quilllearning/ncc-ai-chatbot

The project is a self‑hosted AI chatbot that powers a documentation search (RAG) for the Elysia ecosystem, using vector and BM25 indexes stored in Postgres, semantic caching with Dragonfly, and locally hosted LLMs for answering questions. It’s aimed at developers and maintainers of the Elysia project who want a low‑cost, fully customizable alternative to third‑party API‑based RAG services that can be integrated with their existing Vitepress documentation site.