Ex-MI6 Officer Wants To Break Big Tech’s Grip On Enterprise AI

British intelligence once sent Tyler Edwards to a training installation with instructions that involved a van ride with a stranger and a box for his phone. He remembers the destination as “the fort.” Years later, when he began exploring AI for intelligence work, establishing control over the technology proved more complicated.

“We simply couldn’t trust AI to make life-or-death decisions. A major part of the problem was that we had to rely on a third party to build the technology, without having much control over how it worked or the outcomes it produced,” Edwards says.

Overmind, the company he co-founded and leads as CEO, initially focused on helping businesses supervise AI agents. But customers, he soon discovered, worried less about the risks of deploying AI than about competitors adopting it faster. That realization pointed him toward a bigger opportunity. “If you can develop intelligence around your own needs, shape how it operates and determine the outcomes you want to achieve, you have far greater control over the technology and more confidence in what it can deliver,” he says.

Founded in 2025, the London startup raised £2 million in a seed round led by Osney Capital in February. In October, it released an open-source platform to help developers turn proprietary business data into specialized AI models without a dedicated machine-learning research team. Edwards previously worked at MI5 and MI6, where he helped develop tools for British offensive cyber operations and later led teams exploring how AI could automate intelligence analysis. Cofounder and COO Sam Brunt brings experience scaling Funding Circle and Vertice.

Turning Agent Behavior Into Specialized AI Models

The company’s original tools tracked how agents executed tasks and flagged behavior that went off course. Those records now pinpoint failures and become training examples for specialized models.