What is Genie Ontology and how does it work?

Summary

• Genie Ontology is a unified context layer that gives Databricks Genie a business-aware map of your organization, so it can interpret data questions the way your teams mean them.
• It combines two kinds of context: governed Unity Catalog semantics (metric views and domains you certify) and inferred context that Genie automatically extracts from your existing metric views, dashboards, SQL, and Genie Agents.
• It works automatically. Genie continuously extracts knowledge from connected assets, builds a knowledge graph, and injects relevant, permission-aware context into the agent at query time, with no manual modeling required to start.
• It is permission-aware. Snippets are gated by Unity Catalog permissions, so Genie only uses knowledge from assets you are allowed to see, and you can click citation icons to see which sources answered your question.
• Genie Ontology is in Public Preview. As of August 2026, ontology snippets are available to all customers without requiring special access requests.

What is Genie Ontology and how does it work?

Genie Ontology is a unified context layer that gives Databricks Genie a business-aware map of your organization. Rather than inventing business meaning on its own, Genie interprets meaning that is already encoded in your data, metrics, and governance. A well-curated semantic layer is the foundation of reliable answers. Genie Ontology combines two categories of context: Unity Catalog semantics (modeled context you govern and certify through features like metric views and domains) and inferred context (a map of snippets Genie automatically extracts and maintains from your existing assets and usage, including metric views, dashboards, SQL queries, and Genie Agents). Example snippets include metric definitions (for instance, "an active user is a distinct user, deduplicated across all platforms"), authoritative sources, and business rules.

Genie Ontology is in Public Preview. Availability and behavior may change before general availability.

Why Databricks Genie Ontology for trusted natural-language answers

• Automatic context, no manual modeling to start. Genie Ontology continuously extracts knowledge from assets already connected to Databricks, such as notebooks, dashboards, pipelines, and connected apps, builds a knowledge graph, and injects relevant context into the agent at query time. It requires no manual modeling to begin and improves as your team uses it.
• Authority-ranked knowledge. Each snippet receives an authority score based on where it was generated, how often it is used, and how fresh it is. Genie ranks knowledge by authority using an algorithm called OntoRank, so more trusted sources carry more weight.
• Permission-aware by design. Snippets are gated using Unity Catalog permissions, so Genie only uses snippets extracted from assets you have permission to see. The right people see the right information.
• Accurate answers with lower latency. When you ask a question, Genie ranks the most relevant snippets, resolves conflicts, and answers using only the sources you are permitted to see. Injecting focused context reduces how much data Genie needs to crawl and query, which supports accurate responses and lower latency.
• Transparent citations. You can see which knowledge sources Genie used to answer your question by clicking the citation icons in a response.
• Complements the Unity Catalog Business Glossary. Where the ontology infers meaning, the Business Glossary lets you state it authoritatively, with a version-controlled way to define concepts, group them into domains, and link concepts to specific tables and columns.

Getting started

• Read the Genie documentation and Genie Agents concepts to understand the context model.
• Build a strong semantic foundation: define consistent metrics with metric views, add instructions with business context and terminology, and define filters for common query patterns.
• Let the ontology work automatically as your team uses Genie; it extracts and ranks context from your connected assets without upfront modeling.
• Use the Unity Catalog Business Glossary to authoritatively define concepts and domains that complement the inferred ontology.
• Learn more in Pushing the frontier of data agents: Genie and AI/BI Genie is now generally available.

FAQs

Is Genie Ontology generally available?

Genie Ontology is in Public Preview. As of August 2026, ontology snippets are available to all customers without requiring special access requests.

Do I have to model the ontology manually?

No. Genie Ontology builds and maintains context automatically from your connected assets and usage. You can strengthen it by curating metric views, instructions, and a Business Glossary, but it requires no manual modeling to start.

Does Genie Ontology respect data permissions?

Yes. Snippets are gated using Unity Catalog permissions, so Genie only uses knowledge from assets you have permission to see.

How is Genie Ontology different from the Unity Catalog Business Glossary?

The ontology infers meaning automatically from your assets, while the Business Glossary lets you state meaning explicitly and authoritatively through version-controlled concepts, domains, and semantic assignments. They work together.