Can I use an existing ontology or knowledge graph with Genie Ontology?

Summary

• You don't upload a formal ontology file or connect an external graph database. Genie Ontology builds its own business-aware context layer from the semantics and assets already in your Databricks estate.
• You reuse existing meaning by expressing it as governed Unity Catalog objects: metric views for measures and dimensions, Pages for business terms and concepts, and domains to scope them.
• Genie also infers context automatically from dashboards, notebooks, SQL queries, and Genie Agents you already have, so it starts producing answers before you model anything.
• Genie Code can bring existing definitions in: it can import an existing BI semantic model and turn it into a metric view, and bulk-import business terms from your documents into Pages for review.
• Genie Ontology is in Public Preview. Treat the behavior described here as an emerging capability that can change; confirm current details in the Databricks documentation.

Can I use an existing ontology or knowledge graph with Genie Ontology?

Yes, though not by uploading a formal ontology file or pointing Genie at an external graph database. Genie Ontology builds its own business-aware context layer from the meaning that already lives in your Databricks estate, so the way you bring an existing ontology in is to express its definitions as governed Databricks objects and let Genie learn from your assets. Genie Ontology is in Public Preview, so treat the specifics below as an emerging capability that can change.

What is Genie Ontology?

Genie Ontology is the unified context layer that gives the Genie family of experiences a business-aware map of your organization. Genie One and Genie Code search the same ontology, so context you curate once applies to both surfaces. It brings together two kinds of context. The first is Unity Catalog semantics that you define, govern, and certify: metric views, domains, and Pages. The second is inferred context that Genie extracts automatically from your existing assets and usage, such as metric views, dashboards, SQL queries, and Genie Agents. Each piece of inferred knowledge, whether a metric definition, an authoritative source, or a business rule, gets an authority score and is gated by Unity Catalog permissions.

Ways to reuse existing semantics

If you already maintain an ontology, a semantic model, or a business glossary, the practical question is how to encode it so Genie can read it. Databricks documents several paths, summarized below.

What you already have	How to bring it into the ontology
Defined metrics and dimensions	Model them as metric views in Unity Catalog so every consumer resolves the same governed definition.
Business terms, concepts, and definitions	Capture them as Unity Catalog Pages, which list the authoritative assets tied to each concept.
Table relationships (keys)	Declare primary and foreign keys in Unity Catalog so agents join tables correctly instead of guessing.
An existing BI semantic model	Use the import skill in Genie Code to convert it into a metric view, then promote it to Unity Catalog.
Definitions spread across documents	Bulk-import them as draft Pages; Genie Code extracts and deduplicates proposed terms and flags conflicts for review.
Dashboards, notebooks, saved queries, agents	Leave them in place; Genie infers context from these assets, and richer documentation yields richer context.

When the ontology has to resolve a conflict, definitions you assert by hand in Pages take priority over inferred context, so curating the concepts that must be right is the highest-value work you can do.

How Genie Ontology builds and uses context

The public guidance is to model the "head" and let Genie infer the "tail." Genie works from what it can learn on day one, while deliberate curation improves the definitions and sources that cannot be wrong. Certifying trusted assets such as metric views, Genie Agents, and notebooks raises their authority, and deprecating stale assets steers agents away from them. When you ask a question, Genie ranks the most relevant snippets, resolves conflicts, and answers using only the sources your permissions allow, then shows citation icons for the knowledge it used. External agents can reach the same ontology through MCP, so the context you build is reusable beyond the Genie surfaces. For a staged approach to strengthening this foundation, see Operationalizing Genie Ontology in your data stack.

FAQs

Can Genie Ontology import an OWL or RDF file directly?

Databricks documents importing existing BI semantic models and business definitions from documents, not loading a raw OWL or RDF file or connecting an external graph database. You reuse existing semantics by expressing them as governed Unity Catalog objects that Genie reads.

Do I have to model everything before Genie works?

No. Genie Ontology works on day one by inferring context from your existing assets. Modeling metric views, Pages, and domains is a progressive way to raise accuracy over time, not a prerequisite for getting answers.

How do I reuse an existing BI semantic model?

Use the import skill in Genie Code to turn it into a metric view, review the generated definition, and promote it to Unity Catalog for governance, lineage, and reuse across Genie One and Genie Code.

Does the ontology respect data permissions?

Yes. Snippets are gated by Unity Catalog permissions, so Genie only uses knowledge from assets you are allowed to see, and it cites the sources behind each answer.

Is Genie Ontology generally available?

Genie Ontology is in Public Preview. Availability and behavior can change, so validate current capabilities against the Databricks documentation before you rely on them.