Skip to content

QUICK_START.md

Latest commit

 

History

History
198 lines (151 loc) · 4.65 KB

QUICK_START.md

File metadata and controls

198 lines (151 loc) · 4.65 KB
Raw

Quick Start Guide - AI Insurance Assistant

Get up and running with the LangChain-powered AI assistant in 5 minutes.

Prerequisites

  • Python 3.9+
  • PostgreSQL database access (Supabase)
  • OpenAI API key

Installation

# Navigate to the ai_assistant folder
cd ai_assistant

# Install dependencies
pip install -r requirements.txt

Configuration

Create a .env file:

# Required
DATABASE_URL=postgresql://your-connection-string
OPENAI_API_KEY=sk-your-key-here

# Optional (defaults shown)
LLM_MODEL=gpt-4o-mini
LLM_TEMPERATURE_SQL=0.1
LLM_TEMPERATURE_RESPONSE=0.7
MEMORY_WINDOW_SIZE=3
MAX_RETRIES=3

Run the App

streamlit run ui/app.py

Open http://localhost:8501 in your browser.

Test Individual Components

Test Database Connection

python -c "from config.database import test_connection; print('✅ Connected!' if test_connection() else '❌ Failed')"

Test SQL Generation

from agents.sql_agent import LangChainSQLAgent

agent = LangChainSQLAgent(company_id=29447)
result = agent.generate_sql("What are the contact details?")
print(f"SQL: {result['sql']}")

Test Conversation Memory

from memory.conversation_memory import ConversationMemoryManager

memory = ConversationMemoryManager()
memory.add_exchange("test_session", "What's the email?", "contact@example.com")
memory.add_exchange("test_session", "What about phone?", "+1234567890")

history = memory.get_conversation_history("test_session")
print(f"History: {history}")

Test Full Execution

from core.executor import execute_with_retry

result = execute_with_retry(
    user_question="What are the contact details?",
    company_id=29447
)

print(f"Success: {result['success']}")
print(f"SQL: {result['sql']}")
print(f"Response: {result['natural_response']}")

Example Queries

Try these in the UI:

  1. Phone Calls

    • "What was the last phone call?"
    • "Show me recent calls"
    • "Any voicemails?"

  2. Phone Messages (SMS)

    • "Show me text messages"
    • "Any SMS from the client?"

  3. Email Communications

    • "What quotes were sent?"
    • "What's going on with this account?"
    • "Show me recent emails"

  4. Company Data

    • "What are the contact details?"
    • "What's the email address?"
    • "How many employees do they have?"

  5. General Queries

    • "Show me all activity from last month"
    • "What happened recently?"

Architecture Overview

┌─────────────────┐
│   Streamlit UI  │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│  LangChain      │
│  SQL Agent      │
├─────────────────┤
│ • Skill Router  │
│ • SQL Chain     │
│ • Response Chain│
│ • Memory Mgr    │
└────────┬────────┘
         │
         ▼
┌─────────────────┐
│   Executor      │
├─────────────────┤
│ • Validator     │
│ • PostgreSQL    │
│ • Retry Logic   │
└─────────────────┘

Key Files

  • agents/sql_agent.py - Main LangChain agent orchestrator
  • chains/sql_generation_chain.py - SQL generation chain
  • chains/response_chain.py - Response generation chain
  • memory/conversation_memory.py - LangChain memory manager
  • core/executor.py - SQL execution with retry
  • ui/app.py - Streamlit interface

Troubleshooting

ImportError: No module named 'langchain'

pip install -r requirements.txt

Database connection failed

  • Check DATABASE_URL in .env
  • Test with: python -c "from config.database import test_connection; print(test_connection())"

LangChain memory not working

  • Ensure memory_manager is in st.session_state
  • Check session_id is consistent across requests

SQL validation errors

  • Review error message in UI debug info
  • Check that query includes company_id filter
  • Ensure only SELECT queries (no INSERT/UPDATE/DELETE)

LangSmith Tracing (Optional)

Enable detailed tracing:

# In .env
LANGCHAIN_TRACING_V2=true
LANGCHAIN_API_KEY=ls-your-key
LANGCHAIN_PROJECT=ai-assistant

View traces at https://smith.langchain.com

Next Steps

  1. ✅ Get the app running
  2. Try example queries
  3. Review the architecture in README.md
  4. Customize skills for your use case
  5. Enable LangSmith for production monitoring
  6. Explore advanced LangChain features

Support