kotaemon
An open-source document chat app for Windows, macOS and Linux. Use local models through Ollama or llama-cpp-python, or connect cloud APIs.
kotaemon is a self-hosted document chat app for people who want to ask questions across their files and check where the answers came from. It runs in a browser on Windows, macOS or Linux, with Docker also supported. The project uses the Apache 2.0 license.
You can use local LLMs through Ollama or llama-cpp-python, or connect services such as OpenAI, Azure OpenAI, Cohere and Groq. Local models and local document parsers support private processing on your own hardware. Choosing cloud models or parsing APIs sends the corresponding work to those services. A Hugging Face Space provides an online option, separate from running the app yourself.
Answers include citations that you can inspect in an in-browser PDF viewer, with highlights and relevance scores. The app also warns when retrieved material has low relevance. Its document search combines full-text and vector retrieval with re-ranking, and questions can span multiple documents, including figures and tables.
For shared use, kotaemon has multi-user login, private and public file collections, and chat sharing. Document parsing options include local Docling and PaddleOCR, alongside Azure Document Intelligence and Adobe PDF Extract APIs.
Developers can extend the Python app's Gradio interface and supply their own document indexing or reasoning pipelines. It supports question decomposition for multi-step questions, agent reasoning through ReAct and ReWOO, and GraphRAG indexing.
