Connect Postgres, MySQL, and SQL Server to AI agents

Your own databases are now connections like any other integration. Connect PostgreSQL, MySQL, or SQL Server, and agents can answer questions from your data in chat or use it inside scheduled workflows. When you have more than one, type @ and the connection's name to point the agent at exactly the right one.

How it works

• Schema introspection. On connect, we read the database structure (tables, views, column types, primary and foreign keys), so agents query tables and joins that actually exist instead of guessing.
• A logged proxy. Queries go through a General Input endpoint that opens the connection, enforces a 30-second statement timeout, and caps results at 10,000 rows. Every query is recorded in Access logs against the credential and the run that made it.
• Notes on connections. Any connection can carry notes, like "read-only replica; orders live in public.orders," which agents read when choosing a connection.
• Standard settings. Host, port, database, user, password, an optional Postgres schema, and SSL. SQL Server connections encrypt by default.