
Identity Decoded | The Identity Security Podcast
Winning in securing agentic identity: Start with one team, one use case
In this episode of Identity Decoded, recorded live from Identiverse, hosts Roy and Rob sit down with Sean O'Dell to unpack one of the most talked-about — and least understood — challenges in Identity Security today: agentic identity. Sean shares how his background across healthcare, entertainment (including a stint reverse-engineering an authorization system at The Walt Disney Company), and engineering shaped his product-first approach to identity and access management. The conversation dives into why identity must be a "first directive" baked into agents at creation, the difference between agent identifiers and agent identities, the lack of standards across providers, and Sean's practical advice for security teams overwhelmed by agentic adoption: pick one team, one use case, and deliver one verified agentic identity before trying to scale. They also cover AI security hype, the legal and liability questions raised by LLM decision-making, and why "IGA is the answer to all your questions" might be identity's biggest myth. Episode Highlights [01:29] Sean's background: from healthcare and retail to reverse-engineering an authorization system at The Walt Disney Company. [02:49] Why identity and access management is "the most intellectually complex space" and why companies increasingly want product-minded, business-fluent practitioners over narrow technical specialists. [05:42] Identity’s role in building the agentic AI revolution. [7:15] Sean's framework for why a stable identifier at birth matters even as an agent's role evolves. [12:15] The missing standard: fragmented approaches to agent identifiers (JWT, SPIFFE, WIMSE, or proprietary means) [20:10] Sean's practical starting point for overwhelmed teams: pick one team, one use case, and deliver one verified agentic identity before trying to scale. [32:23] Rapid-fire round: identity myths, leadership mistakes, hard truths, and the most overhyped trend in the space right now (spoiler: AI security). Key Takeaways: ● Takeaway 1: Identity has to be designed into agents at the moment of creation. Provenance and verification are needed at birth. ● Takeaway 2: Don't try to solve agentic identity all at once. Pick one team and one use case, get a single verified agentic identity right, and use that pattern to scale. Trying to build a perfect, overarching framework up front leads to paralysis. ● Takeaway 3: The industry's biggest current gap isn't identifiers. It's authorization and consent: knowing that an agent is only calling what it's supposed to call, and who's liable when an LLM-driven decision goes wrong. Guest Spotlight: Sean O'Dell, Distinguished Engineer at CVS Health Sean O'Dell leads identity and access management, consumer identity, and agentic identity/access security at CVS Health. His path into identity spans healthcare, logistics, and engineering, with an early career highlight reverse-engineering an authorization system at The Walt Disney Company. Known for a product-first, business-fluent approach to IAM, Sean is a vocal advocate for baking identity into agentic systems from the moment of creation rather than retrofitting security after the fact. He's active in the Identity Underground community and speaks regularly at industry events. In this episode, he brings a practical take on how practitioners can start solving agentic identity today rather than waiting for perfect standards. Stay Connected: ● https://www.silverfort.com ●https://linkedin.com/in/rob-ainscough ● https://www.linkedin.com/in/roy-akerman





