At the 2026 Data + AI Summit, Databricks unveiled significant enhancements to Unity Catalog, addressing the governance needs of AI agents acting autonomously on enterprise data. New features expand control beyond simple data access to runtime decision management, deliver shared business semantics for agents, and simplify consistent governance across multi-cloud deployments.
- Govern AI agent actions with Unity AI Gateway runtime controls
- Provide agents shared business meaning through Glossary and Domains
- Enable consistent multi-cloud governance with unified policies and catalogs
Infrastructure signal
Unity Catalog now governs not only static data assets but also runtime behaviors of AI agents, models, tools, and MCPs through the new Unity AI Gateway layer. This shift transforms the catalog from a passive registry into an active decision-making runtime component that controls how agents act on enterprise data and services in real time.
The platform continues to unify data governance policies and catalogs across cloud providers and regions, reducing operational complexity for enterprises managing hybrid and multi-cloud environments. The introduction of centralized governance dashboards like the Governance Hub enables teams to monitor posture, detect risks, and optimize performance and cost from a single pane of glass.
Developer impact
Developers benefit from the enhanced role-based access control (RBAC) system, which supports role assumption and fine-grained permission models tailored to sensitive workloads. This flexibility allows improved data isolation for projects such as clinical trials or country-specific processing and facilitates debugging and privilege management directly from the user interface or programmatic OAuth flows.
With Unity Catalog Semantics extending to include Glossary and Domains, developers can build AI agents and applications that directly consume a shared, authoritative source of business concepts and KPIs. This standardization reduces errors and hallucinations in AI-driven reasoning by ensuring agents operate with clear and consistent business context accessible via SQL, APIs, and MCPs.
What teams should watch
Cloud infrastructure and governance teams should closely evaluate Unity AI Gateway's runtime control capabilities, as these will affect how AI agents interact with data models and services, impacting data security, audit trails, and compliance posture in real time. Teams operating in regulated industries or sensitive data environments should prioritize adopting role-based access control policies as they become generally available.
Data engineering and AI platform teams need to align on deploying and maintaining the shared Glossary and Domain semantic layers to empower AI workflows with accurate business meaning. This collaborative metadata management reduces agent errors and supports decision transparency across tools and regions.
Organizations managing multi-cloud or multi-region pipelines should leverage the unified catalog and policy framework to ensure consistent governance, simplify operational overhead, and better control cloud costs related to querying and processing data and AI workloads.