The Pulse
Snowflake announces Snowflake Business Ontology and Cortex AI Gateway to support enterprise AI agents
Snowflake announced Snowflake Business Ontology and Cortex AI Gateway at Snowflake Expedition 2026 on Nov. 3 U.S. time to provide enterprise AI agents with governed business context and centralized controls. Snowflake says the features cent

AI.info Team · Written with AI · fact-checked ·
Snowflake announced Snowflake Business Ontology and Cortex AI Gateway at Snowflake Expedition 2026 on Nov. 3 U.S. time, saying the additions bring governed business context and controls to enterprise AI agents. Snowflake describes Snowflake Business Ontology as a way to ground agent answers in approved business definitions. Cortex AI Gateway centrally manages agents’ access to models, tools and business systems.
Snowflake Business Ontology enables AI inference based on governed business context
Snowflake describes fragmented business context as a factor in different dashboards, analysts and agents returning different answers to the same question. In a survey conducted by MIT Technology Review with Snowflake’s support, 55% of executives said insufficient business context was behind poor-quality insights from AI agents. A further 45% cited fragmented data or difficulty accessing it.
Business Ontology lets teams approve definitions for business concepts such as net revenue and link them directly to underlying tables and views. Snowflake says those definitions can then be applied across enterprise data, including data outside Snowflake. The release describes the ontology as one part of a framework: it holds the formal meaning of business terms, Semantic Views handle calculations, and Cortex Sense reasons about knowledge that has not been formally defined.
Horizon Context gathers metadata from sources including Tableau, Power BI, dbt, SQL Server and PostgreSQL. Snowflake says it enriches that metadata with data lineage, tags, AI-generated descriptions and usage signals. CoCo can help teams move definitions from third-party catalogs or spreadsheets into Snowflake in a single session as drafts for owners to review and approve; changes are recorded for later audit.
Cortex AI Gateway applies common controls to all agents
The gateway applies common policies across clouds, frameworks and AI providers, Snowflake says. For tool calls made through MCP servers, it applies enterprise access policies, authentication and permissions, while agents inherit controls already set for human users. Cortex Sandboxes provide private environments for running custom code near enterprise data in Snowflake, with the executing user’s permissions applied automatically.
Snowflake also describes Agent Abuse Detection, which analyzes an agent’s execution across a series of actions rather than assessing prompts one at a time. The company says the feature detects behavior that exceeds an agent’s permissions, as well as attempts to bypass approvals or remove sensitive data. Snowflake says it worked with OpenAI as a technical design partner on Private Safety Processing, which supplies privacy-protected safety information to support detection and analysis of dangerous agent activity.
The release says agent activity—including models called, tools used and decisions made—is recorded in each customer’s Snowflake account. Agent Trace Intelligence lets teams query those execution records alongside business data. Cost governance tracks AI spending by user, team, department, agent and application and enforces configured limits, while Dynamic Model Routing sends requests to organization-approved models based on cost, accuracy and latency.
Companies using Snowflake’s AI-agent features include Glean, Kargo, dentsu and Merkle
Snowflake says Glean and Kargo use Cortex AI Gateway to extend enterprise agent use while managing security, compliance and spending. Uniphore is extending the gateway’s controls to agents it orchestrates for shared customers. The release also names dentsu and Merkle among companies promoting agent operations based on business-approved definitions through Snowflake Business Ontology.
Snowflake’s product executive vice president, Christian Kleinerman, said: 「エージェント時代においても、ガバナンスの役割は変わりません。ガバナンスは信頼を醸成し、AIの導入を加速させます。」
Snowflake says the new features add to Horizon Context within Snowflake Horizon Catalog, which integrates enterprise data environments across clouds and catalogs and provides common governance and context for AI agents and analytics. Snowflake says the setup keeps definitions tied to approved business meanings, records agent actions and applies policies across the data environment.