RAG AI Agent

A chat app for asking questions about your own documents. Upload PDFs, ask anything, and every answer comes with citations to its sources, so you can check its work.

The backend is FastAPI with a LangChain pipeline: documents are embedded into a FAISS vector store for semantic retrieval, and Google Gemini writes the answers. Uploading a new PDF updates the vector store automatically. The frontend is a real-time chat interface built with Next.js and Tailwind CSS.