rag-supportbot is a local, document-based AI assistant designed to help technical support teams answer user questions based on existing documents.
The system is built using the Retrieval-Augmented Generation (RAG) architecture, which enables reliable and source-grounded responses by retrieving relevant document chunks and generating answers using a local large language model.
- Offline operation with no external API dependencies
- Source-based, grounded answer generation
- Local vector search using FAISS and LlamaIndex
- Modular architecture for future extensions
- Python 3.11
- FAISS
- LlamaIndex
- LangChain (optional)
- LLaMA 3 Instruct (via
llama.cpp) - Streamlit (for UI)
MIT License
This project is in active development. The architecture and components may change as the system evolves.