How do I get AI to understand my business?

Summary

• AI is only as smart as the context it has. On Databricks, you give AI real business understanding with Genie Ontology (in Public Preview) — a continuously learned context layer that maps raw tables and columns to the concepts, metrics, and definitions your business actually uses.
• Model context once in Unity Catalog: metric views (governed KPIs), domains (business-aligned asset curation), and glossary pages (shared terms and definitions). Every tool then reuses the same trusted definitions.
• The ontology also infers context automatically from the notebooks, dashboards, SQL, and connected apps already in your workspace, ranking knowledge by authority so AI resolves to the definition your organization trusts.
• Because that context is governed by Unity Catalog and permission-aware, answers stay accurate, consistent, and secure — grounded in your business, not guesses.
• Genie and any agent, app, or BI tool can reach the same context (including over MCP), so business users finally get an AI coworker that understands the business.

How do I get AI to understand my business?

AI is only as good as the context it is given. Out of the box, an AI agent sees your data as tables and columns with no business meaning: it cannot tell which of many revenue tables is the trusted one, or whether an "active customer" means someone who logged in this month or someone who paid. The hard part isn't generating an answer — it's knowing which information is trustworthy enough to act on. The way to solve this on Databricks is to give AI a governed model of your business with Genie Ontology, a continuously learned enterprise-context layer that gives AI a business-aware map of your organization.
Genie Ontology combines two kinds of context: the semantics you model and govern explicitly in Unity Catalog, and knowledge it infers automatically from the assets and activity already in your Databricks environment. Together they let Genie — and any connected agent — answer questions grounded in enterprise context instead of guessing.

Why Databricks Genie Ontology helps AI understand your business

• Two kinds of context, one governed layer. Genie Ontology unites modeled semantics you define in Unity Catalog (metrics, domains, glossary pages, certifications) with inferred knowledge automatically learned from your notebooks, dashboards, SQL, and connected apps. Business context is modeled once and reused everywhere.
• Metric views govern your KPIs once. Metric views are reusable SQL objects that define business KPIs as governed Unity Catalog securable objects. Because they compile into logical queries at runtime, every visual, filter, dashboard, and agent uses the same certified measures and dimensions, so the numbers are consistent across teams and tools. Metric views are generally available, and you can author measures, dimensions, and relationships with AI assistance.
• Domains scope context to the business. Domains and subdomains organize data assets into business-aligned groups, giving an agent scoped, relevant context instead of the entire catalog. Asked to forecast revenue, an agent can recognize the Finance domain as relevant and draw on the trusted, curated assets there.
• Glossary pages define shared meaning. Governed pages document business concepts, synonyms, terms, metrics, and their definitions, so every agent, user, and tool works from the same trusted semantic source of truth.
• Authority-ranked, permission-aware knowledge. The ontology builds a permission-aware context graph and ranks knowledge by authority, so when multiple definitions of a metric exist, Genie resolves to the one your organization trusts most. Certification and deprecation signals steer it toward assets you vouch for and away from stale ones, and every response respects each user's permissions.
• Grounded answers in milliseconds. When a user asks a question, Genie retrieves and ranks the relevant modeled and learned knowledge from the ontology, injects it into its reasoning loop, and returns an answer grounded in your business context — all behind the scenes. User feedback continuously improves the ranking over time.
• Reusable across every tool, over open interfaces. Query the same semantics from SQL, BI, and custom apps, and reach them from any agent or client over MCP, so every surface sees the same definition. Governing context once means more accurate answers, faster responses, and lower token cost, because agents don't re-establish context from scratch on every request.

Getting started

Giving AI an understanding of your business is a modeling exercise in Unity Catalog, followed by connecting Genie and your agents on top.

• Define metric views for your key KPIs — revenue, active users, customer acquisition cost — as governed Unity Catalog objects so every tool computes them the same way.
• Organize assets into domains and subdomains by business function (Sales, Finance, Marketing) so agents get scoped, relevant context.
• Author glossary pages that document business concepts, synonyms, and the assets they map to.
• Certify trusted assets so Genie prioritizes them and deprecate the ones that are out of date.
• Build Genie Agents on your metric views, and connect other agents and clients over MCP so they inherit the same governed definitions.
• Follow the docs on Unity Catalog semantics and Genie Ontology, and read Redefining the semantics data layer for the future of BI and AI for the full picture.

FAQs

What is Genie Ontology?

Genie Ontology is Databricks' continuously learned enterprise-context layer. It combines the business semantics you model in Unity Catalog with knowledge it infers automatically from your existing assets, giving AI a business-aware map of your organization.

What are metric views and why do they matter?

Metric views are reusable, governed Unity Catalog objects that define your business KPIs in SQL. They compile into logical queries at runtime, so every dashboard, query, and agent uses the same certified definitions and returns consistent numbers.

Why does AI give wrong answers about company data?

Usually because it lacks business context: it sees raw tables and columns and has to guess which metric definition or which source table is correct. Modeling that context in the ontology — metrics, domains, glossary pages, and certifications — lets AI answer from your organization's trusted definitions instead of guessing.

Can other agents and tools use the same business context?

Yes. Context modeled in the ontology is reusable everywhere: query it from SQL, BI, and custom apps, and reach it from any agent or client over MCP, so every surface works from the same governed definitions and respects the same permissions.

How does Databricks keep AI answers trustworthy?

The ontology ranks knowledge by authority, respects each user's permissions on every response, and uses certification and deprecation signals so AI is steered toward the assets your organization trusts and away from stale ones.