Session

What’s New in Unity Catalog With Live Demos!

Overview

ExperienceIn Person
TrackGovernance & Security
IndustryEnterprise Technology
TechnologiesUnity Catalog
Skill LevelIntermediate

Unity Catalog is evolving fast! Join the product team for a demo-packed dive into the latest capabilities shipping now and what’s on the horizon for data governance.

 

We’ll unveil powerful new features across access control, data lineage, AI governance, observability, discovery and many more!

 

Walk away with a complete understanding of what Unity Catalog can do for your organization today — and what advancements are coming next!

Session Speakers

Lianne Zelsman

/Sr. Product Manager
Databricks

Speaker placeholderIMAGE COMING SOON

Paul Roome

/Senior Staff Product Manager
Databricks

Full Summary

Extending Unity Catalog for the agentic era: AI primitives, access control, and context

Autonomous agents are rapidly becoming a dominant source of queries and actions across data and AI systems. The session outlines how Unity Catalog on the Databricks Platform is evolving to govern AI-native assets, scale access control for thousands of automated actors, and supply the business context agents need to answer correctly. As explored in enterprise AI agent trends, top use cases, governance, and evaluations, the patterns shaping agentic adoption make unified governance more critical than ever.

FAQ


AI securable objects, such as skills, agents, model services, and MCP servers, are governed assets inside Unity Catalog with access control, lineage, and audit. Unity AI Gateway enforces policies at runtime. It controls budgets and token costs, applies guardrails like PII redaction and prompt injection protection, and logs every request and response to a Delta Lake table.

Instead of inheriting the union of all group permissions, a user switches into a specific project role that creates a true isolation boundary. Switching roles is an instant workspace toggle, enabling focused work in one context without logging out or changing workspaces.

No. An AI-powered data classification service continuously detects sensitive patterns, applies governed tags, and supports custom classifiers that learn from already tagged columns. ABAC policies then reference those tags to mask, filter, grant, or deny access automatically.

Yes. The expanded policy model supports request context attributes, including whether the caller is a user or an agent acting for a user. Different rules can be applied based on that distinction.

Glossaries, domains, and metric views provide curated, deterministic definitions that are precise but inevitably incomplete. Genie Ontology learns from real usage in dashboards, notebooks, and queries to infer missing concepts and calculations, filling coverage gaps and informing glossary pages via GenAI code.