Startup Geordie AI Lands $30M to Safe Enterprise AI Brokers

The winner of this 12 months’s RSAC Innovation Sandbox contest raised $30 million to offer visibility into synthetic intelligence brokers throughout cloud, code and endpoint environments.

See Additionally: AI Brokers Introduce a New Insider Risk Mannequin

The Balderton Capital-led Sequence A spherical will present London-based Geordie AI with the assets wanted to broaden engineering capabilities, speed up product improvement and preserve the tempo of innovation clients require, stated co-founder and chief AI officer Hanah-Marie Darley. She stated Geordie AI is growing a safety framework designed particularly round AI agent habits and governance.

“Brokers are inherently working throughout all of those floor areas, usually concurrently,” Darley instructed ISMG. “Once you’re speaking about safety, you possibly can’t simply have one thing on the endpoint, you possibly can’t simply have one thing in cloud options. You want one thing that has protection throughout cloud, code and endpoint providers, as a result of that is the place brokers are working each day.”

Geordie AI, based in 2025, employs 37 individuals and has raised $36.5 million, having beforehand acquired $6.5 million in a September 2025 seed funding spherical co-led by Ten Eleven Ventures and Basic Catalyst. The corporate has been led since its inception by Henry Consolation, who beforehand spent five-and-a-half years at Darktrace, culminating in a job overseeing international enterprise operations (see: OpenAI’s Dawn Bets on Agentic Cyber Protection).

Why Organizations Battle With AI Agent Visibility

Brokers have gotten a part of core enterprise infrastructure, Darley stated, automating workflows, making selections and executing duties that had been beforehand dealt with by workers. Organizations, subsequently, cannot depend on present controls designed for customers, purposes and infrastructure working inside established perimeters. AI brokers behave otherwise, function autonomously and repeatedly evolve.

“This isn’t simply one other small management airplane,” Darley stated. “It isn’t only a small layer or a safety instrument. It is actually altering enterprise infrastructure. It is actually altering materially how companies function. And so the kind of safety paradigm that you just want shouldn’t be going to be a easy instrument layer or one thing that’s simply going to sit down on some extent resolution that is going to be off to the aspect.”

Organizations usually underestimate how completely different AI brokers could be from each other, with a Claude-based agent, a Copilot Studio agent and a custom-built enterprise agent all working otherwise, utilizing completely different instruments and creating completely different safety dangers. And new agent frameworks seem frequently, forcing safety groups and distributors to repeatedly adapt their visibility, monitoring and governance methods.

“That floor space goes to proceed to broaden, and we should be proper on the entrance of that,” Darley stated. “It entails constant protection and safety as we evolve not solely our fashions and the way in which that we proceed to prioritize proprietary threat and enforcement, but additionally the way in which that we proceed to remain forward because the area adjustments.”

Many organizations at present lack even a primary understanding of what their AI brokers are doing on a day-to-day foundation, with present instruments not capturing instrument utilization, agent-to-agent interactions and autonomous workflows, Darley stated. Geordie AI’s strategy entails profiling brokers, establishing behavioral baselines and performing anomaly detection, which helps organizations extra simply determine uncommon habits.

“One of many issues that we repeatedly do is actually profile brokers,” Darley stated. “We perceive how they function. We provide you with behavioral baselines, and we additionally go into anomaly detection. And that is solely going to evolve as we proceed to deepen not solely our understanding and our clients’ understanding of brokers, but additionally because the area continues to evolve and alter.”

How Routing Exercise By way of Gateways Introduces Complexity

Safety architectures that rely closely on gateways and proxies can present worth throughout early-stage deployments, however they turn into more and more tough to handle as organizations broaden their use of AI brokers, Darley stated. As brokers entry extra instruments, APIs and SaaS providers, routing all exercise by means of centralized gateways can introduce operational complexity and efficiency bottlenecks, she stated.

“As an alternative of getting to architect your entire workflows by means of a gateway, you are capable of function with coverage enforcement throughout all of those brokers,” Darley stated.

Simply as human workers want insurance policies, procedures and coaching to function safely, AI brokers require contextual data that guides their selections in machine-readable codecs and sometimes want it delivered in actual time, Darley stated. Safety and governance controls should subsequently be built-in instantly into agent workflows quite than delivered by means of periodic coaching or static documentation, Darley stated.

“For us, that additionally facilities on context engineering,” Darley stated. “Giving the brokers again the identical context that you’d give your safety engineers, however in an agent-readable means.”

The business is starting to maneuver past single-agent deployments towards extra dynamic fashions during which brokers autonomously hand off duties to 1 one other and coordinate actions throughout a number of techniques, Darley stated. Organizations should subsequently perceive not solely particular person agent habits but additionally how brokers work together with one another, trade data and collectively execute workflows.

“It’s important to have purpose-built safety in terms of brokers,” Darley stated. “You will not be capable of simply pivot one thing out of your present resolution, since you’ll have gaps – whether or not that is context gaps, whether or not that is safety management gaps, or whether or not that could be a mixture of visibility and management gaps. The important thing problem round brokers is that they are not the identical as conventional software program.”