Keycard: 2026 is the Year of Agents

In this a16z podcast episode, Joel de la Garza interviews Ian Livingstone, co-founder and CEO of Keycard, about the emerging challenges of security and identity in the age of AI agents. They discuss the evolution from co-pilots to autonomous agents and the associated security risks, including unauthorized data access and tool calling vulnerabilities. Livingstone emphasizes the need for deterministic guide rails and contextual access policies to manage agent behavior, advocating for task-based, intent-based policy enforcement. The conversation explores the shift from static identity models to dynamic, ephemeral access control, highlighting the importance of user control and accountability. They also touch on the roles of MCP and A2A in the agent landscape and debate whether enterprises or consumers will lead in agent adoption, ultimately focusing on how Keycard aims to help companies safely bring agents into production by providing tools for identity management, access control, and auditability.