As enterprises expand the use of AI agents across business applications, one challenge has become increasingly difficult to manage: understanding what those agents are doing, what resources they consume, and whether they operate within corporate policies. Snowflake, a cloud data platform provider, has introduced Cortex AI Gateway to address that problem with a runtime governance layer for AI workloads.
Built on technology Snowflake acquired through its purchase of Natoma in May, Cortex AI Gateway sits between AI agents, models, enterprise systems, and data. The platform applies governance policies from Snowflake's Horizon Catalog while routing requests, enforcing access controls, and tracking resource consumption across both Snowflake services and third party AI tools.
"Its role is to provide a trusted control plane that governs how AI agents securely access models, tools, MCP servers, enterprise systems, and data across increasingly diverse AI environments as enterprises adopt more models and agents for their applications and workflows," said Artin Avanes, head of core data platform at Snowflake.
Industry analysts said the product addresses an operational challenge that has emerged as organizations move beyond isolated AI deployments. Michael Leone, principal analyst at Moor Insights and Strategy, said most AI gateways focus on routing model requests and logging prompts rather than governing agent behavior across enterprise environments.
"Most enterprises cannot see or govern agent activity consistently across models, tools, MCP servers, and enterprise systems, as most AI gateways just route models and log prompts," Leone said.
Stephanie Walter, practice lead of AI stack at HyperFRAME Research, argued that organizations also need visibility into how autonomous systems make decisions and consume resources.
"Enterprises need to know which agent acted, who authorized it, what resources it used, and what happened at each step. Without that runtime evidence, firms cannot reliably secure, audit, or contain agentic workflows," she said.
The gateway combines governance with FinOps capabilities, allowing organizations to monitor token consumption and AI costs through the same control plane used to enforce security policies. According to Scott Bickley, advisory fellow at Info-Tech Research Group, that approach reflects the shift toward consumption based pricing for AI models and agents, where organizations need shared usage data and traceability to manage both governance and spending.
For developers, the platform could simplify application development by centralizing credentials, logging, cost instrumentation, and access controls in a single layer. Bickley noted, however, that the opposite could happen if enterprises require teams to work with multiple gateway products, creating additional complexity and slowing development.
Snowflake said Cortex AI Gateway is scheduled to enter public preview soon.
