Okta just dropped $200 million to buy Permiso Security, a Palo Alto-based startup that specializes in detecting identity threats across cloud environments. The deal, announced on July 30, is squarely aimed at one of the fastest-growing attack surfaces in tech: non-human identities like AI agents, automated bots, and service accounts that operate across multi-cloud setups.
What Okta is actually buying
Permiso brings more than 2,500 research-driven detection signals sourced from over 70 identity partners. Its platform uses behavioral analytics to spot anomalies in how identities, both human and machine, interact with cloud infrastructure.
The startup also recently developed what it calls AI agent runtime security capabilities, including a dynamic sandbox environment branded SandyClaw. Think of it as a controlled testing ground where AI agents can be observed for suspicious behavior before they’re let loose in production environments.
Ely Kahn, Okta’s Chief Product Officer, said the acquisition would deliver “comprehensive identity threat protection” for enterprises operating with agentic identities.
