🤖 Agentic RAG System

A comprehensive Agentic RAG (Retrieval-Augmented Generation) system built with CrewAI that demonstrates multi-agent collaboration for business intelligence and document analysis.

✨ Key Features

• 🗄️ ChromaDB Integration: Document storage and retrieval with persistent vector database
• 🔍 Web Search Capabilities: Real-time web search using SerperDevTool
• 🤝 Multi-Agent System: Specialized agents for different tasks (RAG, Web Search, Code Execution)
• ⚡ Fast LLM Inference: Supports Groq and Gemini APIs for high-speed processing
• 📊 Business Intelligence: Automated analysis and visualization generation
• 🏗️ Hierarchical Process: Manager agent coordinates specialized workers
• 🔧 Flexible Configuration: Easy switching between LLM providers

📋 View Detailed Architecture Specification

Complete technical specification of the multi-agent architecture with ChromaDB integration, web scraping capabilities, and hierarchical task coordination for comprehensive business intelligence analysis.

🚀 Quick Start

Prerequisites

• Python >=3.10, <=3.13
• UV for dependency management

Installation

1. Clone the repository:

   git clone <your-repo-url>
   cd agentic-rag-practical-example

2. Install UV:

   pip install uv

3. Install dependencies:

   crewai install

Configuration

1. Copy environment template:

   cp .env.example .env

2. Add your API keys to .env:

   SERPER_API_KEY=your_serper_api_key_here
   GROQ_API_KEY=your_groq_api_key_here
   LLM=groq/llama3-70b-8192
   OPENAI_API_KEY=your_groq_api_key_here  # Used for CrewAI validation

3. Load documents into ChromaDB (optional):

   python src/agentic_rag/load_chroma_docs.py

Running the Project

crewai run

This will:

• Analyze business documents using RAG
• Generate comprehensive business trends report
• Create visualization code for data insights
• Output results to outputs/ directory

🏗️ Architecture

Agents

• 📚 Document RAG Agent: Retrieves information from ChromaDB vector database
• 🌐 Web Agent: Performs real-time web searches using SerperDevTool
• 💻 Code Execution Agent: Generates and executes analysis code

Tasks

• 📋 Fetch Tax Docs: Retrieves relevant documents from internal database
• ❓ Answer Question: Provides detailed answers using RAG
• 📈 Business Trends: Analyzes trends and generates insights
• 📊 Graph Visualization: Creates data visualization code

Tools

• ChromaDBTool: Custom tool for vector database operations
• SerperDevTool: Web search capabilities
• ScrapeWebsiteTool: Website content scraping and analysis

📁 Project Structure

agentic-rag-practical-example/
├── src/agentic_rag/
│   ├── config/
│   │   ├── agents.yaml          # Agent configurations
│   │   └── tasks.yaml           # Task definitions
│   ├── tools/
│   │   └── chromadb_tool.py     # Custom ChromaDB tool
│   ├── crew.py                  # Main crew definition
│   ├── main.py                  # Entry point
│   └── load_chroma_docs.py      # Document loader
├── internal_docs/               # Sample documents
├── db/                         # ChromaDB storage
├── outputs/                    # Generated reports
├── .env.example               # Environment template
└── README.md

🔧 Configuration Options

Switching LLM Providers

The system supports multiple LLM providers. Update your .env file:

For Groq:

LLM=groq/llama3-70b-8192
GROQ_API_KEY=your_groq_api_key
OPENAI_API_KEY=your_groq_api_key

For Gemini:

LLM=gemini/gemini-1.5-pro
GEMINI_API_KEY=your_gemini_api_key
OPENAI_API_KEY=your_gemini_api_key

Customizing Queries

Edit src/agentic_rag/main.py to change the analysis query:

inputs = {
    "query": "Your custom business question here",
    "company": "Your Company Name",
    "company_description": "Brief company description",
}

📊 Output Examples

The system generates:

1. Business Trends Report (outputs/business_trends.md)

   • Financial analysis over multiple years
   • Revenue and expense trends
   • Profitability insights

2. Visualization Code (outputs/visualize.ipynb)

   • Python matplotlib scripts
   • Interactive data visualizations
   • Chart generation code

🛠️ Development

Adding Custom Tools

1. Create a new tool in src/agentic_rag/tools/
2. Import and add to agent configuration in crew.py
3. Update agent YAML configurations as needed

Training the Crew

crewai train <n_iterations> <filename>

Testing

crewai test <n_iterations> <model_name>

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

🙏 Acknowledgments

• CrewAI for the multi-agent framework
• ChromaDB for vector database capabilities
• Groq for fast LLM inference