A web-based system for creating, training, and interacting with custom RAG-powered AI agents.
This project aims to build a complete system where users can define custom AI "agents" with specific purposes. Users can upload documents (like PDFs, TXT, etc.) to provide a knowledge base for each agent. The system will then use a Retrieval-Augmented Generation (RAG) pipeline to allow users to chat with their agents, receiving answers grounded in the provided documents.
- Backend: Go
- Frontend: React
- Vector Database: Weaviate
- Persistence Database: MongoDB
This project is in the initial setup phase. The foundational structure for the Go backend is currently being built.
- Set up Go backend server skeleton & API structure.
- Implement document ingestion pipeline (parsing, chunking, embedding).
- Develop the core RAG chat endpoint.
- Integrate MongoDB for persistence and Weaviate for vector storage.
- Build the React frontend for agent and chat management.