kotaemon: An Open-Source RAG Tool for Document Chat

kotaemon: An Open-Source RAG Tool for Document Chat

Summary

kotaemon is an open-source, RAG-based tool designed to facilitate interactive conversations with your documents. It provides a clean and customizable UI, catering to both end-users seeking document Q&A and developers building RAG pipelines.

Repository Info

Updated on November 22, 2025
View on GitHub

Tags

chatbotllmsopen-sourceragPythondocument-qaai-tools

Click on any tag to explore related repositories

Introduction

kotaemon is an open-source, RAG-based tool designed to facilitate interactive conversations with your documents. It offers a clean and customizable UI, built for both end-users who want to perform QA on their documents and developers looking to build their own RAG pipelines. This project aims to provide a functional RAG UI, supporting various LLMs, easy installation, and a robust framework for RAG pipeline development.

For end-users, kotaemon provides a clean and minimalistic UI for RAG-based QA, compatibility with various LLM API providers (OpenAI, AzureOpenAI, Cohere, etc.), and local LLMs (via ollama and llama-cpp-python), along with easy installation. Developers benefit from a comprehensive framework for building RAG pipelines, a customizable UI built with Gradio, and a dedicated Gradio theme.

Key features include hosting your own document QA web-UI with multi-user login and file organization, support for various LLM and embedding models, a hybrid RAG pipeline for optimal retrieval quality, multi-modal QA support for documents with figures and tables, and advanced citations with in-browser document preview. The system also supports complex reasoning methods like question decomposition and agent-based reasoning (ReAct, ReWOO), and offers a configurable settings UI.

Installation

To get started with kotaemon, ensure you meet the system requirements: Python >= 3.10. Docker is optional but recommended for an easier setup. For processing additional file types beyond .pdf, .html, .mhtml, and .xlsx, you may need to install Unstructured.

With Docker (Recommended)

kotaemon provides lite, full, and ollama Docker images. The full version includes unstructured for broader file type support, while ollama bundles Ollama for local RAG.

docker run \
-e GRADIO_SERVER_NAME=0.0.0.0 \
-e GRADIO_SERVER_PORT=7860 \
-v ./ktem_app_data:/app/ktem_app_data \
-p 7860:7860 -it --rm \
ghcr.io/cinnamon/kotaemon:main-lite

Access the WebUI at http://localhost:7860/. You can specify the platform (e.g., --platform linux/arm64) if needed.

Without Docker

  1. Clone the repository and install required packages:
git clone https://github.com/Cinnamon/kotaemon
cd kotaemon
pip install -e "libs/kotaemon[all]"
pip install -e "libs/ktem"
  1. Create a .env file based on .env.example in the project root for initial model configuration.
  2. (Optional) For in-browser PDF viewer, download and extract PDF_JS_DIST to libs/ktem/ktem/assets/prebuilt.
  3. Start the web server:
python app.py

The app will launch in your browser. Default login is admin/admin.

Examples

kotaemon offers several ways to experience its capabilities:

  • Live Demos: Explore interactive versions on Hugging Face Spaces:

  • Local RAG with Colab: Try a local RAG setup using the provided Colab Notebook.

  • Visual Previews: The repository includes preview images showcasing the user interface, such as the chat tab and advanced citation features with in-browser PDF viewer.

Why Use kotaemon?

kotaemon stands out as a powerful and flexible solution for document Q&A due to several compelling reasons:

  • Open-Source & Community-Driven: Being open-source, it fosters transparency, community contributions, and continuous improvement.

  • Comprehensive RAG Capabilities: It provides a robust, hybrid RAG pipeline with full-text and vector retrieval, re-ranking, and support for complex reasoning methods like question decomposition and agent-based approaches (ReAct, ReWOO).

  • Versatile LLM Support: Seamlessly integrate with popular LLM API providers (OpenAI, AzureOpenAI, Cohere) or leverage local models via Ollama and llama-cpp-python for private RAG solutions.

  • User-Friendly & Customizable UI: The clean and minimalistic Gradio-based UI is intuitive for end-users, while its extensibility allows developers to customize or add new UI elements and integrate custom RAG pipelines.

  • Advanced Features: Benefit from multi-modal QA support (figures, tables), detailed citations with in-browser PDF preview and relevance scoring, and configurable settings to fine-tune retrieval and generation processes.

  • Easy Deployment: With Docker support and clear installation guides, setting up your own document QA web-UI is straightforward.

Links