BlockEdge

Technology and digital life

Snowflake outlines Business Ontology for governed AI data analysis

In a report published on November 18, 2026, Snowflake described Business Ontology, a feature it had announced on November 3 and said was due to enter public preview. The proposed governance layer is intended to give people and AI agents shared, approved definitions of business concepts and metrics rather than leaving each application to interpret them independently.

Business Ontology is part of Snowflake Horizon Context, within Horizon Catalog. Snowflake says teams will be able to assign owners, set approval status and define relationships among terms such as customers, orders and net revenue. They can then link those definitions to the data and semantic views that implement them. For example, a finance team could approve a definition of “net revenue” and connect it to metrics in more than one semantic view without requiring every view to use the same calculation implementation. Teams could also retain separately approved definitions for different brands or business domains.

Snowflake outlines Business Ontology for governed AI data analysis (1)

Snowflake describes three distinct parts of its approach:

  • Cortex Sense, also described as forthcoming in public preview, is intended to infer context from metadata and usage patterns, including queries, dashboards and dbt models. Its suggestions would serve as drafts for review.
  • Business Ontology would govern the meaning, ownership and approval of business definitions. Snowflake says approved definitions would take precedence over conflicting automatically inferred context.
  • Semantic Views define how metrics are calculated and expose metrics, dimensions and joins for queries. Business Ontology would link approved meanings to those views, not change their calculation logic.
Snowflake outlines Business Ontology for governed AI data analysis (2)

Snowflake says Horizon Context can collect metadata from tools including Tableau, Power BI, dbt, Sigma and PostgreSQL. It also says Business Ontology supports importing and exporting business definitions through open formats, while Semantic Views can exchange definitions using Apache Ossie, an incubating project. According to Snowflake, agents outside its platform can query Semantic Views through the Model Context Protocol, subject to Horizon Catalog access controls.

The practical aim is to reduce conflicting answers when an analyst, dashboard or AI agent uses the same business term. The report describes how Snowflake intends the components to work; it does not establish that Business Ontology or Cortex Sense was generally available at the time of publication, or independently measure improvements in AI answer quality.