RAGapp
A self-hosted RAG app built on LlamaIndex. Run it with Docker, use local models through Ollama, or connect to hosted OpenAI and Gemini models.
RAGapp is a self-hosted app for teams that want an AI assistant using retrieval-augmented generation (RAG) on their own infrastructure. It pairs a browser chat interface with an admin interface for configuring the assistant, taking an approach similar to OpenAI's custom GPTs. It's open source under the Apache 2.0 license.
The app runs in Docker on your own hardware or cloud infrastructure. You can use local models through Ollama or choose hosted models from OpenAI and Gemini. That choice matters for where the AI runs: Ollama provides local model execution, while the OpenAI and Gemini options rely on external model services. Hosting the app yourself doesn't make those cloud models local.
Built on LlamaIndex, RAGapp includes an API alongside its chat interface, so teams can use the assistant directly in a browser or connect it to another application. Its Docker Compose deployments cover an Ollama and Qdrant setup, as well as multiple RAGapp instances with a shared management interface.
Authentication is a separate concern. The standalone container has no built-in authentication layer; it relies on an API gateway to control incoming access. Teams choosing it for an internal service need to provide that access layer as part of their deployment.
