Snowflake Unveils Cortex AI Gateway: Managing AI Agents and Maximizing Cost Efficiency

Snowflake recently introduced Cortex AI Gateway, a centralized control layer aimed at overseeing how AI agents, even those from competitors, interact with enterprise data, tools, and models. The company also revealed security integrations with various identity vendors to ensure secure agent interoperability.

This announcement represents Snowflake’s strategic move to position itself not only as a data repository but also as the orchestrator determining the actions AI agents can perform. Mayank Upadhyay, Snowflake’s chief security and trust officer, emphasized the importance of secure agent interoperability for fostering innovation and scalability in AI across businesses.

The core argument put forth in the announcement challenges traditional enterprise security models, which were designed under the assumption that human actors were behind every access request. With the advent of AI agents, these models are no longer sufficient, as agents can operate at machine speed, combining access across systems and potentially exposing vulnerabilities.

To address these challenges, Snowflake’s Cortex AI Gateway serves as a central hub for managing trusted agent activity, both internally developed and third-party. It enforces access policies, authentication, permissions, and audit logging, while also providing visibility into AI consumption and cost attribution.

The integration of dual attribution and task-scoped access ensures that actions taken by AI agents are traceable back to both the agent itself and the human who authorized the task. This approach aims to maintain the original intent of each action and prevent unauthorized behavior.

SailPoint, one of the identity vendors collaborating with Snowflake, highlighted three common failures in enterprise identity systems when dealing with AI agents: scale-driven shallowness, drift, and missing data context. They revamped their architecture to treat AI identities as first-class objects, ensuring a unified lineage from human to agent to tool to data.

The collaboration between Snowflake and various identity vendors underscores the importance of a shared trust framework to address the challenges posed by AI agents. By combining trust and a system of record, the partners aim to create a more secure ecosystem for autonomous agents.

As the use of AI agents continues to grow, the governance of these agents becomes a critical aspect of enterprise security. Analyst forecasts predict a significant increase in AI agent deployment, highlighting the need for robust governance frameworks to prevent incidents and ensure compliance.

In conclusion, Snowflake’s approach to agent governance, alongside its partnerships with leading identity vendors, signals a shift towards a more secure and controlled environment for AI-driven operations. By prioritizing trust, visibility, and control in the agentic enterprise, Snowflake aims to set a new standard for secure AI interactions.