Snowflake: trust and security are key to the agentic enterprise
At Snowflake World Tour London, Snowflake chief security and trust officer Mayank Uphadyay said this year has been about figuring out how to govern AI, which he framed as two sides of one coin: keeping other companies' AI out of your perimeter, and keeping your own AI well-behaved so it does not exfiltrate data or wander outside its boundaries. He likened the AI model to a brain and the agent to a body, resting on data that must be "rock solid," and argued security is now an enabler rather than a blocker. Snowflake pointed to its Trust Center tool to connect and protect those layers and demoed its Coco and CoWork platforms; Uphadyay stressed that agent permissions matter, and that automation should climb a pyramid of tasks with humans required for mission-critical work.
As enterprises deploy agents at scale, security and trust governance are becoming a precondition for adoption, and Snowflake's stance reflects that industry shift.