Build a Governed AI Agent with Starburst MCP Server

AI agent prototypes often stall when they need real enterprise data. Connecting an agent to a database can be easy, but giving that agent the right business context, enforcing access controls, and keeping queries read-only and predictable is the real challenge.

In this workshop, you’ll connect an MCP-compatible client to Starburst MCP Server. You’ll build an agent that discovers data products, runs governed SQL, and completes a multi-step workflow across your lakehouse, all using your existing authentication and permissions. You’ll leave with a working pattern for grounding agent responses in governed business context instead of raw tables or custom integrations.

You’ll learn how to:

• Connect an MCP-compatible client to Starburst MCP Server and see how MCP servers and clients fit into an agent workflow
• Discover data products and pull metadata that helps agents write better queries
• Run governed, read-only SQL through existing authentication and access controls
• Design a multi-step agent workflow grounded in governed data