "TFIDFRetriever",
+            "params": {
+                "TFIDFVectorizer": {
+                    "component": "TFIDFVectorizerModel",
+                    "params": {},
+                },
+                "similarity_function": "cosine",
+                "top_k": 5,
+            },
+        }
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "TFIDFRetriever"
+        assert ret["params"]["similarity_function"] == "cosine"
+        assert ret["params"]["top_k"] == 5
+
+    def test_encoding_tfidf_custom(self, client: TestClient, test_doc_id: int):
+        """TFIDFRetriever with similarity_function=manhattan, top_k=15."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc TFIDF Custom"
+        params["parameters"]["retriever_model"] = {
+            "component": "TFIDFRetriever",
+            "params": {
+                "TFIDFVectorizer": {
+                    "component": "TFIDFVectorizerModel",
+                    "params": {},
+                },
+                "similarity_function": "manhattan",
+                "top_k": 15,
+            },
+        }
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "TFIDFRetriever"
+        assert ret["params"]["similarity_function"] == "manhattan"
+        assert ret["params"]["top_k"] == 15
+
+    def test_encoding_dense_sentence_transformer(
+        self, client: TestClient, test_doc_id: int
+    ):
+        """DenseEmbeddingRetriever + SentenceTransformerEmbedding (all-MiniLM-L6-v2)."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc Dense ST"
+        params["parameters"]["retriever_model"] = _dense_st_retriever(top_k=10)
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "DenseEmbeddingRetriever"
+        embed = ret["params"]["embedding_model"]
+        assert embed["component"] == "SentenceTransformerEmbedding"
+        assert embed["params"]["model_name"] == ST_MINI_LM
+        assert embed["params"]["normalize"] is True
+        assert embed["params"]["device"] == "CPU"
+        assert ret["params"]["similarity_metric"] == "cosine"
+        assert ret["params"]["top_k"] == 10
+
+    def test_encoding_dense_openai(self, client: TestClient, test_doc_id: int):
+        """DenseEmbeddingRetriever + OpenAIEmbedding (text-embedding-3-small)."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Enc Dense OpenAI"
+        params["parameters"]["retriever_model"] = _dense_openai_retriever(top_k=10)
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "DenseEmbeddingRetriever"
+        embed = ret["params"]["embedding_model"]
+        assert embed["component"] == "OpenAIEmbedding"
+        assert embed["params"]["model_name"] == "text-embedding-3-small"
+        assert embed["params"]["api_key"] == "test-key"
+        assert ret["params"]["similarity_metric"] == "cosine"
+        assert ret["params"]["top_k"] == 10
+
+
+# ===================================================================
+# Category 2: Ranking Functions  (4 tests)
+# ===================================================================
+
+
+class TestRankingFunctions:
+    """4 ranking function configurations evaluated with different encoding models."""
+
+    def test_ranking_dense_cosine(self, client: TestClient, test_doc_id: int):
+        """DenseEmbeddingRetriever with cosine similarity."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Rank Cosine"
+        params["parameters"]["retriever_model"] = _dense_st_retriever(
+            similarity_metric="cosine", top_k=10
+        )
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["params"]["similarity_metric"] == "cosine"
+        assert ret["params"]["top_k"] == 10
+
+    def test_ranking_dense_euclidean(self, client: TestClient, test_doc_id: int):
+        """DenseEmbeddingRetriever with euclidean similarity."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Rank Euclidean"
+        params["parameters"]["retriever_model"] = _dense_st_retriever(
+            similarity_metric="euclidean", top_k=10
+        )
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["params"]["similarity_metric"] == "euclidean"
+        assert ret["params"]["top_k"] == 10
+
+    def test_ranking_sparse_manhattan(self, client: TestClient, test_doc_id: int):
+        """BM25Retriever with manhattan similarity."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Rank Manhattan"
+        params["parameters"]["retriever_model"]["params"]["similarity_function"] = "manhattan"
+        params["parameters"]["retriever_model"]["params"]["top_k"] = 10
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "BM25Retriever"  # guard: depends on _base_params
+        assert ret["params"]["similarity_function"] == "manhattan"
+        assert ret["params"]["top_k"] == 10
+
+    def test_ranking_mmr_reranker(self, client: TestClient, test_doc_id: int):
+        """MMRRerankerRetriever (lambda=0.7, factor=4) wrapping DenseEmbeddingRetriever."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Rank MMR"
+        params["parameters"]["retriever_model"] = {
+            "component": "MMRRerankerRetriever",
+            "params": {
+                "mmr_lambda": 0.7,
+                "retrieval_factor": 4,
+                "top_k": 10,
+                "children": [
+                    _dense_st_retriever(similarity_metric="cosine", top_k=40)
+                ],
+            },
+        }
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["component"] == "MMRRerankerRetriever"
+        assert ret["params"]["mmr_lambda"] == 0.7
+        assert ret["params"]["retrieval_factor"] == 4
+        assert ret["params"]["top_k"] == 10
+        children = ret["params"]["children"]
+        assert len(children) == 1
+        assert children[0]["component"] == "DenseEmbeddingRetriever"
+        assert children[0]["params"]["embedding_model"]["component"] == "SentenceTransformerEmbedding"
+        assert children[0]["params"]["top_k"] == 40
+
+
+# ===================================================================
+# Category 3: Top K  (6 tests)
+# ===================================================================
+
+
+class TestTopK:
+    """6 top-K configurations using different encodings and ranking functions."""
+
+    def test_topk_bm25_3(self, client: TestClient, test_doc_id: int):
+        """BM25Retriever with top_k=3."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK BM25 3"
+        params["parameters"]["retriever_model"]["params"]["top_k"] = 3
+        stored = _post_and_get(client, params)
+        assert stored["parameters"]["retriever_model"]["params"]["top_k"] == 3
+
+    def test_topk_tfidf_5(self, client: TestClient, test_doc_id: int):
+        """TFIDFRetriever with top_k=5."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK TFIDF 5"
+        params["parameters"]["retriever_model"] = {
+            "component": "TFIDFRetriever",
+            "params": {
+                "TFIDFVectorizer": {
+                    "component": "TFIDFVectorizerModel",
+                    "params": {},
+                },
+                "similarity_function": "cosine",
+                "top_k": 5,
+            },
+        }
+        stored = _post_and_get(client, params)
+        assert stored["parameters"]["retriever_model"]["params"]["top_k"] == 5
+
+    def test_topk_dense_st_10(self, client: TestClient, test_doc_id: int):
+        """DenseEmbeddingRetriever + SentenceTransformer with top_k=10."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK Dense ST 10"
+        params["parameters"]["retriever_model"] = _dense_st_retriever(top_k=10)
+        stored = _post_and_get(client, params)
+        assert stored["parameters"]["retriever_model"]["params"]["top_k"] == 10
+
+    def test_topk_dense_openai_15(self, client: TestClient, test_doc_id: int):
+        """DenseEmbeddingRetriever + OpenAI with top_k=15."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK Dense OpenAI 15"
+        params["parameters"]["retriever_model"] = _dense_openai_retriever(top_k=15)
+        stored = _post_and_get(client, params)
+        assert stored["parameters"]["retriever_model"]["params"]["top_k"] == 15
+
+    def test_topk_bm25_20(self, client: TestClient, test_doc_id: int):
+        """BM25Retriever with top_k=20."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK BM25 20"
+        params["parameters"]["retriever_model"]["params"]["top_k"] = 20
+        stored = _post_and_get(client, params)
+        assert stored["parameters"]["retriever_model"]["params"]["top_k"] == 20
+
+    def test_topk_mmr_12(self, client: TestClient, test_doc_id: int):
+        """MMRRerankerRetriever with top_k=12 (child retrieves 36)."""
