The RSAC Innovation Sandbox has long been one of the most watched competitions in cybersecurity, and this year’s winner caught much of the industry off guard. Alan Shimel sits down with Henry Comfort, CEO of Geordie AI, to talk about how a startup that was buying laptops just a year ago ended up taking the crown at the industry’s most prestigious launchpad.

Comfort’s path to cybersecurity was anything but conventional. Before founding Geordie AI, he ran operations for a UK football club and helped lead Darktrace through its hypergrowth phase. That experience across very different environments shaped his view of what enterprises actually need as they adopt AI at scale. The core problem he keeps seeing is a lack of visibility. Organizations are deploying AI agents and agentic workflows faster than their security teams can track, creating blind spots that traditional tools were never built to address.

The gap Comfort describes sits between two camps that have dominated the enterprise AI conversation. On one side are organizations that refuse to touch AI out of fear. On the other are those racing ahead without adequate guardrails. Most companies fall somewhere in the middle, wanting to move forward but lacking the ability to see what their AI systems are actually doing across the organization. Giving security teams that visibility and control over rapidly expanding agentic footprints is where Comfort sees the most urgent need.

For security leaders trying to get ahead of the governance challenge that agentic AI is creating, the Innovation Sandbox result signals where the industry thinks the next wave of critical investment needs to go.