The round also includes new investment from Crosspoint Capital and follow-on funding from existing backers General Catalyst and Ten Eleven Ventures. The new investment brings the total Geordie has raised to date to $36.5 million and comes less than a year after the company first emerged from stealth with a $6.5 million seed round in September.

The startup, which was cofounded by veterans of the U.K. cybersecurity company Darktrace—which had been one of the first cybersecurity firms to apply machine learning to detecting unusual network activity—and Snyk, a company that created software for automatically identifying and fixing vulnerabilities in code bases, has caught the attention of the cybersecurity industry. Geordie won the 2026 RSAC Innovation Sandbox contest in March—a closely watched startup competition for cybersecurity companies, whose past finalists have included Wiz and SentinelOne.

Geordie is now deployed across roughly 30 customer environments, cofounder and CEO Henry Comfort told Fortune. Those customers include upstart financial data company AlphaSense, where Geordie covers “tens of thousands of agents,” and French-American AI biotech company Owkin, an early customer that runs hundreds of agents across more than 50 petabytes of data, according to Geordie.

“We believe agents are going to change the nature of work,” Comfort said. They are, he said, “going to need purpose-built security and governance capabilities.”

Comfort co-founded Geordie in early 2025 with Hanah Darley, the former director of security and AI strategy at Darktrace, where Comfort previously served as COO for the Americas, and Benji Weber, the former senior director of engineering at developer-security firm Snyk. Weber is now Geordie’s chief technology officer, while Darley is its chief AI and product officer.

The company currently has 37 employees split between offices in London and New York, and Comfort said he expects headcount to reach about 50 in the next three months. Money from the Series A funding round will be used primarily to expand engineering and U.S. go-to-market teams.

Geordie’s pitch to enterprise security teams is that they need an independent view of AI agent activity that does not come from the vendors providing the agents and that works across all agent types and every kind of agent deployment.

The company’s platform discovers AI agents wherever they are running—whether that’s locally on an employee’s laptop, on the company’s own cloud, inside cloud-hosted software platforms, or within a company’s code base. It then maps all the tools, interfaces, plug-ins, and data sources each agent can reach, flagging risks in real time. A separate module called Beam, which Geordie describes as an “AI agent remediation suite,” uses “context engineering” to dynamically shape and constrain agent behavior.

Geordie is competing against an increasingly crowded field of AI agent orchestration and governance platforms, many of them from vendors that bundle agent oversight into broader software packages—Microsoft, for instance, offers Agent 365, and ServiceNow has its AI Control Tower, while OpenAI has introduced AI agent governance features into its Frontier enterprise platform. But Comfort argues those offerings are a poor fit for enterprises that, in practice, run agents across multiple model providers.

“We’re playing a different game to those guys,” he said. He noted that many of the companies selling AI models and offering “agent building” software are also making it harder to integrate those agents with data and plug-ins from other vendors in an effort to keep more value within their own ecosystem.  “From a governance perspective, you really feel like you need some kind of independent layer,” Comfort said. He likened Geordie’s role to “being the Switzerland of the future when it comes to agents.”

Comfort, who described Geordie’s longer-term ambition as becoming a kind of “air traffic control” for enterprise AI agents, said the principal threat the company faces is whether “incumbents eventually use their distribution advantage to great effect in this space.”

At Owkin, Comfort said, plugging in Geordie revealed that the company had three times more AI agents running than it had previously realized. During a proof-of-concept trial of Geordie’s software, Owkin was able to mitigate risk exposure that, using Owkin’s own risk-quantification methodology, totaled between $12 million and $13 million, Geordie said.

“We’re seeing the iceberg that rocked the Titanic weeks in advance rather than the moment it appears on screen,” Leo Cunningham, chief information security officer at Owkin, said in a statement.

Comfort says fast deployment is another differentiator for the startup. Geordie aims to allow customers to begin “mitigating risk within 24 hours” of installation, and installation itself only takes a few hours, versus the consultant-heavy roll-outs that often accompany enterprise agent projects.

Correction, May 28: This story has been updated to include a different post-money valuation figure for Geordie AI derived from calculations using filings at the British business registry Companies House.