Founded by the seller of Dome9 to Check Point, the former CIO of T Mobile USA, and the former CTO of Nym Health, the startup argues that legacy identity platforms built for human workers will fail once autonomous agents do real work.
For a year, companies have been issuing AI agents the digital equivalent of a permanent master key: long-lived API credentials with broad permissions, no real offboarding, no real audit trail. NewCore, a startup emerging from stealth on Monday, is betting that model is about to collapse, and it has a $66 million seed round at a $300 million post-money valuation to build what it argues is the first identity platform designed for a workforce where humans and software agents share the same permissions plumbing.
The pitch is less about the raise than the category. CEO Zohar Alon, who sold his previous cloud-security startup Dome9 to Check Point, frames identity platforms from Okta and Microsoft Entra as the wrong shape for the new work. They were built, he told TechCrunch, for tens of thousands of human employees and a small population of service accounts. Agent identities, in his view, will be orders of magnitude more numerous, more autonomous, and more consequential, with a blast radius an old-style service account never had.
"We know for sure that the scale and the complexity that those things [AI agents] are going to add to 15- or 20-year-old identity platforms are going to break them," Alon said, per TechCrunch's launch piece.
What NewCore is actually selling is the same plumbing a human employee gets on day one, day ninety, and last day, but built for software. That means a distinct identity per agent, scoped permissions that match the agent's job, rotation and monitoring, an audit log of what each agent touched, and a clean way to revoke access the moment the agent is retired or goes off-task. The company also describes a "split-key" architecture that divides the most sensitive credentials between the customer and the platform so a NewCore breach alone would not be enough to impersonate an agent, plus an "Agentic Skill" integration package that lets coding assistants from Anthropic, OpenAI, and Cursor pick up managed identities inside the customer's environment.
The market context the company leans on is real, even if the numbers need sourcing. Goldman Sachs said last year it was testing the AI coding agent Devin "as a new employee," and McKinsey has described a workforce of roughly 25,000 AI agents working alongside its 60,000 human employees. Both claims are currently in circulation via the companies' own framing, and a serious buyer of the category thesis should ask what "as a new employee" actually meant in practice: a sandbox experiment, a scoped production task, or a genuine headcount substitution.
The round itself is unusual — a $66M seed that reads more like a Series A in everything but the label. With more than 50 employees already on the payroll, the company is describing the raise as a seed, but the size and the team footprint are more typical of a late seed or Series A. Readers should treat the round name as the company's marketing rather than a statement of fact. The founders' prior exits — Alon's Dome9 sale to Check Point and CTO Amihai Neiderman's Nym Health — are more useful indicators of how this category bet will be priced and sold to enterprise security buyers.
CRO Erez Yarkoni, the former CIO of T-Mobile USA and Telstra, is the third founder, and the one whose customer-side résumé raises the real go-to-market question: whether large enterprises will buy agent identity as a standalone line item, or treat it as a feature they expect their existing identity vendor to ship.
NewCore's own site repeats three headline statistics: that more than 70% of attacks are rooted in identity, that more than 75% of enterprises are deploying AI agents, and that more than 80% lack real governance for those agents. The numbers are useful for signaling the category but are not independently sourced, and should be read as company claims rather than market facts.
The open question is not whether agents need identity. They do, and that part is no longer in serious dispute. The question is whether the answer is a new category, or a feature on a category that already exists. NewCore's bet, and the $300M post-money attached to it, is that the load of real, autonomous, employee-shaped work will be heavy enough to break the old platforms. Whether the rest of the market agrees will be decided by which vendors can show real revocation, real audit, and real offboarding for an agent workforce in the next 18 months.