An agent gateway moves identity security beyond deciding who gets through the front door to governing what autonomous AI agents do once inside. By sitting in the path of attempted actions, the gateway can enforce authorization decisions at runtime.
Okta Inc. recently launched an AI agent runtime gateway as part of its broader approach to securing autonomous systems. The architecture marks a new role for the identity provider, according to Ric Smith (pictured), president of products and technology at Okta.
“That is a very different transition for Okta,” he said. “Dominantly, we had not been inline, so with us moving in that direction, it means that we can do a lot more decision-making in terms of what we authorize.”
Smith spoke with theCUBE Research’s Krista Case and co-host Rebecca Knight at Okta’s Oktane event, during an exclusive broadcast on theCUBE, SiliconANGLE Media’s livestreaming studio. They discussed runtime threat detection, intent-based authorization and how Okta is expanding further into cybersecurity. (* Disclosure below.)
Why an agent gateway puts Okta in the runtime path
Being inline also lets Okta watch for threats rather than simply grant permissions. That logic drove the company’s acquisition of Permiso Security, which adds detection capabilities as Okta expands further into cybersecurity, according to Smith. Full visibility into an agent’s behavior, however, requires observing both the prompt entering the model and the actions it proposes.
“You really have to sit in two positions around these models,” he said. “One is on the front end of it so that you can capture the conversation being presented. Then, based on the Gateway that we’ve already announced, we already sit on the back end of the model where we can capture all of the actions that the model ultimately wants to take and then also make decisions around that.”
Combining those vantage points could allow Okta to judge whether a proposed action should proceed, Smith explained. In 2027, the company plans to add intent-based authorization to its agent gateway, using an agent’s context and permitted tools to inform access decisions. Determining how narrowly to scope those permissions without blocking legitimate work remains an open research question.
“It’s a hard problem for us to solve, and so that’s why we’re hard at work doing all the research around it,” Smith said. “But we think that is pay dirt in terms of what actually has to happen with the agent workforce.”
Here’s the complete video interview, part of SiliconANGLE’s and theCUBE’s coverage of Okta’s Oktane event:
(* Disclosure: TheCUBE is a paid media partner for Okta’s Oktane event. Neither Okta Inc., the sponsor of theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)