+        params = _base_params(test_doc_id)
+        params["name"] = "TopK MMR 12"
+        params["parameters"]["retriever_model"] = {
+            "component": "MMRRerankerRetriever",
+            "params": {
+                "mmr_lambda": 0.5,
+                "retrieval_factor": 3,
+                "top_k": 12,
+                "children": [
+                    _dense_st_retriever(similarity_metric="cosine", top_k=36)
+                ],
+            },
+        }
+        stored = _post_and_get(client, params)
+        ret = stored["parameters"]["retriever_model"]
+        assert ret["params"]["top_k"] == 12
+        assert ret["params"]["children"][0]["params"]["top_k"] == 36
+
+
+# ===================================================================
+# Category 4: Chunking Strategies  (6 tests)
+# ===================================================================
+
+
+class TestChunkingStrategies:
+    """6 chunking strategy configurations with different algorithms and parameters (chunk_size, chunk_overlap)."""
+
+    def test_chunking_char_small(self, client: TestClient, test_doc_id: int):
+        """CharacterChunkModel with chunk_size=256, chunk_overlap=25."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Char 256/25"
+        params["parameters"]["chunking_model"] = {
+            "component": "CharacterChunkModel",
+            "params": {"chunk_size": 256, "chunk_overlap": 25},
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "CharacterChunkModel"
+        assert chunk["params"]["chunk_size"] == 256
+        assert chunk["params"]["chunk_overlap"] == 25
+
+    def test_chunking_char_paragraph(self, client: TestClient, test_doc_id: int):
+        """CharacterChunkModel with chunk_size=500, chunk_overlap=50."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Char 500/50"
+        params["parameters"]["chunking_model"] = {
+            "component": "CharacterChunkModel",
+            "params": {"chunk_size": 500, "chunk_overlap": 50},
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "CharacterChunkModel"
+        assert chunk["params"]["chunk_size"] == 500
+        assert chunk["params"]["chunk_overlap"] == 50
+
+    def test_chunking_char_page(self, client: TestClient, test_doc_id: int):
+        """CharacterChunkModel with chunk_size=2000, chunk_overlap=200."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Char 2000/200"
+        params["parameters"]["chunking_model"] = {
+            "component": "CharacterChunkModel",
+            "params": {"chunk_size": 2000, "chunk_overlap": 200},
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "CharacterChunkModel"
+        assert chunk["params"]["chunk_size"] == 2000
+        assert chunk["params"]["chunk_overlap"] == 200
+
+    def test_chunking_recursive_custom(self, client: TestClient, test_doc_id: int):
+        """RecursiveCharacterChunkModel with custom separators."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Recursive Custom"
+        params["parameters"]["chunking_model"] = {
+            "component": "RecursiveCharacterChunkModel",
+            "params": {
+                "chunk_size": 1000,
+                "chunk_overlap": 100,
+                "separators": ["\n\n", "\n", ".", " ", ""],
+            },
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "RecursiveCharacterChunkModel"
+        assert chunk["params"]["chunk_size"] == 1000
+        assert chunk["params"]["chunk_overlap"] == 100
+        assert chunk["params"]["separators"] == ["\n\n", "\n", ".", " ", ""]
+
+    def test_chunking_token_e5_mistral(self, client: TestClient, test_doc_id: int):
+        """TokenChunkModel with e5-mistral tokenizer, chunk_size=300, overlap=60."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Token E5"
+        params["parameters"]["chunking_model"] = {
+            "component": "TokenChunkModel",
+            "params": {
+                "tokenizer_name": "intfloat/e5-mistral-7b-instruct",
+                "chunk_size": 300,
+                "chunk_overlap": 60,
+            },
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "TokenChunkModel"
+        assert chunk["params"]["tokenizer_name"] == "intfloat/e5-mistral-7b-instruct"
+        assert chunk["params"]["chunk_size"] == 300
+        assert chunk["params"]["chunk_overlap"] == 60
+
+    def test_chunking_token_bert_spanish(self, client: TestClient, test_doc_id: int):
+        """TokenChunkModel with BERT Spanish tokenizer, chunk_size=512, overlap=50."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Chunk Token BERT ES"
+        params["parameters"]["chunking_model"] = {
+            "component": "TokenChunkModel",
+            "params": {
+                "tokenizer_name": "dccuchile/bert-base-spanish-wwm-uncased",
+                "chunk_size": 512,
+                "chunk_overlap": 50,
+            },
+        }
+        stored = _post_and_get(client, params)
+        chunk = stored["parameters"]["chunking_model"]
+        assert chunk["component"] == "TokenChunkModel"
+        assert chunk["params"]["tokenizer_name"] == "dccuchile/bert-base-spanish-wwm-uncased"
+        assert chunk["params"]["chunk_size"] == 512
+        assert chunk["params"]["chunk_overlap"] == 50
+
+
+# ===================================================================
+# Category 5: Prompts  (4 tests)
+# ===================================================================
+
+
+class TestPrompts:
+    """4 prompt configurations validating formatting and chat session consistency."""
+
+    def test_prompt_default_rag_en(self, client: TestClient, test_doc_id: int):
+        """DefaultRAGGenerationPrompt with language=en."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Prompt RAG EN"
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultRAGGenerationPrompt"
+        assert prompt["params"]["language"] == "en"
+
+    def test_prompt_default_rag_es(self, client: TestClient, test_doc_id: int):
+        """DefaultRAGGenerationPrompt with language=es."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Prompt RAG ES"
+        params["parameters"]["prompt"] = {
+            "component": "DefaultRAGGenerationPrompt",
+            "params": {"language": "es"},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultRAGGenerationPrompt"
+        assert prompt["params"]["language"] == "es"
+
+    def test_prompt_default_qna_en(self, client: TestClient, test_doc_id: int):
+        """DefaultQnARAGGenerationPrompt with language=en."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Prompt QnA EN"
+        params["parameters"]["prompt"] = {
+            "component": "DefaultQnARAGGenerationPrompt",
+            "params": {"language": "en"},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultQnARAGGenerationPrompt"
+        assert prompt["params"]["language"] == "en"
+
+    def test_prompt_custom_template(self, client: TestClient, test_doc_id: int):
+        """CustomRAGGenerationPrompt with user-defined template."""
+        template_text = "Answer the question based on: {chunks}\n\nQuestion: {input}"
+        params = _base_params(test_doc_id)
+        params["name"] = "Prompt Custom"
+        params["parameters"]["prompt"] = {
+            "component": "CustomRAGGenerationPrompt",
+            "params": {"template": template_text},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "CustomRAGGenerationPrompt"
+        assert prompt["params"]["template"] == template_text
+
+
+# ===================================================================
+# Category 6: Generator Models  (8 tests)
+# ===================================================================
+
+
+class TestGeneratorModels:
+    """8 generator model configurations with different models and hyperparameters (≤8B)."""
