Kaveon documentation
Kaveon is a self-hosted analytics platform — query live data, build charts, assemble dashboards, and ask questions in plain English. This is the reference for using and operating it.
What Kaveon is
Kaveon sits between you and your data. Point it at a warehouse or database and you get a private analytics workspace: a full SQL Lab, a drag-and-drop chart builder, composable dashboards, reusable semantic datasets, and a DLM (Data Language Model) that turns plain-English questions into charts with no hosted LLM — answering the common ones straight from precomputed context. It is open source, MIT-licensed, and runs entirely on infrastructure you control.
Data sources are configurable from the UI — no config-file edits, no restarts to add one. Authentication providers (GitHub, Google, Microsoft Entra) activate automatically when their env vars are set.
How it fits together
Kaveon is a small monorepo with two services and a data layer:
| Piece | Stack | Responsibility |
|---|---|---|
| kaveon-web | Next.js 15 · TypeScript · React 19 | The UI — SQL Lab, chart & dashboard builders, AI chat. Handles sign-in (NextAuth) and proxies data calls. |
| kaveon-api | FastAPI · Python 3.11 | Query execution, semantic SQL generation, the DLM engine, RBAC, connection pooling to every data source. |
| Data layer | Azure PostgreSQL · Fabric SQL · Azure SQL · MySQL · StarRocks | A two-plane store — kaveonmeta (Kaveon’s own state + DLM context) and the kaveon warehouse — plus the data sources you register. |
How these docs are organized
- Getting Started — what Kaveon is, a hands-on quickstart, and the core concepts you’ll reuse everywhere.
- Features — deep dives on SQL Lab, the AI / NL→SQL engine, the chart builder, dashboards, semantic datasets, and data sources.
- Platform — architecture, authentication & RBAC, and how to deploy.
Each page has an outline on the right; use the search box in the sidebar to jump anywhere.