RasaGPT

Why we included this project

Rasa remains a solid choice for dialogue management, but wiring it to a modern LLM stack is exactly the kind of integration work that eats weekends. RasaGPT is a headless reference implementation that already did that glue work: a FastAPI backend, Langchain and LlamaIndex for indexing and retrieval, pgvector for vector storage, and Telegram as the chat channel, with document upload and context injection built in. The author also worked out the fiddly parts most projects discover too late, like multi-tenancy, session handling, and passing metadata between Rasa and your own backend, plus a Dockerized setup. If you already build on Rasa and want LLM-powered answers behind existing dialogue flows, this is a strong starting point to adapt rather than a finished product to deploy as-is.