+
+    def test_generator_llama_1b_default(self, client: TestClient, test_doc_id: int):
+        """LlamaModel 3.2-1B with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Llama 1B Default"
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "LlamaModel"
+        assert gen["params"]["model_name"] == LLAMA_1B
+        assert gen["params"]["max_tokens"] == 100
+        assert gen["params"]["temperature"] == 0.7
+        assert gen["params"]["frequency_penalty"] == 0.1
+        assert gen["params"]["context_window"] == 512
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_llama_3b_custom(self, client: TestClient, test_doc_id: int):
+        """LlamaModel 3.2-3B with custom hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Llama 3B Custom"
+        params["parameters"]["generation_model"] = {
+            "component": "LlamaModel",
+            "params": {
+                "model_name": LLAMA_3B,
+                "max_tokens": 512,
+                "temperature": 0.5,
+                "frequency_penalty": 0.0,
+                "context_window": 2048,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "LlamaModel"
+        assert gen["params"]["model_name"] == LLAMA_3B
+        assert gen["params"]["max_tokens"] == 512
+        assert gen["params"]["temperature"] == 0.5
+        assert gen["params"]["frequency_penalty"] == 0.0
+        assert gen["params"]["context_window"] == 2048
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_mistral_7b_default(self, client: TestClient, test_doc_id: int):
+        """MistralModel 7B with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Mistral 7B Default"
+        params["parameters"]["generation_model"] = {
+            "component": "MistralModel",
+            "params": {
+                "model_name": "bartowski/Mistral-7B-Instruct-v0.3-GGUF",
+                "max_tokens": 100,
+                "temperature": 0.7,
+                "frequency_penalty": 0.1,
+                "context_window": 512,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "MistralModel"
+        assert gen["params"]["model_name"] == "bartowski/Mistral-7B-Instruct-v0.3-GGUF"
+        assert gen["params"]["max_tokens"] == 100
+        assert gen["params"]["temperature"] == 0.7
+        assert gen["params"]["frequency_penalty"] == 0.1
+        assert gen["params"]["context_window"] == 512
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_qwen_0_5b_default(self, client: TestClient, test_doc_id: int):
+        """QwenModel 0.5B with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Qwen 0.5B Default"
+        params["parameters"]["generation_model"] = {
+            "component": "QwenModel",
+            "params": {
+                "model_name": "Qwen/Qwen2.5-0.5B-Instruct-GGUF",
+                "max_tokens": 100,
+                "temperature": 0.7,
+                "frequency_penalty": 0.1,
+                "context_window": 512,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "QwenModel"
+        assert gen["params"]["model_name"] == "Qwen/Qwen2.5-0.5B-Instruct-GGUF"
+        assert gen["params"]["max_tokens"] == 100
+        assert gen["params"]["temperature"] == 0.7
+        assert gen["params"]["frequency_penalty"] == 0.1
+        assert gen["params"]["context_window"] == 512
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_qwen_1_5b_custom(self, client: TestClient, test_doc_id: int):
+        """QwenModel 1.5B with custom hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Qwen 1.5B Custom"
+        params["parameters"]["generation_model"] = {
+            "component": "QwenModel",
+            "params": {
+                "model_name": "Qwen/Qwen2.5-1.5B-Instruct-GGUF",
+                "max_tokens": 256,
+                "temperature": 0.3,
+                "frequency_penalty": 0.5,
+                "context_window": 1024,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "QwenModel"
+        assert gen["params"]["model_name"] == "Qwen/Qwen2.5-1.5B-Instruct-GGUF"
+        assert gen["params"]["max_tokens"] == 256
+        assert gen["params"]["temperature"] == 0.3
+        assert gen["params"]["frequency_penalty"] == 0.5
+        assert gen["params"]["context_window"] == 1024
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_smol_1_7b_default(self, client: TestClient, test_doc_id: int):
+        """SmolLMModel 1.7B with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen SmolLM 1.7B Default"
+        params["parameters"]["generation_model"] = {
+            "component": "SmolLMModel",
+            "params": {
+                "model_name": "HuggingFaceTB/SmolLM2-1.7B-Instruct-GGUF",
+                "max_tokens": 100,
+                "temperature": 0.7,
+                "frequency_penalty": 0.1,
+                "context_window": 512,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "SmolLMModel"
+        assert gen["params"]["model_name"] == "HuggingFaceTB/SmolLM2-1.7B-Instruct-GGUF"
+        assert gen["params"]["max_tokens"] == 100
+        assert gen["params"]["temperature"] == 0.7
+        assert gen["params"]["frequency_penalty"] == 0.1
+        assert gen["params"]["context_window"] == 512
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_phi4_mini_default(self, client: TestClient, test_doc_id: int):
+        """Phi4MiniInstructModel with default hyperparams."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen Phi4 Mini Default"
+        params["parameters"]["generation_model"] = {
+            "component": "Phi4MiniInstructModel",
+            "params": {
+                "model_name": "unsloth/Phi-4-mini-instruct-GGUF",
+                "max_tokens": 100,
+                "temperature": 0.7,
+                "frequency_penalty": 0.1,
+                "context_window": 512,
+                "device": "CPU",
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "Phi4MiniInstructModel"
+        assert gen["params"]["model_name"] == "unsloth/Phi-4-mini-instruct-GGUF"
+        assert gen["params"]["max_tokens"] == 100
+        assert gen["params"]["temperature"] == 0.7
+        assert gen["params"]["frequency_penalty"] == 0.1
+        assert gen["params"]["context_window"] == 512
+        assert gen["params"]["device"] == "CPU"
+
+    def test_generator_openai_gpt4o_mini_custom(
+        self, client: TestClient, test_doc_id: int
+    ):
+        """OpenAITextToTextGenerationModel with gpt-4o-mini, temp=0.3, max_tokens=512."""
+        params = _base_params(test_doc_id)
+        params["name"] = "Gen OpenAI GPT4o Mini"
+        params["parameters"]["generation_model"] = {
+            "component": "OpenAITextToTextGenerationModel",
+            "params": {
+                "model_name": "gpt-4o-mini",
+                "API_key": "test-key",
+                "temperature": 0.3,
+                "max_completions_tokens": 512,
+            },
+        }
+        stored = _post_and_get(client, params)
+        gen = stored["parameters"]["generation_model"]
+        assert gen["component"] == "OpenAITextToTextGenerationModel"
+        assert gen["params"]["model_name"] == "gpt-4o-mini"
+        assert gen["params"]["API_key"] == "test-key"
+        assert gen["params"]["temperature"] == 0.3
+        assert gen["params"]["max_completions_tokens"] == 512
diff --git a/tests/back/RAG/test_rag_pipeline_api_configs.py b/tests/back/RAG/test_rag_pipeline_api_configs.py
new file mode 100644
index 000000000..883fa6060
--- /dev/null
+++ b/tests/back/RAG/test_rag_pipeline_api_configs.py
@@ -0,0 +1,451 @@
+"""Integration tests validating RAG pipeline configurations from published research.
