Refactor documentation for clarity and remove metadata exports

src/app/docs/api/ai-service/page.mdx

@@ -1,8 +1,3 @@
-export const metadata = {
-  title: 'AI Service API - NeuroLab Documentation',
-  description: 'API Reference for the NeuroLab AI and Signal Processing service.',
-}
-
 # AI Service API
 
 The AI Service provides endpoints for EEG analysis, mental state classification, and voice processing.

src/app/docs/api/backend/page.mdx

@@ -1,8 +1,3 @@
-export const metadata = {
-  title: 'Backend API - NeuroLab Documentation',
-  description: 'Current route overview for the NeuroLab Express backend.',
-}
-
 # Backend API
 
 The NeuroLab backend is an Express service that owns authentication, user and doctor workflows, appointments, uploads, reports, billing, notifications, and chat orchestration.

src/app/docs/api/page.mdx

@@ -0,0 +1,49 @@
+import Link from 'next/link'
+import { ArrowRight, Brain, Server, Code2 } from 'lucide-react'
+
+# API Reference
+
+This section groups the two API surfaces documented in this repository:
+
+- the backend API, which handles platform workflows
+- the AI service API, which handles analysis and inference endpoints
+
+If you need a route map first, start with one of the cards below.
+
+<div className='not-prose mt-8 grid gap-4 sm:grid-cols-2'>
+  <Link href='/docs/api/backend' className='group rounded-2xl border border-zinc-200 bg-white p-6 transition-colors hover:bg-zinc-50 dark:border-zinc-800 dark:bg-zinc-950 dark:hover:bg-zinc-900/50'>
+    <Server className='mb-4 size-5 text-zinc-700 dark:text-zinc-300' />
+    <h2 className='flex items-center gap-2 text-xl font-bold'>
+      Backend API
+      <ArrowRight className='size-4 opacity-0 transition-transform group-hover:translate-x-1 group-hover:opacity-100' />
+    </h2>
+    <p className='mt-2 text-sm text-zinc-600 dark:text-zinc-400'>
+      Route groups, operational notes, and the current backend surface.
+    </p>
+  </Link>
+
+  <Link href='/docs/api/ai-service' className='group rounded-2xl border border-zinc-200 bg-white p-6 transition-colors hover:bg-zinc-50 dark:border-zinc-800 dark:bg-zinc-950 dark:hover:bg-zinc-900/50'>
+    <Brain className='mb-4 size-5 text-zinc-700 dark:text-zinc-300' />
+    <h2 className='flex items-center gap-2 text-xl font-bold'>
+      AI Service API
+      <ArrowRight className='size-4 opacity-0 transition-transform group-hover:translate-x-1 group-hover:opacity-100' />
+    </h2>
+    <p className='mt-2 text-sm text-zinc-600 dark:text-zinc-400'>
+      EEG analysis, voice analysis, and model/status endpoints.
+    </p>
+  </Link>
+</div>
+
+## What to expect
+
+<div className='not-prose mt-8 rounded-2xl border border-zinc-200 bg-zinc-50 p-6 dark:border-zinc-800 dark:bg-zinc-950/40'>
+  <div className='flex items-start gap-3'>
+    <Code2 className='mt-0.5 size-5 text-zinc-700 dark:text-zinc-300' />
+    <div className='space-y-2'>
+      <p className='font-semibold text-zinc-950 dark:text-zinc-50'>This section is documentation only.</p>
+      <p className='text-sm leading-7 text-zinc-600 dark:text-zinc-400'>
+        The pages here summarize the API surfaces that the wider NeuroLab platform exposes. They do not imply that the service source code lives in this repository.
+      </p>
+    </div>
+  </div>
+</div>

src/app/docs/architecture/page.mdx

@@ -4,137 +4,113 @@ import {
   Callout,
 } from '@/components/mdx-components'
 
-export const metadata = {
-  title: 'Architecture - NeuroLab Documentation',
-  description: 'Documented runtime architecture for the current NeuroLab codebase.',
-}
-
 # Architecture
 
-NeuroLab currently spans a backend application, an AI service, client applications, and supporting infrastructure. This page focuses on the runtime shape that is visible in the repository today.
+NeuroLab is documented here as a layered system: clients, backend API, AI service, and supporting infrastructure.
 
 <Callout
   type='warning'
   title='Source Of Truth'
   icon='△'
 >
-  This page describes the current documented runtime. If architecture notes elsewhere in the repository disagree, trust the active service entrypoints and route files first.
+  This page summarizes the architecture model used by the docs site. It does not assert that the service source code lives in this repository.
 </Callout>
 
 ## System Overview
 
-The current repository is best understood as four operational layers:
+The platform is described as four operational layers:
 
 <ArchitectureGrid>
   <ArchitectureCard
     eyebrow='Client Layer'
-    title='User-Facing Applications'
-    stack='Vite and Next.js clients'
+    title='Client Applications'
+    stack='User-facing web apps'
     icon='network'
   >
-    `platform/`, `webapp/`, `docs/`, and `ecommerce/` sit at the edge of the
-    system. They consume backend HTTP APIs and expose different product
-    surfaces: platform workflows, user-facing web experiences, and
-    documentation.
+    The docs site sits alongside the platform-facing applications and
+    describes the supported workflows, setup, and reference surfaces.
   </ArchitectureCard>
 
