Workday Bought C1.ai to Give AI Agents a Login. Here's What Agentic Identity Management Actually Is

Workday is an HR software company. On Monday it announced a definitive agreement to buy C1.ai, an identity management company — not because Workday cares about logins, but because something in your company is using logins now that didn't exist three years ago: AI agents. The deal, announced at Workday Rising 2026, should close by January 31, 2027. Terms weren't disclosed, but the reasoning is plain: companies now run on "a combined workforce of people and AI agents," and most of them have no idea what their agents can touch.

Here's the number that made this acquisition happen: C1.ai's research says 95% of large U.S. companies already run autonomous agents, while Deloitte's State of AI research says only 21% have a mature way to govern them. That gap — between the agents in production and the governance that was supposed to contain them — is what the industry has started calling agentic identity management. This week it got its first big-name buyer.

What Workday actually bought

C1.ai is six months into a new name. It was ConductorOne until April 2026, an identity governance company founded by Alex Bovee, who used to run security products at Okta. It raised $79 million last fall from Greycroft, Accel, and Felicis, and spent the summer building what it calls an Agentic Control Plane — tools to discover AI agents running in the shadows of your company, store their secrets in a vault that never hands the raw credential to the agent, and govern what each agent is allowed to do at runtime.

The pitch Workday gave: most organizations manage employee access across hundreds of applications through manual service tickets, spreadsheets, and periodic reviews "as workers join, move, and leave." AI agents, the argument goes, now need the same treatment — "giving each person and agent the right access at the right time and removing it the moment it's no longer needed."

Join. Move. Leave. That's HR language: "Every new hire, job change, and exit is a Workday event." Your agents get onboarded, reassigned, and offboarded like staff.

The problem: your agent logs in as you

Here's how most AI agents get their access today. A finance team turns on an autopilot that posts journal entries. The agent needs to read your CRM and approve a refund — so it gets handed a service account, an API key, or worse, the employee's own credentials. It inherits an identity designed for a human being.

That creates three problems. First, privilege sprawl: the agent gets a key with far more access than any employee would be allowed. Second, standing credentials: a human's access gets reviewed quarterly, but an agent's API key sits in an environment variable for two years — and when the project ends, there's no offboarding for something nobody remembers creating. Third — attribution. When your agent approves the wrong invoice at 3 AM, the audit log says you did it. There's no way to distinguish the agent from the person it acts for. As CrowdStrike's Scott Kriz put it: "you cannot continuously authorize an identity you were never able to establish." Traditional identity providers were built for people and predictable workloads, not for software that reasons, acts, and delegates to other agents at machine speed.

What agentic identity management actually does

Strip away the marketing and it comes down to four mechanics, borrowed from how good companies already handle human employees.

One: every agent gets its own identity. Not a borrowed service account or a shared API key — a first-class, cryptographically verifiable identity with a defined owner, recording what created the agent, what it's allowed to do, and who's responsible. Some are long-lived (the agent running nightly reconciliation); many are per-execution — one identity for one task, discarded when it finishes.

Two: credentials that die on schedule. Agents never get standing credentials of their own. The identity provider brokers short-lived, tightly-scoped tokens — just enough access for the task at hand, and nothing after. C1's version authenticates the agent without ever handing it the raw secret, so there's nothing to leak or forget to rotate. The industry shorthand is "zero standing privileges."

Three: the full employee lifecycle. Join, move, leave — for agents. Created, it gets provisioned with defined roles; reassigned, its permissions get re-scoped; retired, every token gets revoked. Workday already runs join-move-leave for 11,500 organizations' human workforces — extending that machinery to agents is a smaller leap than building it from scratch.

Four: discovery, because most agents sneak in the side door. Many agents never touch your identity infrastructure — they enter through an endpoint, a SaaS integration, or a vibe-coded internal app someone shipped on a Friday. The governance conversation can't start until you know the agents exist, and right now, most companies don't.

Why now, and why an HR company

October brought Workday opening its own Trusted Agent Platform to be "headless" — any agent that speaks Model Context Protocol can now reach Workday's HR and finance rules without opening Workday. Agents are moving from chatbots that answer questions to operators that execute payroll, post accounting entries, and run hiring workflows. The moment software starts moving money and making personnel decisions, someone has to answer for each action.

And there's a deeper reason identity is becoming the control point. When every actor in the stack is software, the only stable handle you have on any of it is identity. You can't review the judgment of something that makes decisions at machine speed. You can only control what it was allowed to touch, and prove what it did.

The honest counterpoint: this is also a land grab. "Agentic identity" is a category every identity vendor is sprinting to define, and Workday buying C1.ai is as much about owning the vocabulary as owning the technology. The 95%-vs-21% stat comes from C1.ai's own research — read it with that in mind. But the underlying direction is real. The question isn't whether agents need identities. It's whose directory they'll live in.

What to do about it this week

You don't need to buy anything. But if your company runs AI agents — and the odds say it does — three things are worth doing before the auditors ask:

• Inventory your agents like employees. Which agents exist, who owns each one, what credentials do they hold? The ones nobody can name are the ones with standing API keys.
• Kill the standing credentials. Any agent authenticating with a permanent API key or shared service account is the highest-risk item on your list. Scope it down to short-lived tokens for the task it performs.
• Make attribution possible. If your agent acts under your name, change that. Give it its own identity, even a crude one, so the audit log can distinguish its actions from yours.

The boring truth about this acquisition is also the interesting one. The identity industry spent thirty years figuring out how to govern humans joining, moving, and leaving companies. AI agents just inherited all of that machinery — because the hardest problem in enterprise AI was never intelligence. It's knowing who did what, and being able to take the keys back.

References

• Workday to Acquire C1.ai, a Leading Agentic Identity Platform for People and Agents — Workday Newsroom, October 13, 2026
• Workday Signs Deal to Acquire Agentic Identity Platform C1.ai — Unite.AI, October 13, 2026
• Workday Launches Headless HR and Finance for Any Agent, Agrees to Buy C1.ai to Govern Them — Shashi Bellamkonda, October 13, 2026
• C1 – formerly known as ConductorOne — to be acquired by Workday — Silicon Florist, October 13, 2026
• CrowdStrike Announces Agentic Identity Provider — CrowdStrike, September 2, 2026
• AI has created a new identity problem. Agentic IAM is how we solve it. — VentureBeat, September 2026
• Agentic AI and Non-Human Identities Demand a Paradigm Shift in Security: Lessons from NHIcon 2026 — Security Boulevard, January 2026