+
+Each test creates a RAG session through the API with the exact parameter set
+specified in the referenced publications and verifies:
+
+- Session creation returns 201 Created
+- Model name ("RAGPipeline") and task name ("RAGTask") are correct
+- All parameters can be retrieved via GET and match the expected configuration
+
+Note: Session creation validates only the top-level RAGPipeline schema
+({component, params} structure). Sub-component schemas (LlamaModel,
+OpenAIEmbedding, etc.) are validated at pipeline runtime, not during
+session creation.
+"""
+
+import pytest
+from fastapi.testclient import TestClient
+
+from DashAI.back.dependencies.database.models import Document
+
+
+def _create_test_document(client: TestClient, suffix: str = "") -> int:
+    """Create a minimal test document in the DB and return its ID."""
+    session_factory = client.app.container["session_factory"]
+    with session_factory() as db:
+        doc = Document(
+            file_name=f"test_doc{suffix}.txt",
+            file_type="txt",
+            file_path=f"/tmp/test_doc{suffix}.txt",
+            file_hash=f"test_hash_123_{suffix}" if suffix else "test_hash_123",
+        )
+        db.add(doc)
+        db.commit()
+        db.refresh(doc)
+        return doc.id
+
+
+@pytest.fixture(scope="module")
+def test_doc_id(client: TestClient) -> int:
+    """Module-scoped test document ID shared across all pipeline config tests."""
+    return _create_test_document(client, suffix="_pipeline_configs")
+
+
+def test_publication_1_medical_fitness(client: TestClient, test_doc_id: int):
+    """Retrieval augmented generation for 10 large language models and its
+    generalizability in assessing medical fitness.
+
+    Configuration:
+    - Chunking:  RecursiveCharacterChunkModel (1000/100)
+    - Retrieval: DenseEmbeddingRetriever + OpenAIEmbedding (ada-002, cosine, top_k=10)
+    - Prompt:    DefaultRAGGenerationPrompt (en)
+    - Generator: LlamaModel (3.2-3B, 1024 tokens, temp 0.1)
+    """
+    params = {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "RecursiveCharacterChunkModel",
+                "params": {"chunk_size": 1000, "chunk_overlap": 100},
+            },
+            "retriever_model": {
+                "component": "DenseEmbeddingRetriever",
+                "params": {
+                    "embedding_model": {
+                        "component": "OpenAIEmbedding",
+                        "params": {
+                            "model_name": "text-embedding-ada-002",
+                            "api_key": "test-key",
+                        },
+                    },
+                    "similarity_metric": "cosine",
+                    "top_k": 10,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": "bartowski/Llama-3.2-3B-Instruct-GGUF",
+                    "max_tokens": 1024,
+                    "temperature": 0.1,
+                    "frequency_penalty": 0.0,
+                    "context_window": 4096,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Pub 1 - Medical Fitness",
+        "description": None,
+    }
+
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == "Pub 1 - Medical Fitness"
+
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    stored = get_resp.json()
+    chunking = stored["parameters"]["chunking_model"]
+    retriever = stored["parameters"]["retriever_model"]
+    assert chunking["component"] == "RecursiveCharacterChunkModel"
+    assert retriever["component"] == "DenseEmbeddingRetriever"
+    assert retriever["params"]["embedding_model"]["component"] == "OpenAIEmbedding"
+    assert stored["parameters"]["generation_model"]["component"] == "LlamaModel"
+    assert stored["parameters"]["prompt"]["component"] == "DefaultRAGGenerationPrompt"
+
+
+def test_publication_2_ehr_summarization(client: TestClient, test_doc_id: int):
+    """Applying generative AI with retrieval augmented generation to summarize
+    and extract key clinical information from electronic health records.
+
+    Configuration:
+    - Chunking:  CharacterChunkModel (600/40)
+    - Retrieval: MMRRerankerRetriever (lambda=0.5) wrapping
+                 DenseEmbeddingRetriever + SentenceTransformerEmbedding
+                 (all-MiniLM-L6-v2, cosine, retrieval top_k=60)
+    - Prompt:    DefaultRAGGenerationPrompt (en)
+    - Generator: LlamaModel (3.2-3B)
+    """
+    mname = "sentence-transformers/all-MiniLM-L6-v2"
+    params = {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "CharacterChunkModel",
+                "params": {"chunk_size": 600, "chunk_overlap": 40},
+            },
+            "retriever_model": {
+                "component": "MMRRerankerRetriever",
+                "params": {
+                    "mmr_lambda": 0.5,
+                    "retrieval_factor": 3,
+                    "top_k": 20,
+                    "children": [
+                        {
+                            "component": "DenseEmbeddingRetriever",
+                            "params": {
+                                "embedding_model": {
+                                    "component": "SentenceTransformerEmbedding",
+                                    "params": {
+                                        "model_name": mname,
+                                        "normalize": True,
+                                        "device": "cpu",
+                                    },
+                                },
+                                "similarity_metric": "cosine",
+                                "top_k": 60,
+                            },
+                        }
+                    ],
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": "bartowski/Llama-3.2-3B-Instruct-GGUF",
+                    "max_tokens": 1024,
+                    "temperature": 0.1,
+                    "frequency_penalty": 0.0,
+                    "context_window": 4096,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Pub 2 - EHR Summarization",
+        "description": None,
+    }
+
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == "Pub 2 - EHR Summarization"
+
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    stored = get_resp.json()
+    assert stored["parameters"]["chunking_model"]["component"] == "CharacterChunkModel"
+    assert (
+        stored["parameters"]["retriever_model"]["component"] == "MMRRerankerRetriever"
+    )
+    assert stored["parameters"]["generation_model"]["component"] == "LlamaModel"
+    assert stored["parameters"]["prompt"]["component"] == "DefaultRAGGenerationPrompt"
+
+    # Verify the MMRRerankerRetriever wraps a DenseEmbeddingRetriever child
+    retriever = stored["parameters"]["retriever_model"]
+    children = retriever["params"]["children"]
+    assert len(children) == 1
+    assert children[0]["component"] == "DenseEmbeddingRetriever"
+    assert (
+        children[0]["params"]["embedding_model"]["component"]
+        == "SentenceTransformerEmbedding"
+    )
+
+
+def test_publication_3_case_study(client: TestClient, test_doc_id: int):
+    """Development and Testing of Retrieval Augmented Generation in Large
+    Language Models -- A Case Study Report.
+
+    Same retriever and chunking configuration as the medical fitness study
+    but with Llama 3.2-1B (smaller model).
