Enterprises Pilot AI Agents, But Trust Deficit Hinders Production

A significant gap exists between enterprises piloting AI agents and those deploying them, with only 5% moving agents to production despite 85% running pilots. Cisco's Jeetu Patel highlights 'trust' as the primary impediment, emphasizing the need for a robust trust architecture to bridge this divide. The shift from information risk to action risk is critical, as AI agents can take irreversible actions. Cisco is addressing this with new security tools like AI Defense Explorer Edition and Agent Runtime SDK, alongside an open-source framework called Defense Claw integrated with Nvidia's OpenShell, aiming to embed security enforcement from the start.