AI agents are reshaping the enterprise, but they are scaling faster than the security around them. This creates a new identity gap, leaving many organizations unable to answer three basic questions: Where are my agents? What can they connect to? And what can they do?
Without these answers, organizations can't securely scale their agentic workforce, much less stop a rogue agent. Today, Okta for AI Agents is now generally available, giving companies a single control plane to discover, protect, and govern all AI agents.
Early adopters like Yahoo and Ramp are already using Okta to innovate securely. They are laying the groundwork to ensure that as they deploy more AI agents, security is built in from the start, not added as an afterthought.
At Yahoo, it starts with strong governance
For Yahoo, the goal is to create a consistent security model for both humans and agents. This approach simplifies access and strengthens governance across their entire digital ecosystem.
"Making sure that you have strong governance in place — it’s critical," says Bryan Meister, Sr. Principal Architect at Yahoo. "It’s not, ‘We can get to this later.’ You have to start there. You have to be building with that in mind.”
Without that central governance, the risk an agent poses is directly tied to the applications it can access, according to Meister. An ungoverned agent with access to a code repository or a documentation store creates a significant attack surface.
"You can't manage what you can't measure," Meister says. "If you don't have effective control or visibility over what an agent is doing, your risk level is going to increase."
This need for visibility and control is why Yahoo is building its AI strategy on a foundation of identity. By treating every agent as a first-class identity, the company can establish a single source of truth for governance.
Ramp introduces ‘painless’ identity controls
Ramp, a leader in finance automation, is deploying AI agents to help customers and employees save time and money. With agents handling sensitive tasks, the company recognized the need for strong identity controls that enable productivity without introducing friction. Instead of creating barriers, Ramp’s philosophy is to provide a "paved road" for employees, giving them a simple, secure way to access the tools they need.
"Okta for AI Agents is all about secure enablement,” says Cameron Leavenworth, a Staff IT Engineer at Ramp. “We get a clear view into what people are doing and can block risky connections, while making sure they have painless access to the tools they need."
This approach is critical for managing the audit and compliance risks that come with AI. One of Ramp’s core security principles is ensuring every autonomous action can be traced back to a person. This prevents a scenario where an agent operates without clear ownership.
"An agent is an identity taking action," Leavenworth says. "You can't just let it run without actually knowing who was telling it to do the thing." By starting with an identity-first approach, Ramp can ensure that every agent has a clear human owner and operates within defined, auditable boundaries.
An ongoing journey
For the vast majority of organizations, the path to becoming a secure agentic enterprise is ongoing.
But as Bryan Meister of Yahoo notes, the most promising solutions are built through partnership and collaboration. "I know that Okta is listening, and we're having deep conversations every single day about AI security and how Okta can help us solve it."
Learn how customer feedback helped build Okta for AI Agents in the next episode of The Secure Agentic Enterprise.