+
+    Configuration:
+    - Chunking:  RecursiveCharacterChunkModel (1000/100)
+    - Retrieval: DenseEmbeddingRetriever + OpenAIEmbedding (ada-002, cosine, top_k=10)
+    - Prompt:    DefaultRAGGenerationPrompt (en)
+    - Generator: LlamaModel (3.2-1B)
+    """
+    params = {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "RecursiveCharacterChunkModel",
+                "params": {"chunk_size": 1000, "chunk_overlap": 100},
+            },
+            "retriever_model": {
+                "component": "DenseEmbeddingRetriever",
+                "params": {
+                    "embedding_model": {
+                        "component": "OpenAIEmbedding",
+                        "params": {
+                            "model_name": "text-embedding-ada-002",
+                            "api_key": "test-key",
+                        },
+                    },
+                    "similarity_metric": "cosine",
+                    "top_k": 10,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": "bartowski/Llama-3.2-1B-Instruct-GGUF",
+                    "max_tokens": 1024,
+                    "temperature": 0.1,
+                    "frequency_penalty": 0.0,
+                    "context_window": 4096,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Pub 3 - Case Study",
+        "description": None,
+    }
+
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == "Pub 3 - Case Study"
+
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    stored = get_resp.json()
+    chunking = stored["parameters"]["chunking_model"]
+    retriever = stored["parameters"]["retriever_model"]
+    assert chunking["component"] == "RecursiveCharacterChunkModel"
+    assert retriever["component"] == "DenseEmbeddingRetriever"
+    assert retriever["params"]["embedding_model"]["component"] == "OpenAIEmbedding"
+    gen = stored["parameters"]["generation_model"]
+    assert gen["component"] == "LlamaModel"
+    # Verify the 1B model variant is stored
+    assert gen["params"]["model_name"] == "bartowski/Llama-3.2-1B-Instruct-GGUF"
+
+
+def test_publication_4a_ragchecker_dense(client: TestClient, test_doc_id: int):
+    """RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented
+    Generation (dense variant).
+
+    Configuration:
+    - Chunking:  TokenChunkModel (300/60, e5-mistral tokenizer)
+    - Retrieval: DenseEmbeddingRetriever + E5Embedding
+                 (e5-mistral-7b-instruct, cosine, top_k=20)
+    - Prompt:    DefaultRAGGenerationPrompt (en)
+    - Generator: LlamaModel (3.2-3B)
+    """
+    params = {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "TokenChunkModel",
+                "params": {
+                    "tokenizer_name": "intfloat/e5-mistral-7b-instruct",
+                    "chunk_size": 300,
+                    "chunk_overlap": 60,
+                },
+            },
+            "retriever_model": {
+                "component": "DenseEmbeddingRetriever",
+                "params": {
+                    "embedding_model": {
+                        "component": "E5Embedding",
+                        "params": {
+                            "model_name": "intfloat/e5-mistral-7b-instruct",
+                            "device": "cpu",
+                        },
+                    },
+                    "similarity_metric": "cosine",
+                    "top_k": 20,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": "bartowski/Llama-3.2-3B-Instruct-GGUF",
+                    "max_tokens": 1024,
+                    "temperature": 0.1,
+                    "frequency_penalty": 0.0,
+                    "context_window": 4096,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Pub 4a - RAGChecker Dense",
+        "description": None,
+    }
+
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == "Pub 4a - RAGChecker Dense"
+
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    stored = get_resp.json()
+    assert stored["parameters"]["chunking_model"]["component"] == "TokenChunkModel"
+    retriever = stored["parameters"]["retriever_model"]
+    assert retriever["component"] == "DenseEmbeddingRetriever"
+    embed = retriever["params"]["embedding_model"]
+    assert embed["component"] == "E5Embedding"
+    assert embed["params"]["model_name"] == "intfloat/e5-mistral-7b-instruct"
+    assert retriever["params"]["top_k"] == 20
+    assert stored["parameters"]["generation_model"]["component"] == "LlamaModel"
+
+
+def test_publication_4b_ragchecker_sparse(client: TestClient, test_doc_id: int):
+    """RAGChecker: A Fine-grained Framework for Diagnosing Retrieval-Augmented
+    Generation (sparse variant).
+
+    Same as 4a but with BM25Retriever instead of DenseEmbeddingRetriever.
+
+    Configuration:
+    - Chunking:  TokenChunkModel (300/60, e5-mistral tokenizer)
+    - Retrieval: BM25Retriever (cosine, top_k=20, k1=1.5, b=0.75)
+    - Prompt:    DefaultRAGGenerationPrompt (en)
+    - Generator: LlamaModel (3.2-3B)
+    """
+    params = {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "TokenChunkModel",
+                "params": {
+                    "tokenizer_name": "intfloat/e5-mistral-7b-instruct",
+                    "chunk_size": 300,
+                    "chunk_overlap": 60,
+                },
+            },
+            "retriever_model": {
+                "component": "BM25Retriever",
+                "params": {
+                    "BM25Vectorizer": {
+                        "component": "BM25VectorizerModel",
+                        "params": {},
+                    },
+                    "similarity_function": "cosine",
+                    "top_k": 20,
+                    "k1": 1.5,
+                    "b": 0.75,
+                    "delta": 0.0,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": "bartowski/Llama-3.2-3B-Instruct-GGUF",
+                    "max_tokens": 1024,
+                    "temperature": 0.1,
+                    "frequency_penalty": 0.0,
+                    "context_window": 4096,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Pub 4b - RAGChecker Sparse",
+        "description": None,
+    }
+
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == "Pub 4b - RAGChecker Sparse"
+
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    stored = get_resp.json()
+    assert stored["parameters"]["chunking_model"]["component"] == "TokenChunkModel"
+    retriever = stored["parameters"]["retriever_model"]
+    assert retriever["component"] == "BM25Retriever"
+    assert retriever["params"]["top_k"] == 20
+    assert retriever["params"]["k1"] == 1.5
+    assert retriever["params"]["b"] == 0.75
+    assert stored["parameters"]["generation_model"]["component"] == "LlamaModel"
diff --git a/tests/back/RAG/test_rag_prompt_updates.py b/tests/back/RAG/test_rag_prompt_updates.py
new file mode 100644
index 000000000..3621c8746
--- /dev/null
+++ b/tests/back/RAG/test_rag_prompt_updates.py
@@ -0,0 +1,126 @@
+
+
+from DashAI.back.dependencies.database.models import GenerativeSession, RAGPrompt
+
+
+def _get_prompt_list(client):
+    response = client.get("/api/v1/prompt/")
+    assert response.status_code == 200
+    return response.json()
+
+
+def _create_rag_session(session_factory, prompt_id: int, name: str):
+    params = {
+        "documents": [1],
+        "chunking_model": {
+            "component": "CharacterChunkModel",
+            "params": {
+                "chunk_size": 8,
+                "chunk_overlap": 2,
+            },
+        },
+        "retriever_model": {
+            "component": "TFIDFRetriever",
+            "params": {
+                "TFIDFVectorizer": {
+                    "component": "TFIDFVectorizerModel",
+                    "params": {},
+                },
+                "similarity_function": "cosine",
+                "top_k": 1,
+                "similarity_threshold": None,
+            },
+        },
+        "generation_model": {
+            "component": "QwenModel",
+            "params": {
+                "model_name": "Qwen/Qwen2.5-1.5B-Instruct-GGUF",
+                "max_tokens": 32,
+                "temperature": 0.2,
+                "frequency_penalty": 0.0,
+                "context_window": 128,
+                "device": "CPU",
+            },
+        },
+        "prompt_id": prompt_id,
+    }
+
+    with session_factory() as db:
+        session = GenerativeSession(
+            model_name="RAGPipeline",
+            task_name="RAGTask",
+            parameters=params,
+            name=name,
+            description=None,
+        )
+        db.add(session)
+        db.commit()
+        db.refresh(session)
+        return session.id
+
+
+def test_update_prompt_in_place(client):
+    prompts = _get_prompt_list(client)
+    prompt = prompts[0]
+
+    response = client.patch(
+        f"/api/v1/prompt/{prompt['id']}",
+        json={"name": f"{prompt['name']} (updated)"},
+    )
+
+    assert response.status_code == 200
+    data = response.json()
+    assert data["id"] == prompt["id"]
+    assert data["name"] == f"{prompt['name']} (updated)"
+
+
+def test_clone_prompt_for_session(client):
+    prompts = _get_prompt_list(client)
+    prompt = prompts[1]
+    session_factory = client.app.container["session_factory"]
+    session_id = _create_rag_session(session_factory, prompt["id"], "rag-session-clone")
+
+    response = client.post(f"/api/v1/prompt/{prompt['id']}/sessions/{session_id}", json={})
+
+    assert response.status_code == 201
+    data = response.json()
+    assert data["session_id"] == session_id
+    assert data["parameters"]["prompt_id"] == data["prompt"]["id"]
+    assert data["prompt"]["name"].endswith(f"session {session_id}")
+
+
+def test_session_parameter_prompt_reassignment(client):
+    """Verify that updating session parameters can switch the prompt.
