RAG AI Agent for SQL Database

A powerful local RAG (Retrieval-Augmented Generation) AI agent that enables natural language querying of SQL databases using LangChain and Ollama.

Architecture

RAG AI Agent
├── Database Connection Layer
├── Schema Analysis & Vectorization
├── Query Processing Engine
├── LangChain RAG Pipeline
├── Ollama LLM Integration
└── User Interface

Quick Start

1. Install dependencies:

   pip install -r requirements.txt

2. Configure your database connection in config/database.yaml

3. Start Ollama service:

   ollama serve

4. Pull required models:

   ollama pull llama3.1
   ollama pull nomic-embed-text

5. Initialize the database schema:

   python scripts/initialize_schema.py

6. Run the agent:

   python main.py

Usage Examples

• "What are the sales last month?"
• "Show me the top 10 customers by revenue"
• "What's the average order value this quarter?"
• "Which products have the highest profit margins?"

Project Structure

RAG/
├── src/
│   ├── agents/          # AI agent implementations
│   ├── database/        # Database connection and operations
│   ├── embeddings/      # Vector embeddings and storage
│   ├── llm/            # LLM integration (Ollama)
│   ├── rag/            # RAG pipeline components
│   └── utils/          # Utility functions
├── config/             # Configuration files
├── data/              # Sample data and schemas
├── scripts/           # Setup and utility scripts
├── tests/             # Test files
└── ui/                # User interface components

Configuration

See config/README.md for detailed configuration options.

License

MIT License