Why Are We Still Governing AI Agents Like Software? - Tim Tyler, Gravitee
AI agents should be governed as a digital workforce, not as software
“I'm the head of AI in an organization, and I'm torn: I go to my CTO, and he says this isn't really a software application.”
Tim Tyler of Gravitee argues that organizations are deploying AI agents under conditions no one would accept for a human employee—no identity, unclear roles, shared credentials, zero accountability—and proposes reframing agent governance through a 'digital workforce' lens. He introduces an 'agent accountability' framework with three stages, beginning with the assignment stage, which centers on giving each agent a verifiable identity. The talk surfaces a real organizational tension: CTOs see AI agents as software, HR sees them as software, and neither framing is sufficient for the accountability challenges agents introduce.