Unity Catalog was supposed to solve governance. Two years in, you have governance over an estate nobody understands. You have lineage you can't navigate. Ownership nobody can answer for. Dashboards whose data sources nobody remembers. Then agents arrived, and your estate had to start explaining itself to machines that can't ask follow-up questions. Your agent needs to understand data in context, but you can't even do that yourself. Your Agentic MCP needs an ontology. Your ontology needs you to actually know what's in your estate. So where do you even start?

This session works through the chain from sprawl to agent-ready estate. We start with what UC sprawl actually looks like and why standard discovery tooling plateaus there. We'll discuss how you put semantic meaning over a sprawling estate, what an ontology over a lakehouse actually needs to contain, and where the design decisions bite. Then we'll look at what's in UC to enable agents, how an agent talks to your data estate, what it needs to know in order to succeed, and what breaks when the ontology underneath is incomplete. We'll run through the tools we build to actually enable this at scale.

For data platform leads, architects, and governance owners running Databricks estates that have to serve agents but can't yet serve their own humans. You'll leave with a vocabulary for the sprawl problem, a clear-eyed view of what an ontology over a lakehouse needs to do, and patterns for building the MCP layer your agents will actually need.