Okta buys AI security startup Permiso; source says for about $200M
TechCrunch Jagmeet Singh
Okta is buying AI security startup Permiso for roughly $200 million, almost all cash. It's chasing a new problem: keeping tabs on rogue AI agents, not just human logins.
Based on reporting by TechCrunch, Jagmeet Singh — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Okta announced Thursday it's acquiring Permiso Security, a three-year-old startup that watches what happens after someone (or something) gets into your cloud systems. Neither company put a number on the deal publicly, but a source told TechCrunch it's just under $200 million, paid almost entirely in cash, and Okta didn't push back when asked. The deal is expected to close in Q3 of Okta's fiscal 2027.
The logic here is straightforward once you think about where identity security is heading. For years, companies like Okta made their money verifying who you are at the login screen. That's no longer the hard part. The hard part is what happens after — especially now that AI agents are wandering around corporate networks with their own credentials, making decisions, touching data, sometimes acting on behalf of a human who isn't even watching. Permiso built its business around exactly that gap, flagging suspicious behavior from compromised or stolen identities inside cloud environments, and earlier this year it launched SandyClaw, a sandbox tool built specifically to test whether an AI agent's skills might be used maliciously before anyone lets it loose in production.
Permiso was founded by Paul Nguyen and Jason Martin, both FireEye alumni, and has raised a modest $29 million total, including an $18.5 million Series A in April 2024 led by Altimeter Capital that reportedly valued the company around $80 million. A $200 million exit less than two years later is a solid multiple, and it says something about how quickly the market for machine-identity security has heated up.
Okta's chief product officer Ely Kahn framed the acquisition as bolting proven threat detection and a strong research team onto Okta's existing identity infrastructure. Read between the lines and it's really about not getting left behind: as more of what happens inside a company is decided and executed by autonomous software rather than employees, the vendors who can prove they're watching those agents closely will be the ones enterprises trust with the keys.
My take — AI-written commentary, not fact-checked reporting
This is less a bold AI bet than Okta buying insurance against irrelevance — if agentic AI takes off the way vendors keep promising, 'who logged in' stops mattering and 'what did the agent do next' becomes the whole ballgame. $200 million for a three-year-old startup with $29 million raised is a steep multiple, but in this cycle that's basically a rounding error, and I'd bet we see three more of these machine-identity acquisitions before the year's out.
Read more about this at: TechCrunch