Snowflake

AI · Data Cloud · Enterprise Security

Launch

Snowflake launches Cortex AI Gateway to govern enterprise AI agents

It treats every AI agent as a machine identity, forcing rival security vendors like Okta and 1Password to plug into one governance layer.

• Snowflake introduced Cortex AI Gateway on July 28, 2026, a centralized control layer to secure, orchestrate, and govern both its own AI agents (CoWork, CoCo) and third-party agents like Claude Code and Cursor.
• Snowflake is the AI Data Cloud company (NYSE: SNOW); the gateway builds on its May 2026 acquisition of Natoma, whose enterprise MCP platform now underpins agent interoperability across the product.
• The gateway supports over 100 MCP servers and centralizes access policies, authentication, permissions, and audit logs so agent activity can be tracked like human user activity.
• A first wave of identity integrations spans 1Password, Aembit, Linx Security, Okta Cross App Access, SailPoint, and Saviynt, letting enterprises consistently govern and audit agents across platforms.
• 1Password's CTO Nancy Wang warned that simply handing an agent credentials can grant it admin-level access, opening the door to data leaks via prompt injection and audit trails that can't distinguish human from agent behavior.
• BlackRock and Thomson Reuters are already applying Snowflake's zero-trust principles to their agent deployments, signaling early enterprise demand for this kind of governance layer.
• As autonomous agents start acting at machine speed across enterprise systems, unified identity governance and cost tracking shift from a nice-to-have to core infrastructure for safe AI adoption.