Skip to main content
Last updated on

Genie Ontology

Genie Ontology is the unified context layer that gives AI agents a business-aware map of your organization. Within the Genie family of AI experiences, Genie One and Genie Code search the same ontology, so context that you curate applies to both surfaces. Agents built on other platforms can also use the ontology through the Genie One Model Context Protocol (MCP) server, with the same Unity Catalog permissions. The ontology brings together two categories of knowledge:

  • Modeled knowledge: Context that you define, govern, and certify through Unity Catalog semantics features such as metric views, domains, Pages, and certification.
  • Learned knowledge: A map of snippets that Genie automatically extracts and maintains from your existing assets and usage, including metric views, dashboards, SQL queries, and Genie Agents. These SQL queries include queries that external tools, such as dbt, run on Databricks. Those runs appear in query history. Learned knowledge can also include business definitions from external knowledge sources (Beta), such as BI tools.
note

At this time, customers do not incur costs for curating ontology snippets. However, Databricks may update pricing for the feature at any time by notifying you and/or by updating our pricing page. See Genie pricing.

Learned knowledge​

Genie stores learned knowledge as snippets. Snippets can include:

  • Metric definitions: For example, "An 'active user' is a distinct user, deduplicated across all platforms."
  • Authoritative sources: For example, "Revenue questions should be answered using the curated Finance Genie Agent."
  • Business rules: For example, "A 'qualified lead' only counts after a demo is booked."

Each snippet gets an authority score based on where it was generated from, how often it is used, and how fresh it is. When you ask a question, Genie ranks the most relevant snippets, resolves conflicts, and answers using only the sources you are permitted to see.

Snippets are gated by Unity Catalog permissions, so users only see snippets derived from assets they're allowed to access. For example, a user who doesn't have access to a Genie Agent can't see snippets derived from that Genie Agent. To keep stale or experimental assets out of answers, restrict access to them with Unity Catalog permissions. Snippets created from external knowledge sources are available to all users in your account.

View learned knowledge snippets​

Learned knowledge snippets appear in context as you ask questions. When Genie uses a snippet, you can see it in Genie's thinking steps and in the response citations, along with the assets Genie considered and the ones it chose.

To see which knowledge sources Genie used to answer your question, click the Snippet icon. citation icons in a response. To view the snippets used in a response, click Genie Ontology under the response, then click a snippet to see its description, source asset, author, domains, and lineage.

A Genie One response with a Genie Ontology snippet selected, showing the snippet's description, source asset, author, domains, and lineage.

You can't edit learned knowledge snippets directly. Instead, change what Genie learns by updating the source assets, or define the concept as modeled knowledge, which Genie treats as more authoritative. See Improve the ontology with modeled knowledge.

Modeled knowledge​

Modeled knowledge is the context that you define, govern, and certify with Unity Catalog semantics. Genie treats modeled knowledge as more authoritative than learned knowledge, so it takes precedence when the two conflict. See How Genie resolves conflicts.

Improve the ontology with modeled knowledge​

Genie works from learned knowledge as soon as you start, and answer quality improves as you add modeled knowledge. Model the business-critical concepts that must be right, and let Genie learn the rest:

  • Define your critical KPIs as metric views: Define governed metrics such as revenue and core KPIs. See Unity Catalog metric views.
  • Organize assets into domains: Group data assets by business unit, region, or function. See Domains and subdomains.
  • Define business concepts in Pages: Document key terms and definitions as an authoritative source. See Pages.
  • Certify trusted assets: Mark the data your organization vouches for, and deprecate outdated assets. See Flag data as certified or deprecated.
  • Curate Genie Agents: Genie prioritizes curated Genie Agents as authoritative sources. See Genie Agents.

You don't need to create assets just to generate more snippets. Your standard work in Databricks, such as building dashboards and Genie Agents, feeds learned knowledge.

How Genie resolves conflicts​

Genie resolves conflicts by assessing the authority of each source, based on signals such as certification, popularity, and author. Modeled knowledge, certified assets, and curated Genie Agents generally take precedence over learned knowledge on the same topic. If Genie can't resolve a conflict, it might ask you for clarification.

Extend the ontology​

Account admins can extend the ontology with sources outside Databricks:

  • Genie Search: Index data from sources such as Google Drive, Jira, and Confluence so Genie One can use it as context. Genie Code doesn't use Genie Search content.
  • External knowledge sources (Beta): Add business definitions that your teams already maintain in BI tools such as Power BI.

Account admins can also connect Workday as a people source to power org directories in Genie One.

Additional resources​

On this page