RAGentrix is an intelligent conversational platform that combines the power of Retrieval-Augmented Generation (RAG) with multi-agent architecture. Built on a modern tech stack including LangGraph, FastAPI, Next.js, and ChromaDB, it provides a robust foundation for developing sophisticated AI-powered chat applications.

• Multi-Agent Framework: Leverages LangGraph for complex agent orchestration and workflow management
• Knowledge Integration: ChromaDB-powered vector storage for efficient document retrieval and semantic search
• Real-time Communication: Server-Sent Events (SSE) for seamless streaming responses
• Scalable Backend: FastAPI with full asynchronous support for high-performance operations
• Modern Frontend: Next.js-based interface with responsive design and intuitive UX

• Custom Agent Development: Create specialized agents with unique behavioral patterns
• Tool Integration: Equip agents with external tools and APIs for extended functionality
• Collaborative Workflows: Enable multiple agents to work together on complex tasks

• Document Processing: Automatic ingestion and indexing of various document formats
• Semantic Search: Advanced vector similarity matching for precise information retrieval
• Context-Aware Responses: Generate answers based on retrieved knowledge and conversation history

• Streaming Chat: Real-time message delivery with typing indicators
• Session Persistence: Browser-based chat history for continuity across sessions
• Multi-Language Support: Built-in capabilities for English and Chinese interactions

RAGentrix/
backend/                 # Python backend services
  app/
    api/                # REST API endpoints
    ai/                 # AI agent implementations
    core/               # Core application logic
    db/                 # Database models and configurations
  resource/             # Static resources and data
frontend/               # Next.js frontend application
  app/
    chat/               # Chat interface components
    components/         # Reusable UI components

• Python 3.13+
• Node.js 18+
• Ollama (for local embedding models)

1. Environment Configuration

   cd backend
   cp .env.example .env

2. Install Dependencies

   pip install uv
   uv sync --frozen
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Configure Environment Variables

   # Database
   DATABASE_URL=sqlite+aiosqlite:///resource/database.db
   
   # Application
   APP_NAME=RAGentrix
   DEBUG=True
   
   # AI Models
   OPENAI_API_KEY=your_openai_key
   DEFAULT_MODEL=gpt-4o-mini
   
   # Embeddings (requires Ollama)
   EMBEDDING_MODEL=bge-m3
   CHROMA_PATH=resource/chroma_db

4. Start Server

   python app/run_server.py

1. Install Dependencies

   cd frontend
   pnpm install

2. Launch Application

   pnpm dev

3. Access Interface Open http://localhost:3000 in your browser

For local document processing, deploy the BGE-M3 embedding model:

ollama pull bge-m3

• Purpose: Office automation and employee information management
• Features: Employee database queries, handbook document retrieval
• Implementation: backend/app/ai/agent/oa_assistant.py

• Math Agent: Specialized in mathematical computations and analysis
• Code Agent: Handles code generation, debugging, and optimization
• General Agent: Manages general-purpose conversations and tasks
• Supervisor: Coordinates agent interactions and task distribution
• Implementation: backend/app/ai/agent/multi_agent.py

1. Define agent logic in backend/app/ai/agent/
2. Implement LangGraph workflow for agent behavior
3. Register agent in the frontend interface
4. Test agent interactions through the chat interface

1. Add documents to backend/resource/ directory
2. Update ChromaDB indexing scripts
3. Configure document processing parameters
4. Test retrieval accuracy through queries

Backend:

• FastAPI (web framework)
• LangGraph (agent orchestration)
• ChromaDB (vector database)
• SQLModel (database ORM)
• Pydantic (data validation)

Frontend:

• Next.js (React framework)
• Ant Design (UI components)
• TypeScript (type safety)
• Tailwind CSS (styling)

AI/ML:

• OpenAI/DeepSeek/DashScope (LLM providers)
• BGE-M3 (embeddings)
• LangChain (AI framework)

1. Fork the repository
2. Create a feature branch
3. Implement your changes
4. Add tests and documentation
5. Submit a pull request

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