basjoo
Self-hosted AI customer-support platform — per-KB RAG on Qdrant, 13 LLM providers, SSE streaming, embeddable chat widget. FastAPI + Vite + SQLite, all in Docker
What is basjoo?
Self-hosted AI customer-support platform — per-KB RAG on Qdrant, 13 LLM providers, SSE streaming, embeddable chat widget. FastAPI + Vite + SQLite, all in Docker Compose.
Overview of basjoo
What it is
Basjoo is a self-hosted AI customer-support platform. It turns a website plus a set of documents into a support agent that can be embedded anywhere. It provides a console for analytics, session transcripts, human takeover, a Q&A index, and lead capture. The platform uses per-knowledge-base RAG on Qdrant, supports 13 LLM providers, SSE streaming, and an embeddable chat widget. The backend is built with FastAPI, the frontend console with Vite, and data is stored in SQLite. Everything runs in Docker Compose.
Key features
- Per-knowledge-base RAG retrieval using Qdrant vector store - Support for 13 LLM providers - SSE streaming for real-time chat - Embeddable chat widget with a single `` tag - Console with analytics, session transcripts, human takeover, Q&A index, and lead capture - Self-hosted; all components run in Docker Compose
Use cases
- Turning a website and documentation into an AI-powered customer support agent - Embedding a support chat widget on any website - Managing and analyzing customer support interactions with a self-hosted console - Implementing human takeover when the AI cannot resolve an issue - Capturing leads and tracking unanswered questions
Getting started
bash curl -fsSL https://raw.githubusercontent.com/haoyiyin/basjoo/main/install-deploy.sh | sudo sh One command on a fresh Ubuntu/Debian host: installs Docker if missing, clones the repo, starts the prod profile, waits for health checks, and prints the console URL. The first registered account becomes the workspace super administrator. For development, see the repository README for backend, frontend, and widget setup commands.
License and pricing
The project is licensed under MIT and is open source.
Based on the project's public README and repository metadata.