$66M Bet Tests AI Agent Identity Before NewCore Charges

NewCore raised $66 million before it has even begun charging customers, betting that AI agent identity will become one of enterprise security’s first major stress tests in the agentic era. The startup emerged from stealth on Monday with a platform built to authenticate, govern, and revoke access for both humans and AI agents, according to TechCrunch.

That funding figure matters because NewCore isn’t pitching a better login screen. It’s arguing that companies are rushing to “hire” AI agents before they know how to restrict them, investigate them, or fire them. The company’s thesis is blunt: identity systems built around employees, apps, and service accounts will struggle when software workers can access tools, trigger workflows, write code, and act across enterprise systems.

NewCore’s seed round was led by Cyberstarts, with participation from Index Ventures and Evolution Equity Partners, and valued the company at $300 million after investment. For a startup with fewer than 10 customers and more than 10 design partners, that valuation says investors are buying the category before the market fully forms.

NewCore’s $66M AI agent identity bet starts with a simple fear: agents need offboarding too

The strongest version of NewCore’s case is not that every AI agent needs a username. It’s that every agent needs an accountable identity, a defined owner, narrow permissions, and a clean revocation path.

That is where AI agent identity becomes different from conventional identity management. Employees have managers, HR records, job titles, onboarding flows, and offboarding dates. Agents may have none of that unless companies build it deliberately.

NewCore co-founder and CEO Zohar Alon, who previously founded Dome9 before its acquisition by Check Point, told TechCrunch that AI agents will strain older identity platforms.

“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.

NewCore was co-founded by Alon, CTO Amihai Neiderman, a former Unit 8200 research leader and founder of Nym Health, and chief revenue officer Erez Yarkoni, former CIO of T-Mobile USA and Telstra.

The timing is not theoretical. TechCrunch cites Goldman Sachs testing AI coding agent Devin last year as a new employee, while McKinsey said earlier this year that 25,000 AI agents already work alongside its 60,000 employees. Those examples show why NewCore is framing agents as workplace participants, not background automation.

The numbers behind the AI agent identity problem: $66M, $300M, and fewer than 10 customers

NewCore’s financing creates a sharp contrast: $66 million raised, $300 million post-money valuation, fewer than 10 customers, and more than 10 design partners. The startup expects to begin charging customers this summer, Alon told TechCrunch.

That gap is the story. Investors are not underwriting mature revenue here. They’re underwriting the belief that AI agent identity will become a control layer enterprises can’t avoid once autonomous tools move deeper into production systems.

NewCore’s platform treats agents as first-class identities, with their own permissions, lifecycle controls, and revocation mechanisms. That differs from treating them as traditional service accounts or machine credentials, which often blur ownership and intent.

A useful way to see the shift:

XOOMAR analysis: the hard part won’t be assigning an ID to an agent. It will be proving, in real time, whether that agent should be allowed to do a specific thing in a specific context.

That means enterprises will care less about identity labels and more about operational evidence:

• Scope: Which systems can the agent access?
• Owner: Which human or team is accountable for it?
• Logs: Can the company reconstruct what it did?
• Revocation: Can access be cut without breaking unrelated workflows?
• Approval latency: Does control slow work to a crawl?

Those are buyer questions, not just security architecture questions.

Old identity tools were built for employees, not software workers making choices

NewCore’s critique of incumbent identity systems is that they were designed around human workforces, then extended for machines. Established providers including Okta and Microsoft Entra have begun adding AI-agent capabilities, TechCrunch reported, but Alon argues those additions sit on platforms originally built for employees.

NewCore says its product was built from the ground up for humans, machines, and AI agents in one system. Its architecture includes a “split-key” model that divides critical identity credentials between the customer and NewCore’s platform, a design intended to remove a single point of compromise.

The company also offers an “Agentic Skill” integration package for coding assistants including Anthropic’s Claude Code, OpenAI’s Codex, and Cursor. The point is to let those tools access enterprise systems as managed identities rather than through manually distributed credentials.

That matters because coding agents are a natural early pressure point. They operate near source code, tickets, internal docs, and production workflows. For adjacent security context on agent risk, XOOMAR readers can see Malware Fears Force NanoClaw AI Agents Through JFrog.

NewCore also says employees can use its mobile app to grant, review, and revoke access for AI agents. That gives the platform a human oversight layer, which is essential if enterprises want agents to act autonomously without becoming invisible.

From service accounts to AI coworkers, autonomy creates the new sprawl

Enterprise security has seen versions of this problem before. Passwords led to single sign-on. Cloud adoption created sprawling roles and permissions. Automation created credentials that often outlived the workflows they served.

AI agents add a different kind of sprawl: autonomy. A conventional script usually does what it was written to do. An AI agent may interpret a task, call tools, generate code, query data, and escalate a workflow based on intermediate results.

That flexibility is the value proposition. It is also the control problem.

XOOMAR analysis: NewCore’s best argument is that agents should not inherit identity shortcuts from the automation era. If an agent uses a human’s credentials, the audit trail can mislead investigators. If it uses a broad service account, permissions can exceed the task. If it lacks lifecycle controls, no one may know when it should stop existing.

This is where the broader AI-agent tooling conversation connects. Our coverage of Microsoft SkillOpt Sidesteps Retraining for AI Agents sits in the same practical zone: enterprises want agents that can gain capabilities without turning every deployment into a governance blind spot.

CISOs, CIOs, and vendors will fight over who controls the agent layer

NewCore’s platform lands in a politically sensitive part of the enterprise stack. Security teams will want least-privilege access, detailed logs, and instant revocation. CIOs and business units will want agents that make work faster without routing every action through a ticket queue.

Employees have their own concern: attribution. If an AI agent acts on behalf of a worker, the company needs to know whether the person approved the action, whether the agent acted independently, and whether the credentials used were appropriate.

Vendors will also crowd the control point. TechCrunch names Okta and Microsoft Entra as established identity providers adding agent features. NewCore’s argument is that a platform rebuilt for AI agents has an architectural advantage over extensions to older systems.

That claim is plausible, but still unproven. NewCore has more than 50 employees across the U.S. and Israel, but it has not yet shown public evidence of broad enterprise adoption. Its charging model is also still ahead, not behind it.

The agent identity test: inventory first, control second, trust last

For enterprise buyers, the practical lesson is immediate: inventory AI agents before they become impossible to govern. Assign human owners. Define scopes. Log actions. Set revocation rules before agents spread across production workflows.

NewCore is right about the direction of travel. If AI agents become a real part of the workforce, AI agent identity becomes a core security category sitting between identity governance, cloud security, data security, and AI governance.

The risk is weak implementation. Over-permissioned agents, shared credentials, unclear logs, and automated actions no one can explain afterward would turn agent deployment into an investigation nightmare.

Alon frames the urgency directly:

“It’s inevitable,” Alon said of AI agents becoming a significant part of the workforce. “The question is whether we’re going to build the guardrails in time.”

The next evidence to watch is simple: whether NewCore converts design partners into paying customers this summer, whether large enterprises accept agent-specific identity as a budget line, and whether incumbents can match the same control model without bolting agent governance onto systems built for a different workforce.

The Bottom Line

• NewCore’s $66 million seed round signals investor confidence that AI agent identity could become a major enterprise security category.
• Companies adopting AI agents will need ways to assign ownership, limit permissions, and revoke access when agents are no longer needed.
• The startup’s $300 million valuation before charging customers shows how quickly security risks around agentic AI are becoming a market priority.