AI agents are no longer tools—they’re coworkers. Companies now treat them like employees.

New startups and established firms are racing to build platforms that manage AI identities, permissions, and governance at scale. This shift demands more than patchwork solutions. It requires rethinking identity management from the ground up.

NewCore, a cybersecurity startup, just raised $66 million to tackle this exact problem. Their pitch: existing identity systems, built for humans, crumble under AI’s complexity. NewCore gives AI agents their own identities with lifecycle controls and permission sets. They don’t just treat AI as machine accounts. They treat them like first-class workers.

NewCore’s “split-key” architecture divides identity credentials between customer and platform to eliminate single points of failure. Their mobile app lets human employees grant, review, and revoke AI agent access—adding a crucial oversight layer as autonomy grows.

Meanwhile, NEWWORK launched its AI digital workforce platform to help enterprises deploy governed AI teams across ERP, CRM, HR, and service tools. NEWWORK’s system supports role-based access, separation of duties, and audit trails. It’s designed for regulated industries like finance and healthcare, where governance isn’t optional.

They emphasize that AI workers must operate within corporate structures—not outside them. Their “Agent Operating System” coordinates AI employees with delegated authority and approval workflows. NEWWORK sees this as a new software category: platforms that digitize work itself.

Atomicwork also entered the scene with a governed AI workforce platform targeting enterprise IT. Their AI “Coworkers” handle tasks like incident management and onboarding with scoped permissions and budget limits. Built on Microsoft Azure AI Foundry, Atomicwork layers onto existing ITSM systems rather than replacing them.

Atomicwork insists on full audit trails and spend controls to prevent runaway costs—a common problem in AI deployments lacking governance. Their approach competes with offerings from ServiceNow and Moveworks but claims a unique “governed AI workforce” positioning. Customer numbers and revenues remain undisclosed.

Behind these platforms lies a deeper security crisis. Cyera, a heavyweight in AI security, just secured $600 million at a $12 billion valuation to build what it calls the AI-native “trust layer.”

Cyera’s platform discovers and classifies data at massive scale with 95% precision. It unifies data security posture management, identity, privacy, and data loss prevention. Crucially, it monitors AI-generated actions in real time, flagging anomalies and preventing misuse.

The company warns that 68% of organizations will soon fail to distinguish human from AI actions—an untenable risk. Cyera’s trust layer aims to close that gap, offering enterprises essential visibility and control over AI behavior. Its growth mirrors the urgency enterprises feel about securing AI.

The stakes have never been higher. AI agents operate autonomously, with access to critical systems and sensitive data. Without robust identity and governance frameworks, companies face not just data breaches but unchecked AI behavior that could cripple operations.

Investors have noticed. Cyberstarts backs both NewCore and Cyera, signaling that AI identity and security form the backbone of the next wave of enterprise AI innovation. The convergence of identity management, governance, and security is no longer theoretical—it’s a battlefield.

Expect the number of AI agents in tech companies to surpass human employees soon. The challenge isn’t building AI agents. It’s controlling them before they run wild.