   <ArchitectureCard
     eyebrow='Core Layer'
     title='Backend API'
-    stack='Node.js + Express + MongoDB'
+    stack='Backend service boundary'
     icon='backend'
   >
-    The backend owns authentication, user and doctor workflows, appointments,
-    uploads, reports, notifications, billing, and chat orchestration. It also
-    exposes Swagger at `/api-docs` and acts as the main application boundary
-    for most clients.
+    The backend is the primary HTTP boundary for users, appointments,
+    uploads, reports, billing, notifications, and related workflows.
   </ArchitectureCard>
 
   <ArchitectureCard
     eyebrow='Inference Layer'
     title='AI Service'
-    stack='FastAPI + Python + optional RQ/Redis'
+    stack='Analysis and inference service'
     icon='ai'
   >
-    The active AI runtime lives under `AI Service/backend`. It handles EEG
-    analysis, voice analysis, recommendations, reports, chat workflows, model
-    calibration, and optional streaming. Versioned routes are exposed under
-    `/api/v1/*`.
+    The AI service covers EEG analysis, voice processing, report generation,
+    and model-related endpoints under a versioned API prefix.
   </ArchitectureCard>
 
   <ArchitectureCard
     eyebrow='Infrastructure Layer'
-    title='Data And Messaging'
-    stack='MongoDB, Redis, object storage, MQTT, gRPC'
+    title='Supporting Services'
+    stack='Data, queueing, and storage'
     icon='data'
   >
-    MongoDB is the primary application datastore. Redis supports queue-backed
-    workloads and background processing. MQTT and gRPC exist as integration
-    mechanisms for device and service-to-service communication in specific
-    flows.
+    Supporting services include the datastore, queueing, and object storage
+    layers that the backend and AI service depend on.
   </ArchitectureCard>
 </ArchitectureGrid>
 
 ## Current Service Boundaries
 
 ### Backend Responsibilities
 
-- API surface for `/api/auth`, `/api/users`, `/api/admin`, `/api/doctors`,
-  `/api/devices`, `/api/sessions`, `/api/analysis`, `/api/uploads`,
-  `/api/reports`, `/api/appointments`, `/api/calendar`, `/api/notifications`,
-  `/api/clinics`, `/api/billing`, and `/api/chat`
-- auth and token issuance
+- authentication and token issuance
 - application data persistence
-- appointment, billing, and workflow orchestration
-- chat conversation persistence and AI job reconciliation
+- appointments, uploads, reports, billing, notifications, and chat workflows
+- route groups summarized in the backend API reference
 
 ### AI Service Responsibilities
 
-- versioned FastAPI routing under `/api/v1`
+- versioned API routing
 - EEG analysis endpoints
 - voice analysis endpoints
-- model calibration
-- report generation jobs
-- chat request processing, including queued background mode
+- report generation and model-related workflows
 
 ## Integration Patterns
 
 | Mechanism | Current Role | Notes |
 | :--- | :--- | :--- |
 | **HTTP/REST** | Primary application integration | Main client and service API surface |
-| **Swagger/OpenAPI** | Backend schema discovery | Exposed by the backend at `/api-docs` |
-| **Versioned FastAPI routes** | AI service contract | Canonical AI surface is `/api/v1/*` |
-| **Redis/RQ** | Background processing | Used for queued AI chat and report workflows |
-| **gRPC** | Targeted service-to-service context exchange | Present in current backend and AI chat/context paths |
-| **MQTT** | Device and notification integration | Present as an optional real-time integration mechanism |
+| **OpenAPI/Swagger** | Backend schema discovery | The backend API page points to the generated schema |
+| **Versioned API routes** | AI service contract | The AI service API is documented under a versioned prefix |
+| **Queue-backed jobs** | Background processing | Used for asynchronous analysis and report workflows |
 
 ## Runtime Shape
 
 ```text
 Client applications
-    -> Backend API (Express)
-    -> AI service (FastAPI) for analysis, reports, and chat
-    -> MongoDB / Redis / storage / optional messaging integrations
+    -> Backend API
+    -> AI service for analysis and inference
+    -> Data, queueing, and storage services
 ```
 
 ## Architectural Notes
 
 <Callout title='Important Distinction' icon='◆'>
-  The repository also contains older or alternate architectural traces. For
-  documentation work, treat `AI Service/backend` as the primary AI runtime
-  unless the goal is explicitly historical or comparative.
+  Keep the architecture page aligned with the routes and pages that exist in
+  this docs repository. Avoid adding directory-level claims unless the source
+  code is actually present here.
 </Callout>
 
 ### What is stable
 
-- backend route groups and Swagger exposure
+- backend route summaries and OpenAPI entrypoints
 - AI service versioned API prefix
-- Vite-based `platform` and `webapp` clients
 - docs app as a separate Next.js application
 