+
+    The endpoint accepts a ``prompt_id`` in the payload and converts it to
+    a ``prompt`` dict with ``component`` and ``params`` keys. Prompt-level
+    cleanup is not implemented, so orphaned clones are left in the DB.
+    """
+    prompts = _get_prompt_list(client)
+    base_prompt = prompts[1]
+    session_factory = client.app.container["session_factory"]
+    session_id = _create_rag_session(session_factory, base_prompt["id"], "rag-session-cleanup")
+
+    clone_response = client.post(
+        f"/api/v1/prompt/{base_prompt['id']}/sessions/{session_id}",
+        json={},
+    )
+    assert clone_response.status_code == 201
+    cloned_prompt_id = clone_response.json()["prompt"]["id"]
+
+    update_response = client.put(
+        f"/api/v1/generative-session/{session_id}/parameters",
+        json={"prompt_id": base_prompt["id"]},
+    )
+
+    assert update_response.status_code == 200
+    # The endpoint converts prompt_id → prompt dict with component + params
+    prompt_param = update_response.json()["parameters"]["prompt"]
+    assert prompt_param["component"] == base_prompt["class_name"]
+    assert "params" in prompt_param
+
+    # The cloned prompt is NOT cleaned up by the API (orphan cleanup
+    # only handles retrievers and chunking models, not prompts).
+    with session_factory() as db:
+        orphan = db.get(RAGPrompt, cloned_prompt_id)
+        assert orphan is not None, "Cloned prompt should still exist (no prompt cleanup)."
diff --git a/tests/back/RAG/test_rag_prompts.py b/tests/back/RAG/test_rag_prompts.py
new file mode 100644
index 000000000..818a7892b
--- /dev/null
+++ b/tests/back/RAG/test_rag_prompts.py
@@ -0,0 +1,379 @@
+"""Integration tests for prompt CRUD API and prompt usage in generative sessions.
+
+Covers:
+- Prompt CRUD operations (list, create, update)
+- Prompt validation (duplicates, missing fields, invalid class names)
+- Prompt usage within generative RAG sessions (default/templates, language, custom)
+- Prompt cloning to sessions
+"""
+
+import pytest
+from fastapi.testclient import TestClient
+from sqlalchemy import exc as sa_exc
+
+from DashAI.back.dependencies.database.models import Document
+
+# ---------------------------------------------------------------------------
+# helpers — matching test_rag_component_api_configs.py patterns
+# ---------------------------------------------------------------------------
+
+LLAMA_1B = "bartowski/Llama-3.2-1B-Instruct-GGUF"
+
+
+def _create_test_document(client: TestClient, suffix: str = "") -> int:
+    """Create a minimal test document in the DB and return its ID."""
+    session_factory = client.app.container["session_factory"]
+    with session_factory() as db:
+        doc = Document(
+            file_name=f"test_doc{suffix}.txt",
+            file_type="txt",
+            file_path=f"/tmp/test_doc{suffix}.txt",
+            file_hash=f"test_hash_123_{suffix}" if suffix else "test_hash_123",
+        )
+        db.add(doc)
+        db.commit()
+        db.refresh(doc)
+        return doc.id
+
+
+@pytest.fixture(scope="module")
+def test_doc_id(client: TestClient) -> int:
+    """Module-scoped test document ID shared across all prompt tests."""
+    return _create_test_document(client, suffix="_prompts")
+
+
+def _base_session_params(test_doc_id: int) -> dict:
+    """Return the minimal default RAG session payload."""
+    return {
+        "model_name": "RAGPipeline",
+        "task_name": "RAGTask",
+        "parameters": {
+            "documents": [test_doc_id],
+            "chunking_model": {
+                "component": "CharacterChunkModel",
+                "params": {"chunk_size": 400, "chunk_overlap": 40},
+            },
+            "retriever_model": {
+                "component": "BM25Retriever",
+                "params": {
+                    "BM25Vectorizer": {
+                        "component": "BM25VectorizerModel",
+                        "params": {},
+                    },
+                    "k1": 1.5,
+                    "b": 0.75,
+                    "delta": 0.0,
+                    "similarity_function": "cosine",
+                    "top_k": 5,
+                },
+            },
+            "generation_model": {
+                "component": "LlamaModel",
+                "params": {
+                    "model_name": LLAMA_1B,
+                    "max_tokens": 100,
+                    "temperature": 0.7,
+                    "frequency_penalty": 0.1,
+                    "context_window": 512,
+                    "device": "CPU",
+                },
+            },
+            "prompt": {
+                "component": "DefaultRAGGenerationPrompt",
+                "params": {"language": "en"},
+            },
+        },
+        "name": "Prompt Test",
+        "description": None,
+    }
+
+
+def _post_and_get(client: TestClient, params: dict):
+    """POST a session, assert 201 + metadata, then GET and return stored JSON."""
+    response = client.post("/api/v1/generative-session/", json=params)
+    assert response.status_code == 201, f"Session creation failed: {response.text}"
+    data = response.json()
+    assert data["id"] is not None
+    assert data["model_name"] == "RAGPipeline"
+    assert data["task_name"] == "RAGTask"
+    assert data["name"] == params["name"]
+    session_id = data["id"]
+    get_resp = client.get(f"/api/v1/generative-session/{session_id}")
+    assert get_resp.status_code == 200
+    return get_resp.json()
+
+
+# ===================================================================
+# Prompt CRUD Operations
+# ===================================================================
+
+
+class TestPromptCRUD:
+    """Prompt CRUD operations via the prompt API."""
+
+    def test_list_prompts_returns_defaults(self, client: TestClient):
+        """GET /api/v1/prompt/ returns the lazily-seeded default prompts."""
