Wren AI
A self-hosted data agent that turns questions into SQL and dashboards, connects to Claude and ChatGPT over MCP, and can run fully air-gapped.
Wren AI is a self-hosted data agent for teams and agent builders who need answers based on agreed business definitions. It turns plain-language questions into SQL and interactive dashboards, using the same definitions when someone asks directly or through Claude, ChatGPT or Gemini over MCP.
Its context layer holds approved metrics, relationships between tables, units and business knowledge that database schemas alone don't capture. Teams keep that context in Git as reviewable YAML and Markdown. This gives agents a shared definition of terms such as revenue rather than leaving each agent to interpret them independently.
Wren queries data where it already lives, without requiring ETL or a database migration. Connections include PostgreSQL, Snowflake, BigQuery, ClickHouse, Databricks and DuckDB. Dashboards support drill-down, roll-up and filters, while product teams can embed analytics under their own brand through an iframe, API or MCP.
The commercial platform includes row- and column-level access controls, role-based permissions and audit logs. Its agent can reason in an isolated sandbox, with replayable traces for inspecting its work. Developers can also connect it to LangChain and LangGraph.
You can run Wren on your own servers, in a VPC or fully air-gapped on-premises; Wren AI Cloud is the hosted offering. The context engine can run without a vendor account, and the engine uses Apache DataFusion.
