
Show Notes
Episode 422 is the debut of Decoded by Identity at the Center, a new sub-series hosted by Jeff Steadman and Sean O'Dell dedicated to unpacking the specifications and standards powering IAM. Joining them is Pieter Kasselman, VP of Open Standards at Defakto and chair of the WIMSE working group. The conversation covers why traditional non-human identity approaches break at agentic scale, how SPIFFE and SPIRE enable short-lived automated credential provisioning without long-lived secrets, and why treating agents as workloads unlocks a decade of existing standards. Pieter walks through critical OAuth specs including JWT authorization grant, token exchange, client ID metadata, and the emerging transaction tokens draft. Sean connects these to practical gateway architecture, continuous access evaluation, and policy-based authorization. The episode closes with real-world deployment examples and a clear takeaway: the tools to secure agentic identity are available today.
Episode Links:Pieter Kasselman: https://www.linkedin.com/in/pieter-kasselman-0259862/AI Agent Authentication and Authorization: https://datatracker.ietf.org/doc/draft-klrc-aiagent-auth/Workload Identity in Multi-system environments (WIMSE): https://ietf-wg-wimse.github.io/OAuth SPIFFE Client Authentication: https://datatracker.ietf.org/doc/draft-ietf-oauth-spiffe-client-auth/Transaction Tokens: https://datatracker.ietf.org/doc/draft-ietf-oauth-transaction-tokens/08/Agentic Identity Control Framework. You Already Have the Pieces. Now Build It. by Sean O'Dell: https://www.linkedin.com/pulse/agentic-identity-control-framework-you-already-have-pieces-o-dell-61b5e/
Timestamps: 00:00 Introduction to Decoded by Identity at the Center 00:13 The mission of the Decoded sub-series 03:02 Guest intro: Pieter Kasselman, VP of Open Standards at Defakto 06:21 Why agentic identity is urgent: scale, multi-platform, and shifting threat landscape 10:42 The real cost of API keys and credential sprawl in agentic systems 13:23 Agentic identity identifiers and how SPIFFE assigns unique workload IDs 21:00 Credential types: X.509, JWTs, and workload identity tokens 31:00 Connecting SPIFFE to OAuth and dynamic registration with client ID metadata 38:18 SPIFFE SVIDs, multiple credentials per agent, and governance traceability 41:44 Authentication versus authorization: delegation versus impersonation 47:00 Transaction tokens: binding access to specific transactions to stop token theft 51:21 Identity chaining and cross-domain authorization 55:00 Shared Signals Framework and dynamic authorization 57:00 Gateways, CAEP, and mid-flight token revocation for rogue agents 59:31 What you can deploy today with SPIFFE, OAuth, and existing IDPs 01:02:58 Policy-based access control and why instance-level governance cannot scale 01:04:58 Workload identity federation: Anthropic and Google Agent ID updates 01:07:13 Cross-platform federation and the law of agentic utility 01:11:55 Elevator pitch: agents are workloads and 95% of the problem is solved now 01:17:03 What is coming next: a transaction tokens deep dive
Keywords: agentic identity, SPIFFE, SPIRE, OAuth, transaction tokens, Shared Signals Framework, WIMSE, workload identity, non-human identity, authorization delegation, JWT, CAEP, API gateway, IAM standards, AIMS, Jeff Steadman, Sean O'Dell, Pieter Kasselman, IDAC, Identity at the Center, Jim McDonald, Decoded by Identity at the Center
Decoded by Identity at the Center:
Jeff Steadman: https://www.linkedin.com/in/jeffsteadman/
Sean O'Dell: https://www.linkedin.com/in/seanodentity/
Jim McDonald: https://www.linkedin.com/in/jimmcdonaldpmp/
Visit the show on the web at https://idacdecoded.com/
More from Identity at the Center
Most cloud security incidents trace back to the customer, and most of those trace back to identity. Jeff and Jim sit down with Mike Small, Senior Analyst at KuppingerCole, to unpack Cloud Native Application Protection Platforms (CNAPP) and why identity practitioners should care. Mike explains how identity controls in the cloud have become data defined in code, why ephemeral workloads break traditional scan-then-deploy security, and how CNAPP evolved from static configuration analysis into runtime monitoring and SOC integration. The conversation covers the shared responsibility model, SOC 2 attestations and complementary user entity controls, multi-cloud complexity, and why standards matter for portability and sovereignty. Mike also looks at CNAPP as a security plane for AI, agents as another layer of non-human identity, recent market consolidation, and the cloud as national critical infrastructure. Plus, a detour into amateur radio, Morse code exams, and low-power QRP contacts across the Atlantic.
Bojan Simic, co-founder and CEO of HYPR, returns to Identity at the Center for a Sponsor Spotlight conversation with Jim McDonald. Bojan traces HYPR's path from passwordless authentication to a broader set of identity assurance controls and explains how assurance differs from verification across the full employee lifecycle. He describes what organizations find when they look for worker impersonators, from remote employees outsourcing their jobs to North Korean operatives, and why verification has to continue past day one. The conversation then turns to AI agents: why IAM teams are becoming the HR of AI, how agent identity differs from human identity, why humans can't manually govern agents at scale, and why securing MCP alone misses the full context of an agent's session. Bojan also covers what is slowing passwordless adoption, adaptive risk, alert fatigue, and how identity teams and the SOC can work together. This episode was made possible with support from HYPR.
Eve Maler, founder of Venn Factory and author of Mastering Digital Identity, joins Jim McDonald and Jeff Steadman for her tenth appearance on Identity at the Center. Eve shares what she has learned since the book came out, from writing for non-technical executives (and keeping AI out of the manuscript) to the strange places identity ends up on the org chart. The conversation covers treating identity as a product, why identity can be a business model in disguise, how CISOs can influence CIAM when they are not the boss of it, and the identity debt that AI is now calling due. Eve also explains why intent looks more like purpose of use than finer-grained authorization, how contract language can help frame agent mandates, what trust means to humans versus security teams, and the early work on UMA for Agents (U4A) to govern third-party agents through a resource-centric, protect surface mindset. The episode wraps with travel tips from all three frequent flyers.
Shea McGrew, Chief Technology Officer at Maricopa County, returns to Identity at the Center one year after episode 376 to take on a question she first raised at our Identity After Dark session: when does digital identity become critical infrastructure? Jeff and Jim talk with Shea about how a county broadband and digital equity project led her to see digital trust as the thread running through every digital government service, and why that trust depends on identity. The conversation covers building an external identity portal for residents, designing for consent and privacy from the start, why residents extend trust to private platforms more readily than to government, the right to be forgotten, data governance across departments, and where decentralized identity and verifiable credentials might fit. Shea also shares her view on access reviews as a learning journey, what it would take to trust AI agents with certifications, and whether an agent could ever be considered a citizen. Plus, potatoes and the proper way to cut a sandwich.
Alex Bovee, CEO and co-founder of C1.ai (formerly ConductorOne), returns for his third appearance on Identity at the Center. Alex walks Jim and Jeff through the rebrand to C1.ai and why the .ai matters: identity now covers humans, workloads, and agents. The conversation opens on launch week and the new App Hub, built around a question many CIOs and CISOs are facing. What happens when everyone in the company becomes a builder and vibe-coded apps start running critical workflows?