The internal marketplace that helps people find trusted, relevant data and AI assets — and the business-context layer that helps agents use them reliably.
by Stef Bran Melendez and Kelly Albano Today, we're announcing the Public Preview of Domains and the Discover page, powered by Unity Catalog.
As data and AI estates grow, it becomes increasingly difficult for people and agents to determine which assets are relevant, high quality, and safe to use. Get it wrong, and the consequences are significant: an agent grounds its answer in the wrong table, a business user builds a report using the wrong metric, or a team recreates something that already exists elsewhere. And as agents take on more of this work, that risk compounds.
Discover and Domains were created to solve this problem. Domains organize data and AI assets around how your business is structured - such as by function, business unit, or geography - making it easier to find relevant, trusted assets. The Discover page is the human-facing experience, where people can browse domains and find the assets they need. That same domain structure helps agents understand which assets are relevant to a given business area and which ones can be trusted.
Domains extend Unity Catalog Semantics by capturing how your organization structures and understands its data in business terms. That context feeds Genie Ontology, helping agents find and use the right assets within each user’s existing permissions.
The Discover page gives your organization an internal marketplace for data and AI assets, including tables, dashboards, notebooks, queries, metric views, Genie Agents, and apps, all organized by domain rather than by technical location. Instead of hunting through schemas and catalogs, people can browse by business area, search across certified assets, and land on something they can trust without having to track down an owner to ask "is this the right one?"
Out of the box, Discover helps you find high-quality, relevant assets by surfacing:
Data stewards can shape that experience further. They can:
The result is fewer dead ends and more effective discovery. Instead of pinging a channel to ask, “Does anyone know which table is the source of truth?”, people can browse by business area, explore its subdomains, or use AI-powered recommendations to find the trusted assets that support their work.
As organizations adopt AI, agents need business context to deliver accurate, reliable outputs: how an organization defines key terms and metrics, which assets are relevant to different areas of the business, and which sources can be trusted. Without it, they may use the wrong tables, misinterpret important business concepts, or overlook the assets that matter most.
Domains provide a foundation for that context. For example, when an agent is asked to forecast revenue, it can recognize the Finance domain - or a narrower subdomain like Revenue - as relevant and draw on the trusted, high-quality assets curated there, rather than searching across the entire data estate. Glossary, coming soon, will add shared definitions for terms like revenue, bookings, and annual recurring revenue, helping the agent understand those concepts as the business intends.
Every domain you curate becomes part of Unity Catalog Semantics and feeds Genie Ontology, which combines the context your organization defines with what it continuously learns across your environment. Agents across Databricks can draw on that shared foundation to narrow retrieval to the most relevant domains and subdomains, prioritize trusted assets and metrics, and interpret results using the language of the business - all within each user’s existing permissions. The work you do to organize data for people also strengthens the context available to agents, making existing agents more reliable and new ones easier to build.
The Discover page and Domains are in Public Preview today. Open the new Discover page in your Databricks workspace to create your first domain and begin curating trusted data and AI assets for your organization. Every domain you build improves discovery for your teams today while creating a stronger foundation for AI tomorrow.
See the Discover documentation to get started, and tell us what you'd like to see next.
Subscribe to our blog and get the latest posts delivered to your inbox.