Building a Viral Open Source AI Chatbot: Journey from Concept to Reality
summary
An open source chatbot called Quivr goes viral in about two months, claiming 14k stars and 6.5k active users, built as a document-embedding RAG assistant with local-model support. Backend uses Docker/Fargate, Supabase/Postgres with PGVector, LangChain, Streamlit prototype; team spins off genos GPT and codery GPT and a custom non‑deterministic testing framework. Early wins include a free hosted demo and heavy GitHub traffic; failures include frequent site crashes, tight Superbase coupling, LLM API sprawl and accumulated tech debt now being fixed as UX work.
AI-written summary · about
chapters (11)
- 0:00Introduction and background
- 1:05What is Quiver and its features
- 3:32Open source challenges
- 5:14Balancing user feedback
- 7:40Knowing your product
- 10:29Unexpected growth and tech debt
- 12:55Understanding embeddings and RAG
- 14:18Lack of GenAI tooling
- 17:29Testing generative models
- 20:00Current product and UX focus
- 21:47Future of GenAI and takeaways
related talks
51mEvent Driven AI: Supercharging ChatGPT with RAG & LangStream
28mWhat the RAG? Demystifying Retrieval Augemented Generation
32mChatBot Learning Assistant with Large Language Models
36mGENAI-Powered chatbot to increase team efficiency
48mUnlocking LLM Potential: From Exploration to Integration
41mFrom RAGs to riches: Build an AI document interrogation app in 30 mins
39mBuilding Your Own Bot with GPT: Harnessing the Power of AI...
37mOpen-Source Generative AI: A Product Manager’s Blueprint