Why are organizations delaying AI agent implementation?

“The guardrails for authorization in agentic AI flows are almost non-existent today.” — Jagadeesh Kunda

As AI agents move from experiments to production, identity and access management is being tested in ways traditional models never anticipated. In this episode, Ravi sits down with Jagadeesh Kunda (Co-founder & CPO, Oleria) to unpack how the rise of non-human identities and agentic workflows is reshaping security, governance, and developer priorities.

You’ll come away with insights on:

• Why AuthZ guardrails are lagging behind in AI agent flows, and what CIOs/CISOs are most worried about.
• How autonomous vs. automated governance differs—and why real-time evaluation is now table stakes.
• The blurred lines between workforce IAM, customer IAM, and non-human IAM, and what that means for product teams.
• Practical challenges of data exfiltration, impersonation, and continuous access evaluation in AI-first environments.
• Why identity is a business-first problem that AI alone can’t solve—and how engineering leaders should rethink their approach.

Packed with real-world anecdotes and forward-looking perspectives, this conversation is a blueprint for leaders navigating the messy, fast-evolving intersection of AI, identity, and enterprise security.

If you’re looking to dig deeper into how these ideas play out in practice, check out our developer resources curated for the agent-first workflow era.