Snowflake introduced its Business Ontology feature on November 3, enabling companies to assign a single approved definition for each business term, accompanied by an owner and an audit trail. This tool aims to reduce discrepancies in business data, as different departments may provide varying definitions for terms such as “net revenue.” The announcement was made at the company’s Expedition 2026 virtual conference, which included additional features such as an expanded Cortex AI Gateway and enhanced autonomous data engineering capabilities.
Business Ontology is designed to ensure that all agents within a company, including analysts and AI systems, reference the same definition, thereby streamlining information retrieval. Kyt Dotson of SiliconANGLE noted that without universally accepted definitions, agents can produce different and incompatible answers to the same question, even with accurate data. Currently, Business Ontology is in private preview, with Snowflake indicating that a public preview will be available soon.
According to Snowflake, effective use of AI agents requires three types of context: a governed definition of each term, a Semantic View indicating how the term is calculated, and insights inferred from operational knowledge. In cases where an agent’s inference conflicts with an approved definition, the company stated that the approved definition will take precedence.
Built-in connectors will allow the software to gather schemas, query logs, and dashboard definitions from various sources, including Tableau and Power BI. This metadata will be enhanced with lineage tracking and automated tags, facilitating easier migration of existing definitions from third-party catalogs to Snowflake’s system. The significance of the draft status in this process emphasizes the importance of approval; each change is logged, allowing for easy retrieval of the approved definitions at any time.
Dentsu is one of the early adopters of Business Ontology. Joe Tobey, the head of data product engineering at Dentsu, commented that it is essential for definitions to remain dynamic rather than static, especially as AI agents become central to data queries. The company believes that having accountable owners for each definition enhances the integrity of the data being queried.
The Cortex AI Gateway, initially introduced in July 2026, offers governance and monitoring tools for AI agents. It enforces existing data protection measures, ensuring that agents operate under the same permissions as human users. Recent updates include features for detecting potential abuse, better monitoring of agent activity, and cost governance functionalities.
Looking ahead, the Business Ontology will soon transition to a public preview, which will be a key indicator of its reception beyond initial users like Dentsu. Snowflake’s roadmap includes future integrations with Oracle Cloud Infrastructure and increased capabilities for managing AI-driven operations, positioning the company to better serve clients navigating complex data landscapes.
