Snowflake Launches Business Ontology to Move AI Agents from Pilot to Production

Snowflake Tackles Enterprise AI Agent Trust Gap

Snowflake announced Business Ontology and expanded Cortex AI Gateway capabilities at Snowflake Expedition 2026 on November 4, targeting a critical challenge preventing enterprises from moving AI agents from pilot to production1

1

2

2

. According to a recent MIT Technology Review report sponsored by Snowflake, 55% of executives identify lack of business context as the primary cause of low-quality AI agent insights, while 45% cite fragmented or inaccessible data4

4

. "The enterprises that win won't be the ones with the most agents. They'll be the ones whose agents know exactly what the business means by accessing the right data, abide by governance and guardrails specified, and can account for actions taken by agents," said Christian Kleinerman, EVP of Product at Snowflake2

2

.

Business Ontology Establishes Single Source of Truth

Business Ontology, integrated within Snowflake Horizon Catalog, allows enterprises to define and approve business concepts, assign owners to definitions, and link them directly to underlying tables and views1

1

. This addresses a common problem where different departments interpret the same metric differently. For instance, a sales leader requesting a weekly net revenue report might receive three different answers from a dashboard, an analyst, and an AI agent4

4

. Business Ontology ensures AI agents access authoritative business definitions rather than relying on competing interpretations across systems1

1

. The framework works alongside Semantic Views for calculations and Cortex Sense for inferring undefined meanings, though approved definitions always take precedence4

4

.

Simplifying Agent Development and Scaling

For developers, Business Ontology reduces the effort required to build context for AI agents because they no longer need to embed business logic into every agent they create, according to Ashish Chaturvedi, executive research leader at HFS Research1

1

. Agents can draw on centrally maintained definitions in Business Ontology, allowing developers to reuse the same business context across multiple agents and applications1

1

. This capability reduces inconsistencies between agents interpreting business data differently and simplifies maintenance as business requirements evolve1

1

. For CIOs managing over 14,500 global enterprises using Snowflake, establishing common business definitions makes it easier to scale AI agents across departments without interpretation conflicts4

4

.

Cortex AI Gateway Expands Governance Capabilities

Cortex AI Gateway serves as the single front door for agents within an organization, applying enterprise access policies, authentication, and permissions across different AI providers and frameworks2

2

. The expanded capabilities include agent abuse detection, observability and tracing, Cortex Sandboxes, cost governance, dynamic model routing, and tools integration using technology from Snowflake's Natoma acquisition3

3

. The gateway provides fine-grained authorization, protection against prompt injection and jailbreaks, and records agent activity for monitoring usage and spending2

2

. Cortex Sandboxes give agents isolated environments to execute custom code alongside enterprise data, while Agent Trace Intelligence allows developers and data teams to examine agent execution traces against business data1

1

.

Autonomous Execution with CoCo and CoWork Enhancements

Snowflake introduced automation features for CoCo, its coding agent for data teams, and CoWork, its personal work agent for knowledge workers2

2

. CoCo Automations allow data teams to configure recurring workflows that agents execute autonomously, including pipeline health checks and weekly reports, with each execution maintaining an audit trail and task history2

2

. The new CoCo and CoWork Bots operate continuously in the background, act directly in tools teams already use like Slack, Excel, and Google Sheets, and check in only when decisions or sign-offs are required2

2

4

4

. These bots inherit users' access rights and permissions while maintaining governance controls2

2

.

Competitive Positioning in Agentic Enterprise Race

"AI is industrializing software creation. Actually, it's industrializing intelligence," said Sridhar Ramaswamy, CEO of Snowflake3

3

. Ramaswamy noted that while AI changes the economics of software development as agents take over specialized tasks, the competitive advantage shifts from specialists to proprietary knowledge3

3

. Snowflake's push comes amid broader industry competition to establish the control layer for enterprise AI agents, with SAP expanding its Autonomous Enterprise architecture and Salesforce introducing its AI Control Plane3

3

. Snowflake approaches the challenge from the data platform side, positioning itself as the foundation through which AI agents access trusted business information and operate under enterprise policies for the agentic enterprise3

3

. Business Ontology is expected to be available in public preview soon, while Cortex Training entered public preview to allow customers to tailor open-weight AI models with proprietary information1

1

3

3

.