Your Okta tenant is your weakest agent-era link
This is not an Okta criticism. We use Okta. The argument is structural: a product designed for the era when humans logged into apps cannot be the trust root for the era when their agents act on their behalf.
The structural mismatch
Okta's architecture assumes one human, occasionally accessing apps via a session, with the session as the unit of authorization. The enterprise has 100 non-human identities for every human, agents acting hundreds of times per session, and the session as the unit of authorization is too coarse to reason about.
You can bolt agent extensions onto Okta. Many enterprises have. The bolt-on satisfies a checkbox; it does not change the fact that the audit log, the policy engine, and the customer success motion are all human-shaped.
Where Okta fails the agent test
- Audit logs index by user; agent attributions are second-class.
- Sessions, not delegations: scope cannot narrow per agent action.
- Tenant lock-in: the human's identity dies when they leave the company.
- No work attestation primitives: Layer 3 is out of scope.
- Revocation propagation is built for app-level access, not 200ms agent-level kill.
Why this is the weakest link
"Weakest link" is not "the weakest control." It is the link where the consequence of failure is highest and the architectural fit is lowest. In an agent fleet, that is the identity layer above the agent. Get that wrong and every other control downstream — network security, data classification, audit logging — operates on the wrong assumption.
What we do instead
Run Okta for human SSO. Run Manav for the human-to-agent boundary. The two compose: Manav reads Okta sessions as a Layer 1 anchor. Manav handles delegation, attestation, score, and settlement — the layers Okta wasn't built for. Each protocol does the job it was designed to do.
The Okta angle on this
Okta is a $15B enterprise that does its job well. Recommending against it would be silly. Recommending it as the agent-trust root is a different mistake — one Okta itself has not made in writing. Their roadmap for "agent support" is feature-shaped; the structural answer is not their job to ship.
Common objections
The strongest counter-arguments we have heard. The incumbent will catch up — possibly inside their boundary; the cross-platform shape is architecturally hard for them. The category is too narrow — we believe it broadens as agent autonomy compounds; we may be wrong; the data over the next year will tell.
Frequently asked questions
What are the strongest counter-arguments? The two we hear most: (1) the incumbent will eventually ship this, and (2) the category is too narrow to support a category-defining company. We address both head-on; we believe the incumbent's architecture cannot ship this without a rebuild, and we believe the category broadens as agent autonomy compounds.
Are we ignoring legitimate criticism? We try not to. The honest criticisms — slow adoption, immature SDKs in some languages, unclear regulator response — are documented openly. We answer with progress, not with marketing.
What would make us change our mind? Three signals. A major incumbent shipping a comparable cross-platform delegation primitive. A regulator explicitly preempting the category with a different spec. A customer cohort showing they prefer the platform-bound alternative even when the audit trail is broken. None of those have appeared.
Where to start
For the steel-manned counter-position, read manav vs okta. For the alternative we agree could win, see agent identity is a category. We do not need to be right for the category to be real.
Adjacent reading
For the steel-manned counter-position, see Manav vs Worldcoin and the vendor map. For the alternative that could win in another universe, see the AgentKit thought experiment. We do not need to be right for the category to be real; the contrarian read is at least worth the disagreement.
What CISOs we work with quietly admit
In off-record conversations with CISOs across the Fortune 500, the same admission surfaces. The Okta deployment is the most secure piece of their stack and the most concentrated risk in their stack at the same time. They have added MFA, conditional access, behavioral analytics. They have not addressed the structural fact that the entire identity system funnels through one vendor, one signing infrastructure, and one breach-class away from a re-authentication wave they cannot manage. The CISOs who say this on-record are vanishingly few. The CISOs who say it off-record are most of them. The fix is not replacing Okta; it is adding portability to the identity layer above Okta, so that an Okta breach degrades rather than collapses the trust posture. Manav is one shape of that portability layer. Other shapes will emerge. The CISO who hedges this risk before the breach is the CISO who keeps their job through it.
The weakest link is not the weakest control. It is the misfit at the layer that matters most.