+        response = client.get("/api/v1/prompt/")
+        assert response.status_code == 200
+        data = response.json()
+        assert isinstance(data, list)
+        assert len(data) >= 2, "Expected at least 2 default prompts in the list"
+
+        class_names = [p["class_name"] for p in data]
+        assert "DefaultRAGGenerationPrompt" in class_names, (
+            "DefaultRAGGenerationPrompt not found in prompt list"
+        )
+        assert "DefaultQnARAGGenerationPrompt" in class_names, (
+            "DefaultQnARAGGenerationPrompt not found in prompt list"
+        )
+
+    def test_create_custom_prompt(self, client: TestClient):
+        """POST /api/v1/prompt/ creates a new CustomRAGGenerationPrompt and it appears in the list."""
+        template = "Answer using: {chunks}\n\nQuestion: {input}"
+        payload = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Test Custom Prompt",
+            "parameters": {"template": template},
+        }
+        response = client.post("/api/v1/prompt/", json=payload)
+        assert response.status_code == 201, f"Expected 201, got {response.status_code}: {response.text}"
+        data = response.json()
+        assert data["id"] is not None, "Created prompt should have an id"
+
+        list_resp = client.get("/api/v1/prompt/")
+        assert list_resp.status_code == 200
+        prompts = list_resp.json()
+        custom = [p for p in prompts if p["name"] == "Test Custom Prompt"]
+        assert len(custom) == 1, "Custom prompt not found in list"
+        assert custom[0]["class_name"] == "CustomRAGGenerationPrompt"
+        assert custom[0]["parameters"]["template"] == template
+
+    def test_create_duplicate_prompt_fails(self, client: TestClient):
+        """Creating the same prompt twice (same class_name + parameters) is rejected via IntegrityError."""
+        template = "Q: {input}\nContext: {chunks}"
+        payload = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Dup Prompt 1",
+            "parameters": {"template": template},
+        }
+
+        resp1 = client.post("/api/v1/prompt/", json=payload)
+        assert resp1.status_code == 201, f"First creation should succeed: {resp1.text}"
+
+        # Second creation with same class_name + parameters but different name.
+        # The TestClient re-raises unhandled exceptions from the app, so the
+        # IntegrityError on the UNIQUE constraint propagates directly.
+        payload2 = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Dup Prompt 2",
+            "parameters": {"template": template},
+        }
+        with pytest.raises(sa_exc.IntegrityError):
+            client.post("/api/v1/prompt/", json=payload2)
+
+    def test_update_prompt_name(self, client: TestClient):
+        """PATCH /api/v1/prompt/{id} updates the prompt name."""
+        # Create a prompt first
+        create_payload = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Original Name",
+            "parameters": {"template": "A: {chunks} Q: {input}"},
+        }
+        resp = client.post("/api/v1/prompt/", json=create_payload)
+        assert resp.status_code == 201, f"Creation failed: {resp.text}"
+        prompt_id = resp.json()["id"]
+
+        # Update the name
+        patch_resp = client.patch(
+            f"/api/v1/prompt/{prompt_id}", json={"name": "Updated Name"}
+        )
+        assert patch_resp.status_code == 200, f"PATCH failed: {patch_resp.text}"
+        updated = patch_resp.json()
+        assert updated["name"] == "Updated Name", "Name should be updated"
+        assert updated["id"] == prompt_id
+
+        # Verify persistence
+        get_resp = client.get("/api/v1/prompt/")
+        prompts = get_resp.json()
+        match = [p for p in prompts if p["id"] == prompt_id]
+        assert len(match) == 1
+        assert match[0]["name"] == "Updated Name"
+
+    def test_update_prompt_parameters(self, client: TestClient):
+        """PATCH /api/v1/prompt/{id} updates the prompt template."""
+        original_template = "Docs: {chunks}\nQuery: {input}"
+        create_payload = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Params Test",
+            "parameters": {"template": original_template},
+        }
+        resp = client.post("/api/v1/prompt/", json=create_payload)
+        assert resp.status_code == 201, f"Creation failed: {resp.text}"
+        prompt_id = resp.json()["id"]
+
+        new_template = "Context: {chunks}\n\nUser: {input}\nAnswer:"
+        patch_resp = client.patch(
+            f"/api/v1/prompt/{prompt_id}",
+            json={"parameters": {"template": new_template}},
+        )
+        assert patch_resp.status_code == 200, f"PATCH failed: {patch_resp.text}"
+        updated = patch_resp.json()
+        assert updated["parameters"]["template"] == new_template
+
+    def test_create_prompt_missing_required_field(self, client: TestClient):
+        """POST without class_name or without name returns 422."""
+        # Without class_name
+        resp1 = client.post(
+            "/api/v1/prompt/",
+            json={"name": "Missing Class", "parameters": {"template": "Q: {input}"}},
+        )
+        assert resp1.status_code == 422, (
+            f"Expected 422 for missing class_name, got {resp1.status_code}"
+        )
+
+        # Without name
+        resp2 = client.post(
+            "/api/v1/prompt/",
+            json={
+                "class_name": "CustomRAGGenerationPrompt",
+                "parameters": {"template": "Q: {input}\nC: {chunks}"},
+            },
+        )
+        assert resp2.status_code == 422, (
+            f"Expected 422 for missing name, got {resp2.status_code}"
+        )
+
+    def test_create_prompt_invalid_class_name(self, client: TestClient):
+        """POST with a non-existent class_name returns 400."""
+        payload = {
+            "class_name": "NonExistentPrompt",
+            "name": "Bad Prompt",
+            "parameters": {"template": "Q: {input} - {chunks}"},
+        }
+        response = client.post("/api/v1/prompt/", json=payload)
+        assert response.status_code == 400, (
+            f"Expected 400 for invalid class_name, got {response.status_code}: {response.text}"
+        )
+
+
+# ===================================================================
+# Prompt Session Integration
+# ===================================================================
+
+
+class TestPromptSessionIntegration:
+    """Prompt usage within generative RAG sessions."""
+
+    def test_session_with_default_prompt_en(self, client: TestClient, test_doc_id: int):
+        """Session creation with DefaultRAGGenerationPrompt (language=en) stores prompt correctly."""
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session EN Prompt"
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultRAGGenerationPrompt", (
+            "Prompt component mismatch"
+        )
+        assert prompt["params"]["language"] == "en", "Language should be en"
+
+    def test_session_with_default_prompt_es(self, client: TestClient, test_doc_id: int):
+        """Session creation with DefaultRAGGenerationPrompt (language=es) stores prompt correctly."""
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session ES Prompt"
+        params["parameters"]["prompt"] = {
+            "component": "DefaultRAGGenerationPrompt",
+            "params": {"language": "es"},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultRAGGenerationPrompt"
+        assert prompt["params"]["language"] == "es", "Language should be es"
+
+    def test_session_with_qna_prompt(self, client: TestClient, test_doc_id: int):
+        """Session creation with DefaultQnARAGGenerationPrompt (language=en) stores prompt correctly."""
