Here's a detail most people miss when they hear "AI agents at work": the problem isn't that companies are deploying AI. It's that nobody has figured out how to give those agents a legitimate place in the identity systems that already run every enterprise.
Zohar Alon, NewCore's CEO and a veteran who founded cloud-security startup Dome9 before it got acquired by Check Point, put it bluntly. "We know for sure that the scale and the complexity that those things are going to add to 15- or 20-year-old identity platforms are going to break them," he told TechCrunch.
Think about that for a second. Okta, Microsoft Entra — the identity providers every Fortune 500 company relies on to control who gets into what — were designed for humans. Badges, passwords, MFA prompts. Now you're asking them to manage thousands of autonomous software workers that never sleep, never call in sick, and multiply faster than any HR team can hire for.
Alon's epiphany came in 2023 while reviewing a company's technology budget. He saw the identity platform bill, assumed the customer was thrilled with the product, and asked the wrong question: "You must be extremely happy with them." The answer? "No, I'm not."
That's the market NewCore is walking into — a massive, stagnant identity space where incumbents face limited competitive pressure and customers are quietly unhappy. Okta and Microsoft have started adding agentic capabilities, sure. But as Alon puts it, "The traditional vendors give you an agentic way to deal with identity, but it's on the side — it's not integrated."
It's bolted on. Not built for.
Treating AI Agents as Employees, Not Service Accounts
This is where NewCore actually gets interesting.
The industry standard for managing non-human identities has always been service accounts — clunky, over-privileged machine credentials that exist in a gray zone between "person" and "thing." Nobody really owns them. Nobody revokes them on a schedule. They accumulate permissions like dust in attic corners until someone accidentally grants a script admin access to the production database.
NewCore's answer is to treat AI agents as first-class identities. Full stop. Each agent gets its own permissions, lifecycle controls, and revocation mechanisms — the same toolkit humans get when they onboard. Not a service account with extra steps.
The architecture behind this is called "split-key," and it's worth understanding because it solves a real problem. Critical identity credentials get divided between the customer and NewCore's platform, eliminating any single point of compromise. If one side gets breached, the attacker walks away empty-handed. It's a design choice that signals NewCore takes the threat model seriously.
Then there's the Agentic Skill integration package — a practical bridge between NewCore's identity layer and the actual AI tools people use every day. Anthropic's Claude Code, OpenAI's Codex, Cursor: these coding assistants can now access enterprise systems as managed identities rather than through manually distributed credentials that someone typed into a config file at 2 AM.
And here's the human piece most people overlook. NewCore built a mobile app that lets employees grant, review, and revoke access for AI agents themselves. Not IT admins. Not security teams. The actual humans who'll be working alongside these autonomous systems. It's a human oversight layer, and it matters more than you'd think — because the people doing the work are usually the ones who understand what an AI agent actually needs access to. For deeper insights into how human oversight shapes AI agent security, see Gartner's framework on guardian agents.
The Numbers Behind the AI Workforce
You don't build an identity platform for a hypothetical problem. You build it because the math is already breaking.
Goldman Sachs tested Devin — the autonomous AI coding agent from Cognition — as a new employee last year. Not as a tool. As an employee. That's not a thought experiment anymore; it's a pilot program at one of the world's most data-driven firms.
McKinsey reported earlier this year that 25,000 AI agents already work alongside its 60,000 human employees. That's a near 1:2 ratio of software workers to humans at a firm that prides itself on operational rigor. And TCS chairman N. Chandrasekaran has publicly predicted AI agents could eventually rival the Indian IT services giant's workforce in size.
Alon predicts something similar for tech organizations broadly. AI agents could outnumber human employees at many technology-focused companies within a few years.
Let that sink in. When the ratio flips — when there are more AI agents than humans on the network — identity stops being an HR problem and becomes a security infrastructure problem of unprecedented scale. Every agent needs authentication. Every agent needs authorization boundaries. Every agent needs a way to be revoked when it goes rogue, gets compromised, or simply becomes obsolete.
"It's inevitable," Alon said. "The question is whether we're going to build the guardrails in time."
For a comprehensive look at how unmanaged agent lifecycles create security gaps, see The Lifecycle Crisis: Managing the Birth, Life, and Death of AI Agents.
The Money and the Team
NewCore emerged from stealth with $66 million in seed funding, led by Cyberstarts — a cybersecurity-focused venture firm that clearly sees the writing on the wall. Index Ventures and Evolution Equity Partners also participated, valuing the company at $300 million post-money.
That's a serious bet on a category that didn't exist as a standalone market six months ago.
The founding team has the right mix of enterprise security, intelligence, and commercial experience for this moment:
- Zohar Alon (CEO) — Founded Dome9, which Check Point acquired. Deep enterprise security DNA.
- Amihai Neiderman (CTO) — Former Unit 8200 research leader (Israel's elite signals intelligence unit) and founder of healthcare AI startup Nym Health. The technical depth here is unusual for a seed-stage company.
- Erez Yarkoni (CCO) — Previously CIO of T-Mobile USA and Telstra. He understands how to sell identity infrastructure to enterprises that have been buying it for decades.
The team has grown to more than 50 employees across the U.S. and Israel — reasonable for a seed-stage cybersecurity startup with this kind of funding.
The customer picture is still early. Fewer than 10 customers, plus more than 10 design partners. NewCore expects to begin charging this summer (2026). That's a tight timeline, but it makes sense: the design partners are effectively stress-testing the product before the paying customers arrive.
What Comes Next
The identity market is old. The problem NewCore is solving is new — not because AI agents are brand new, but because the scale at which enterprises are deploying them has crossed a threshold where legacy systems literally can't handle it.
NewCore's thesis is simple: if companies are going to treat AI agents as workers, they need worker-grade identity infrastructure for them. Not service accounts with extra steps. Not a bolted-on module on Okta's platform. A system built from the ground up for a workforce made up of humans, machines, and AI agents operating together.
The split-key architecture, the Agentic Skill integrations with Claude Code and Codex and Cursor, the mobile app for human oversight — these aren't features. They're the foundation of a category.
Whether NewCore wins that category is another question entirely. Okta and Microsoft aren't going to sit still. But NewCore has something incumbents can't easily replicate: a platform designed for this moment, not retrofitted for it. And at $300 million valuation with $66 million in the bank, they've got enough runway to prove it.
The identity layer for the AI workforce is going to exist. The question Alon keeps coming back to — "whether we're going to build the guardrails in time" — is really a question about execution. NewCore just bet $66 million that they can build those guardrails fast enough.