cinnamon/kotaemon

kotaemon is an open-source, self-hosted chat interface that lets you have conversations with your own documents using AI. It solves the problem of needing to search through large collections of PDFs, reports, or other files manually, instead, you upload your documents and ask questions in plain language, and the AI finds relevant passages and answers you. The technology behind it is called RAG, which stands for Retrieval-Augmented Generation. This means the AI doesn't just rely on its training knowledge; it first searches your uploaded documents to find relevant sections, then uses that retrieved content to generate an accurate, cited answer. kotaemon uses a hybrid retrieval approach, combining traditional keyword search with semantic (meaning-based) vector search, to improve the quality of what it finds. Answers come with citations, and you can see exactly which passages were used, highlighted directly in a built-in PDF viewer. The tool supports multiple AI providers, including OpenAI, Azure, Groq, and locally-run models via Ollama, and handles images, tables, and complex multi-step questions. It has a multi-user login system, supports private and shared document collections, and is built on Gradio (a Python framework for building web UIs). You can run it with Docker for the easiest setup. You would use kotaemon if you are a researcher, lawyer, analyst, or any knowledge worker who needs to quickly extract information from large document collections. The tech stack is Python.