
The podcast explores the evolving landscape of AI agents, particularly focusing on the critical challenges of identity, authentication, and authorization in enterprise adoption. Ian Livingstone, Keycard co-founder and CEO, joins the conversation to discuss the continuum of agentic behavior, from co-pilots to fully autonomous agents, and highlights the security risks associated with tool calling and data access. A key concern raised is the potential for agents to access and expose sensitive data across different compute boundaries, emphasizing the need for dynamic, intent-based access policies. The discussion also covers the limitations of existing security standards like SAML and OAuth in the context of AI agents, advocating for a hybrid deterministic and non-deterministic system to manage agent behavior and ensure user control. Ultimately, the conversation suggests enterprises are likely to lead in agent adoption due to clear benefits in operating efficiency and the imperative to maintain business defensibility.
Sign in to continue reading, translating and more.
Continue