MCP Server – Building Your Agent with Starburst

Most AI agent projects stall when they hit real data: brittle integrations, custom glue code, and no clear way to enforce governance at scale. In this 90‑minute, hands‑on workshop, you’ll learn how to use Starburst’s MCP Server as the governed gateway between your agents and your production data—so you can build powerful agents without sacrificing control.

We’ll walk through what MCP is, how Starburst exposes data products and SQL endpoints as MCP resources, and how to wire up popular MCP clients (like Claude or ChatGPT Enterprise) to Starburst. From there, we’ll build and iterate on a simple agent that can discover data products, issue governed queries, and run multi-step workflows on top of your lakehouse.

You’ll learn how to:

• Configure and connect an MCP-compliant client to the Starburst MCP Server for secure, read-only access to governed data
• Treat Starburst data products as first-class MCP resources so agents work with business-ready context
• Design and test an end-to-end agent workflow that you can adapt to your own environment