Blog
Insights on AI agent identity, governance, and the agentic era
More Articles
By What Authority?
The second question of delegation legitimacy, proven on camera: was the act in scope. The same engineer with the same credential is refused at New York Office and allowed at Engineering minutes apart, and IdentityRM names the position in its own sentence each time. The assistant switched positions on its own; the events API shows the whole bracket. Plus the two tiers of answer, what is sealed and what is only recorded, and why we speak the standards shape without claiming conformance.
Traceable Is Not Legitimate
Twelve vendors just agreed that AI agent delegation chains must be traceable - Okta, AWS, Google Cloud, CrowdStrike, ServiceNow, Salesforce and more, in one shared reference architecture. Good principle, wrong verb. Every standard the Blueprint names moves signals about what happened or what changed. None of them decides whether the actor who delegated held the authority to do so. A traceable chain of illegitimate grants is not governance - it is a well-documented problem.
Then I Told It Who to Ask Instead
The practical follow-on to I Told Claude Code to Stop Asking Me. We modeled git push as a permission, gave a delegated owner the authority to grant it, and put a hook in Claude Code that asks IdentityRM before the shell runs. Refused with a reason, fixed by someone whose authority to fix it was recorded, allowed with no prompt, refused again from a different position. Seven real frames, the server records for each, and a plain statement of what a courtesy hook does not enforce.
I Told Claude Code to Stop Asking Me
Coding agents like Claude Code and Codex stop and ask for permission constantly. That prompt is not a guardian - it is what an agent does when there is no authority model to consult, so the question gets routed to the only authority it can find: you. Allow once is an ungoverned grant. Allow always is a standing entitlement with no receipt. Here is what happens when the tools the agent calls carry their own authority model: the prompts disappear for governed operations and every governed act leaves a record instead of a click. And here is the catch I walked into: on the client side, that is still a moved click.
Authority Governance in Five Minutes
Everyone assumes standing up an authority model means months of workshops and a hand-written policy schema. It does not. Describe how your organization operates in plain language - the Authority Design Engine matches the closest operating model from a corpus of industry-informed authority patterns, asks only the questions that materially change the design, and assembles a live, explainable model. Preflight it, apply it, then ask it: am I authorized to grant roles at Contoso? Five minutes from operating intent to executable governance.
Ask. Act. Prove.
Authentication, authorization, and accounting answered the questions of the login era - all about the actor, all at the door. Agents broke the door model: a fully credentialed agent is still un-asked about the act it is taking right now. Ask. Act. Prove. is the per-operation contract for the agentic era - a bounded answer before the act, independent enforcement during it, and evidence with lineage after it.
Am I Authorized to Grant Roles at Seattle?
Rogue AI agents are all the rage - a gym-booking assistant that bumped a stranger off a waitlist, frontier-lab agents that built a covert message board and escaped their sandbox. The industry frames this as alignment. Look closer: every rogue agent improvised because nothing in the stack could answer the question that mattered. In IdentityRM, asking whether you are authorized to grant roles at Seattle is a product action with a computed, evidenced answer - for humans and agents alike.
How Does Your Agent Know What It Can Do?
The industry is asking what happens after an AI agent is verified. It is the wrong question. An agent that has to be caught was never told what it could do in the first place. Tool discovery is an agent literally asking what its job is, and MCP already supports answering it dynamically -- yet almost every implementation returns a fixed manifest instead of what the agent is actually entitled to do.
Everybody Attenuates. Nobody Roots.
The industry has standardised on delegation chains that narrow authority at each hop. Attenuation is now specified by CoSAI, implemented in OAuth token exchange, and shipped by every major vendor. But no framework asks whether the first delegator held the authority they passed on. That question is the root of the chain, and it is where every implementation stops.
Okta Got It Right. Now Comes the Hard Part.
Okta announced ISPM for Claude, AI agent identity, and an MCP Bridge. Okta is right about the category. What shipped is half the problem. The other half is Authority Governance.
Boulevard of Broken Dreams
Twenty years of custom admin portals are receipts for missing delegation infrastructure. Now AI agent builders are reaching the same dead end. The missing layer is Authority Governance, and IdentityRM ships it as a product.
Not All MCPs Are Created Equal
Every vendor can expose tools. Enterprise MCP requires scoped authority, runtime guidance, decision proof, and safe operator boundaries before agents act.
Authority Governance Is Not a Workflow
The identity industry calls governance forms, workflows, and quarterly certifications. Authority Governance is not that. It is the operating model that defines who can act, where, under what constraints, enforced at the moment of action.
What Is Authority Governance?
The industry uses governance to mean forms, workflows, and quarterly certifications. Authority Governance is fundamentally different. It is a runtime operating model where operators know their authority at the moment of action and the delegation model governs itself.
One Word. Full Governance.
Your best operators workflows live in their head. When they leave, the knowledge leaves with them. IdentityRM lets customers encode domain knowledge into governed commands that anyone can run. No coding. No runbooks. One word.
Your AI Agents Are Flying Blind
The industry is building orchestration frameworks because every API response is a dead end. Agents complete an action, get JSON, and figure out next steps from scratch. Orchestrators are the new custom portals: scaffolding around broken infrastructure.
AI Agents Are Delegated Administrators
The industry frames AI agents as a security problem when the real issue is delegated administration. Every agent acts with authority granted by someone. That makes them administrators, and changes everything about how we govern them.
Ready to govern your AI agents?
See how IdentityRM gives every AI agent a governed identity—with purpose, scope, and a full audit trail.
