For architects, the harder question is where this control plane sits relative to existing IAM, API management, endpoint management and SOC workflows. If agents are treated as first-class identities, teams will need consistent lifecycle rules across human users, service accounts and autonomous agents, including joiner-mover-leaver processes, token rotation, audit retention and incident response. The value of a kill switch depends less on the button itself than on how quickly revocation propagates across clouds, SaaS tools, MCP servers and downstream sessions.
The Blueprint Alliance is potentially more significant than any single feature if it produces interoperable signals rather than another vendor-specific overlay. Enterprises should watch for practical evidence: shared schemas, cross-platform revocation flows, reference integrations and test results that prove one vendor’s detection can trigger another vendor’s containment.
Near-term, practitioners should inventory where agents already hold non-expiring credentials, identify unmanaged endpoint-based agents and map which business workflows rely on agent-to-agent delegation. Those are the places where governance failures will appear first, long before formal agent programs are fully standardized.
Identity and access management company Okta Inc. used the opening day of its Oktane conference in Las Vegas today to announce runtime enforcement and a wider kill switch for its Okta for AI Agents platform.
Okta also joined 11 other vendors, among them Amazon Web Services Inc., CrowdStrike Holdings Inc. and Google Cloud, in forming the Blueprint Alliance. The group is reworking the agent security framework Okta first published in March into an open, multivendor reference architecture.
The product additions are aimed at agents that end up with more access than anyone intended. An employee who links an AI assistant to everyday tools can give it a path into sensitive systems without realizing it, Okta said. If that person later leaves, nothing stops the agent from running on in the background, ungoverned.
“The challenge businesses face is the same access that makes agents powerful also makes them dangerous,” said Ric Smith, president of products and technology at Okta.
The runtime piece is Agent Gateway, which sits in the execution path between an agent and the tools it calls, enforcing policy and logging each interaction as it happens. Okta’s existing visibility comes from agent events captured in its System Log, which customers stream to security information and event management systems for review after the fact.
The gateway is also where the kill switch is headed. Administrators can already deactivate an agent from the admin console, which blocks new sessions. Once an agent routed through the gateway is deactivated there, Okta plans for every active token it holds to be revoked and every session in flight shut down.
Discovery now extends to employee laptops and desktops. Okta can already register known agents directly or import them from platforms such as Amazon Bedrock and Salesforce Agentforce, and shadow agents working through the browser get picked up too. A new feature, Shadow AI Agent Discovery for Endpoints, looks for unmanaged agents running on those machines.
Once an agent is on the books, the remaining additions govern what it connects to, whether that’s an app, a Model Context Protocol server or another agent. Agent SSO brings Cross App Access, which Okta debuted in June 2025, to all of its single sign-on customers so they can swap non-expiring keys for short-lived tokens tied to an identity.
Rules on which agents may call which other agents come through Agent-to-Agent Connections, with each handoff recorded in an auditable chain. Resource Access Certifications review agent connections over time to catch standing or excessive permissions. For administrators who would rather see the map, Configuration Designer draws agent-to-resource connections visually.
The Blueprint Alliance takes the framework behind those products outside Okta. Founding members also include Databricks Inc., Docker Inc., Lovable Labs Inc., Proofpoint Inc., Salesforce Inc. and ServiceNow Inc. Wiz Inc. has a seat of its own alongside Google Cloud, six months after Google closed its $32 billion purchase of the company. Zscaler Inc. is the 12th member.
“No single technology or vendor can secure the agentic era alone,” said Daniel Bernard, chief business officer at CrowdStrike.
Research from Gartner Inc. cited by the alliance Original Postress-releases/2026-04-28-gartner-identifies-six-steps-to-manage-artificial-intelligence-agent-sprawl" shape="rect">predicts that the average global Fortune 500 enterprise will have more than 150,000 agents in use by 2028. Only 13% of organizations think they have the right AI agent governance in place, Gartner said in the same release.
The March version of the blueprint was built around questions about where agents run, what they can reach and what they do. The alliance’s version adds how an enterprise should respond when an agent is compromised, with containment through token revocation, session termination or network quarantine and a staged, auditable path for restoring it afterward. Members have agreed on principles that include treating every agent as a first-class identity and scoping its access to the task at hand.
Beyond the principles, members are already testing interoperability across open standards, including MCP, the Open Cybersecurity Schema Framework and the Shared Signals Framework. The goal is for a threat signal raised by one member’s runtime monitor to trigger action across every connected control plane. Joint test results and reference integrations will be published on a regular basis.
GE Appliances and the nonprofit World Central Kitchen are serving as strategic advisers. Brian Stoll, chief technology officer at World Central Kitchen, said the organization is using agentic technology to support disaster response and needs governance “as dynamic as the agents themselves.” Taking part in the alliance, he added, might also bring “a better night’s rest for the CTO.”
Of the new Okta features, Agent SSO, Agent-to-Agent Connections and Resource Access Certifications are generally available today. Agent Gateway and Shadow AI Agent Discovery for Endpoints are planned for general availability in the third quarter, followed by Configuration Designer and the gateway kill switch in the fourth. More alliance members are expected to join over time.
Okta adds AI agent runtime gateway, forms Blueprint Alliance with AWS and CrowdStrike
Enjoyed this article? Sign up for our newsletter to receive regular insights and stay connected.