By Mukundan Sivaraj · AIM Media House
At one company evaluating Okta's AI-agent security tools, the number of Claude agents in its environment jumped from 50 to 1,500 in a few weeks.
Todd McKinnon, CEO and Co-Founder of identity provider Okta, shared the example during the company's Q2 FY2027 earnings call as it described the demand emerging around AI agents. Okta reported $805 million in revenue for the quarter ended July 31, up 11% year over year, while subscription revenue rose 12%.
Current remaining performance obligations (cRPO) increased 14% to $2.585 billion. Okta also reported dozens of AI-agent deals, including several worth more than $1 million, although executives said the revenue contribution remains immaterial.
The company's existing workforce and customer identity products still account for most of its business. AI agents are adding a new requirement: organizations need to identify these systems, control their access and track what they do after they are deployed.
AI Agents Are Becoming an Identity Problem An employee accessing an application has an established identity and access policy. An AI agent can be created for a user, connect to multiple applications and continue carrying out tasks after the initial interaction.
Okta is building its AI-agent products around three questions: where agents are, what they can connect to and what they can do. Its Okta for AI Agents product became generally available on April 30, and McKinnon says the company shipped 24 significant enhancements during its first two months.
The issue is also appearing in standards work. The National Institute of Standards and Technology (NIST) launched its AI Agent Standards Initiative in February, covering interoperability and security for AI agents.
NIST is separately examining identity and authorization as agents gain access to data, tools and applications.
Continue on AIM Media House