Snowflake has unveiled its new Cortex AI Gateway alongside several AI security innovations, designed to help enterprises safely scale their agent ecosystems. These new initiatives tackle the two primary obstacles of deploying AI agents at scale: security risks and cost overruns, by providing unified monitoring and expense management.

As AI agents increasingly collaborate across enterprise data, applications, and platforms, they introduce unprecedented security and governance risks while also driving up AI consumption costs. Traditional security architectures are ill-equipped to handle this cross-system agent activity, and enterprises often lack centralized governance tools. The Cortex AI Gateway addresses this challenge, built upon technology Snowflake acquired with its purchase of Natoma in May 2026.

The gateway provides enterprises with a unified foundation to govern how all agents—including Snowflake's own CoWork and CoCo, as well as third-party agents like Claude Code and Cursor—access models, tools, and enterprise systems. Its core capabilities include: centralized management of each agent's connections and permissions; end-to-end agent activity logging for enhanced visibility; unified monitoring and management of AI consumption costs to prevent runaway expenses; and intelligent routing of requests to enterprise-approved models to optimize both performance and cost.

Snowflake's Chief Security and Trust Officer stated that enterprise AI is moving from data interoperability to agent interoperability, and security must be at the heart of this transition. Agent interoperability can only be achieved when enterprises can trust how agents across platforms access data and execute actions. To build a broader agent security ecosystem, Snowflake also announced integrations with security vendors including 1Password, Okta, and SailPoint, helping enterprises more effectively govern and audit third-party agents.