RAG Chatbot That Answers From Your Docs
Rishi Barapatre
React RAG Copilot
An AI assistant that answers questions from your documents and shows the source for every answer. Runs without a GPU, so there are no expensive servers to pay for.
Demo knowledge base: React documentation. The same system can be pointed at any business's docs, FAQs, manuals, or policies.
The problem
Teams lose hours searching long documentation, and generic chatbots make answers up. This project answers only from the documents you give it and cites where each answer came from.
What it does
Answers with sources. Every reply points to the document section it came from, so answers can be checked.
Finds the right passage first. A two-step search pulls 15 candidate passages, then a second model keeps the best 3. This cuts irrelevant context and reduces made-up answers.
Runs on a normal CPU. No GPU or cloud infrastructure needed, which keeps running costs low.
Plugs in two ways. A REST API for websites and apps, and an MCP server so AI assistants like Claude Desktop and Cursor can query the docs directly.
Quality is measured. Scored with RAGAS (Faithfulness and Context Precision) instead of "it seems to work."
Ready to ship. Dockerized, with 15 automated tests.
Where this fits
Customer support bot answering from your help center
Internal assistant for policies, SOPs, or onboarding docs
Product or technical documentation search
Hinglish/Hindi/English document Q&A for Indian businesses
Built with
Python, LlamaIndex, ChromaDB, FlashRank, FastAPI, FastMCP, Groq, RAGAS, Docker
Want something like this for your business?
I build document chatbots and AI assistants for small businesses. Contact: rishibarapatre@gmail.com
Like this project
Posted Oct 9, 2026
Ask questions, get answers from your own docs, with sources. Built with RAG, runs without a GPU, and quality-tested with RAGAS. Demo on React docs.