 ### What is integration-sensitive
 
-- MQTT-dependent flows
-- gRPC-dependent context exchange
-- Redis-backed async jobs
-- deployment assumptions across the various frontend apps
+- queue-backed async jobs
+- deployment assumptions across the documented setup pages

src/app/docs/concepts/eeg-processing/page.mdx

@@ -1,8 +1,3 @@
-export const metadata = {
-  title: 'EEG Signal Processing - NeuroLab Documentation',
-  description: 'Learn about the core EEG signal processing pipeline, from raw signal ingestion to feature extraction.',
-}
-
 # EEG Signal Processing
 
 NeuroLab employs a state-of-the-art EEG signal processing pipeline designed to handle noisy real-world data and extract meaningful features for mental state classification.
@@ -85,25 +80,25 @@ The extracted features are further prepared for the machine learning models:
 
 ## Next Steps
 
-Learn how these features are combined with voice data for multimodal analysis:
+Continue with the mental state model or the AI service reference:
 
 <div className='not-prose mt-6 grid gap-4 sm:grid-cols-2'>
   <a
-    href='/docs/concepts/voice-emotion'
+    href='/docs/concepts/mental-states'
     className='hover:bg-accent block rounded-lg border p-6 transition-colors'
   >
-    <h3 className='mb-2 font-bold'>Voice Emotion Detection →</h3>
+    <h3 className='mb-2 font-bold'>Mental States →</h3>
     <p className='text-muted-foreground text-sm'>
-      Analyze audio for emotional features
+      See how the processed signals map to the documented labels
     </p>
   </a>
   <a
-    href='/docs/concepts/multimodal'
+    href='/docs/api/ai-service'
     className='hover:bg-accent block rounded-lg border p-6 transition-colors'
   >
-    <h3 className='mb-2 font-bold'>Multimodal Analysis →</h3>
+    <h3 className='mb-2 font-bold'>AI Service API →</h3>
     <p className='text-muted-foreground text-sm'>
-      Combining brain and voice data
+      Review the analysis endpoints that expose the service contract
     </p>
   </a>
 </div>

src/app/docs/concepts/mental-states/page.mdx

@@ -1,8 +1,3 @@
-export const metadata = {
-  title: 'Mental States - NeuroLab Documentation',
-  description: 'Understanding the mental state classification system in NeuroLab.',
-}
-
 # Mental States
 
 NeuroLab's primary goal is to classify human brain activity and emotional markers into actionable mental states. The system currently recognizes three primary states.

src/app/docs/installation/ai-service/page.mdx

@@ -1,10 +1,9 @@
-export const metadata = {
-  title: 'AI Service Setup - NeuroLab Documentation',
-  description: 'Step-by-step guide to installing and configuring the NeuroLab AI Service.',
-}
-
 # AI Service Setup
 
+This page documents the AI service setup flow for the broader NeuroLab
+platform. It assumes you have access to the platform source tree outside this
+docs repository.
+
 The AI Service is the computational heart of NeuroLab, handling EEG signal processing, voice emotion detection, and machine learning inference. It is built with FastAPI and Python.
 
 ## Prerequisites

src/app/docs/installation/backend/page.mdx

@@ -1,10 +1,9 @@
-export const metadata = {
-  title: 'Backend Setup - NeuroLab Documentation',
-  description: 'Step-by-step guide to installing and configuring the NeuroLab Express.js backend.',
-}
-
 # Backend Setup
 
+This page documents the backend setup flow for the broader NeuroLab platform.
+It assumes you are working from the platform source tree rather than this docs
+repository alone.
+
 The Backend service provides the core API for NeuroLab, managing users, clinical data, billing, and orchestrating analysis sessions. It is built with Node.js and Express.
 
 ## Prerequisites

src/app/docs/installation/docker/page.mdx

@@ -1,10 +1,8 @@
-export const metadata = {
-  title: 'Docker Setup - NeuroLab Documentation',
-  description: 'Learn how to deploy the entire NeuroLab platform using Docker and Docker Compose.',
-}
-
 # Docker Setup
 
+This page assumes you are deploying the broader NeuroLab platform, not just
+this docs repository.
+
 Docker is the recommended way to deploy NeuroLab in both development and production environments. It ensures consistency across different systems and simplifies the management of multiple services and databases.
 
 ## Prerequisites

src/app/docs/installation/production/page.mdx

@@ -1,10 +1,8 @@
-export const metadata = {
-  title: 'Production Deployment - NeuroLab Documentation',
-  description: 'Learn how to deploy the NeuroLab platform to production environments using Render, Vercel, and managed databases.',
-}
-
 # Production Deployment
 
+This page assumes you are deploying the broader NeuroLab platform. It does not
+describe infrastructure contained in this docs repository itself.
+
 Deploying NeuroLab to production requires a distributed approach to ensure scalability, security, and reliability. We recommend using managed services to reduce operational overhead.
 
 ## Deployment Architecture