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session QnA Prompt"
+        params["parameters"]["prompt"] = {
+            "component": "DefaultQnARAGGenerationPrompt",
+            "params": {"language": "en"},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "DefaultQnARAGGenerationPrompt", (
+            "Prompt component should be DefaultQnARAGGenerationPrompt"
+        )
+        assert prompt["params"]["language"] == "en"
+
+    def test_session_with_custom_prompt_template(self, client: TestClient, test_doc_id: int):
+        """Session creation with CustomRAGGenerationPrompt stores the custom template correctly."""
+        template_text = "Answer the question based on: {chunks}\n\nQuestion: {input}"
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session Custom Prompt"
+        params["parameters"]["prompt"] = {
+            "component": "CustomRAGGenerationPrompt",
+            "params": {"template": template_text},
+        }
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "CustomRAGGenerationPrompt"
+        assert prompt["params"]["template"] == template_text, (
+            "Custom template should be stored exactly as provided"
+        )
+
+    def test_clone_prompt_to_session(self, client: TestClient, test_doc_id: int):
+        """POST /api/v1/prompt/{id}/sessions/{session_id} clones a prompt and attaches it to the session."""
+        # Create a session first
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session For Clone"
+        session_data = _post_and_get(client, params)
+        session_id = session_data["id"]
+
+        # Create a prompt via the prompt API
+        create_payload = {
+            "class_name": "CustomRAGGenerationPrompt",
+            "name": "Prompt To Clone",
+            "parameters": {"template": "Clone: {chunks}\nQ: {input}"},
+        }
+        create_resp = client.post("/api/v1/prompt/", json=create_payload)
+        assert create_resp.status_code == 201, f"Prompt creation failed: {create_resp.text}"
+        prompt_id = create_resp.json()["id"]
+
+        # Clone the prompt to the session
+        clone_resp = client.post(
+            f"/api/v1/prompt/{prompt_id}/sessions/{session_id}",
+            json={},
+        )
+        assert clone_resp.status_code == 201, (
+            f"Clone failed: {clone_resp.status_code} {clone_resp.text}"
+        )
+        clone_data = clone_resp.json()
+        assert "prompt" in clone_data, "Response should contain 'prompt'"
+        assert clone_data["session_id"] == session_id
+        new_prompt_id = clone_data["prompt"]["id"]
+        assert new_prompt_id is not None, "Cloned prompt should have an id"
+        assert new_prompt_id != prompt_id, "Cloned prompt should be a new record"
+
+        # Verify session parameters now reference the cloned prompt
+        session_params = clone_data["parameters"]
+        assert session_params.get("prompt_id") == new_prompt_id, (
+            "Session parameters should reference the cloned prompt ID"
+        )
+
+    def test_session_rejects_invalid_prompt_class(self, client: TestClient, test_doc_id: int):
+        """Session creation accepts unknown component names — validation happens at pipeline runtime.
+
+        The schema `component_field(parent="Prompt")` validates the {component, params}
+        structure but does not verify component existence in the registry at session
+        creation time. The unknown component is stored as-is.
+        """
+        params = _base_session_params(test_doc_id)
+        params["name"] = "Session Invalid Prompt"
+        params["parameters"]["prompt"] = {
+            "component": "NonExistentPrompt",
+            "params": {"language": "en"},
+        }
+        # Session accepts the payload; component resolution happens later at runtime
+        stored = _post_and_get(client, params)
+        prompt = stored["parameters"]["prompt"]
+        assert prompt["component"] == "NonExistentPrompt", (
+            "Unknown component should be stored as-is"
+        )
diff --git a/tests/back/api/test_session_api.py b/tests/back/api/test_session_api.py
index 1572a28eb..40531d60b 100644
--- a/tests/back/api/test_session_api.py
+++ b/tests/back/api/test_session_api.py
@@ -1,6 +1,11 @@
 import pytest
 from fastapi.testclient import TestClient
 
+from DashAI.back.dependencies.database.models import (
+    GenerativeSession,
+    GenerativeSessionParameterHistory,
+)
+
 
 @pytest.fixture(scope="module", name="response_1")
 def create_session_1(client: TestClient):
@@ -160,6 +165,76 @@ def test_get_all_sessions(
     assert response_3.json()["id"] in session_ids, "Session 3 not found in all sessions"
 
 
+def test_update_generative_session_params_merges_and_logs_history(client: TestClient):
+    """Test updating RAG parameters through the dedicated endpoint."""
+    session_factory = client.app.container["session_factory"]
+
+    initial_parameters = {
+        "documents": [1, 2],
+        "chunking_model": {"component": "CharacterChunkModel", "params": {"size": 256}},
+        "retriever_model": {"component": "TFIDFRetriever", "params": {"top_k": 5}},
+        "generation_model": {
+            "component": "QwenModel",
+            "params": {"temperature": 0.7, "max_tokens": 128},
+        },
+        "prompt_id": 1,
+    }
+
+    with session_factory() as db:
+        session = GenerativeSession(
+            task_name="RAGTask",
+            model_name="RAGPipeline",
+            parameters=initial_parameters,
+            name="rag-session-update-test",
+            description=None,
+        )
+        db.add(session)
+        db.commit()
+        db.refresh(session)
+        session_id = session.id
+        original_last_modified = session.last_modified
+
+    response = client.put(
+        f"/api/v1/generative-session/{session_id}/parameters",
+        json={
+            "prompt_id": 2,
+            "generation_model": {
+                "component": "QwenModel",
+                "params": {"temperature": 0.2, "max_tokens": 256},
+            },
+        },
+    )
+
+    assert response.status_code == 200, "Failed to update RAG session parameters"
+    data = response.json()
+    assert data["id"] == session_id
+    assert data["parameters"]["documents"] == [1, 2]
+    assert data["parameters"]["chunking_model"] == {
+        "component": "CharacterChunkModel",
+        "params": {"size": 256},
+    }
+    assert data["parameters"]["generation_model"] == {
+        "component": "QwenModel",
+        "params": {"temperature": 0.2, "max_tokens": 256},
+    }
+    assert data["parameters"]["prompt_id"] == 2
+
+    with session_factory() as db:
+        updated_session = db.get(GenerativeSession, session_id)
+        assert updated_session is not None
+        assert updated_session.parameters["documents"] == [1, 2]
+        assert updated_session.parameters["prompt_id"] == 2
+        assert updated_session.last_modified >= original_last_modified
+
+        history_entries = (
+            db.query(GenerativeSessionParameterHistory)
+            .filter(GenerativeSessionParameterHistory.session_id == session_id)
+            .all()
+        )
+        assert len(history_entries) == 1
+        assert history_entries[0].parameters["prompt_id"] == 2
+
+
 def test_create_session_with_invalid_task(response_4):
     """Test creating a session with an invalid task."""
     assert response_4.status_code == 400
diff --git a/tsconfig.json b/tsconfig.json
new file mode 100644
index 000000000..bc2d6d577
--- /dev/null
+++ b/tsconfig.json
@@ -0,0 +1,15 @@
+{
+    "compilerOptions": {
+        "module": "ESNext",
+        "moduleResolution": "Bundler",
+        "target": "ES2022",
+        "jsx": "react-jsx",
+        "strictNullChecks": true,
+        "strictFunctionTypes": true,
+        "sourceMap": true
+    },
+    "exclude": [
+        "node_modules",
+        "**/node_modules/*"
+    ]
+}
\ No newline at end of file