Multi-Agent Chatbot with LangGraph and Azure Services

A sophisticated chatbot implementation that uses multiple specialized agents to process queries through different search and processing methods, powered by Azure OpenAI, Azure AI Search, and Azure SQL Database.

Features

• Query decomposition for complex questions
• Intelligent intent routing between different search methods
• Multiple search capabilities:
  • Keyword-based search using Azure AI Search
  • Semantic search using FAISS vector store
  • SQL database queries
  • Direct LLM processing
• Answer generation combining multiple data sources
• Web interface for easy interaction

Prerequisites

• Python 3.8+

• Azure OpenAI API accessgit config --global user.name "Your GitHub Username" git config --global user.email "[email protected]"

• Azure AI Search instance

• Azure SQL Database

• FAISS index for vector search

Installation

1. Clone the repository:

git clone <repository-url>
cd multi-agent-chatbot-langgraph

Copy

Insert at cursor
markdown
Install required packages:

pip install pyodbc flask python-dotenv langgraph langchain openai faiss-cpu

Copy

Insert at cursor
bash
Create a .env file in the project root with the following variables:

AZURE_OPENAI_DEPLOYMENT_NAME=<your-deployment-name>
AZURE_OPENAI_API_BASE=<your-api-base>
AZURE_OPENAI_API_KEY=<your-api-key>

AZURE_SEARCH_INDEX_NAME=<your-index-name>
AZURE_SEARCH_SERVICE_NAME=<your-service-name>
AZURE_SEARCH_API_KEY=<your-search-api-key>

AZURE_SQL_SERVER=<your-sql-server>
AZURE_SQL_DATABASE=<your-database>
AZURE_SQL_USERNAME=<your-username>
AZURE_SQL_PASSWORD=<your-password>

FAISS_INDEX_PATH=<path-to-faiss-index>

Copy

Insert at cursor
env
Project Structure
├── app.py                 # Main application file
├── templates/
│   └── index.html        # Web interface template
├── .env                  # Environment variables
└── README.md            # Project documentation

Copy

Insert at cursor
text
How It Works
Query Decomposition : Breaks down complex queries into simpler sub-queries

Intent Router : Determines the most appropriate search method

Search Agents :

Index Search: Uses Azure AI Search for keyword-based search

Vector Search: Uses FAISS for semantic similarity search

SQL Query: Executes database queries

LLM Processing: Direct language model processing

Answer Generation : Combines and refines results from different sources

Usage
Start the Flask application:

python app.py

Copy

Insert at cursor
bash
Open a web browser and navigate to http://localhost:5000

Enter your query in the text input and click "Submit"

Architecture
The application uses LangGraph to orchestrate multiple agents:

Query Decomposition Agent

Intent Router Agent

Index Search Agent

Vector Search Agent

SQL Query Agent

Answer Generation Agent

Each agent specializes in a specific task and communicates through the graph structure.

Security Notes
All sensitive credentials are stored in environment variables

Database connections are properly managed and closed

API keys and connection strings should be kept secure

Web interface uses proper input sanitization

Contributing
Fork the repository

Create a feature branch

Commit your changes

Push to the branch

Create a Pull Request

License
Add your chosen license here


This README provides a comprehensive overview of your project, including:
- What the project does
- How to set it up
- Required dependencies
- Configuration instructions
- Project structure
- How the system works
- Usage instructions
- Security considerations
- Contributing guidelines

You should customize this template by:
1. Adding the actual repository URL
2. Choosing and adding an appropriate license
3. Adding any specific deployment instructions for your environment
4. Updating any specific requirements or configurations unique to your implementation

Would you like me to explain or modify