Alastair Paterson

Alastair Paterson co-founded Harmonic Security in August 2023 with Bryan Woolgar-O’Neil, now its chief technology officer, and is its chief executive. It is his second security company. He co-founded the threat intelligence firm Digital Shadows in 2011 and led it as chief executive until ReliaQuest agreed to buy it for $160m (£120m) in June 2022.

Governing how staff use AI

Harmonic sells an AI governance and control platform. It works on devices and in browsers to show which AI tools employees use, classify what they use them for and stop sensitive data leaving through prompts and uploads. In October 2025 it launched an MCP Gateway that extends those controls to AI agents connecting to company systems through the Model Context Protocol (MCP), and in January 2026 it relaunched its brand.

The company raised a $7m seed round in October 2023 and a $17.5m Series A led by Next47 in October 2024, taking total funding to more than $26m. Harmonic lists Advisor360, Apax and Cisco among its customers, and a partner from In-Q-Tel sits on its board.

What its data shows

Harmonic’s own research is central to its public case. In January 2026 it published an analysis of 22.4 million enterprise prompts sent to AI tools in 2025. Of those, 579,113 contained sensitive company data, and 16.9% of the exposures came through personal free-tier accounts that IT teams cannot see. A second study, published in May 2026, classified 1.9 million minutes of workplace AI use over seven weeks and found that 64.5% of activity on personal accounts was business use. Both studies are Harmonic’s own, based on data from a panel of enterprise organizations.

In May 2026 Harmonic was named to Notable Capital’s Rising in Cyber 2026 list of 30 startups, chosen by votes from about 150 CISOs and senior security executives.

Why it matters to UK technology buyers

For UK IT and security leaders, Harmonic’s research describes a familiar problem: much of the AI use that carries risk happens outside approved tools. Harmonic argues that blanket blocking only drives staff to personal accounts. Its answer is to control usage at the point where data enters a prompt, and to apply the same rules to agents as they are connected to internal systems. That makes it relevant to any organization writing an AI acceptable-use policy it cannot yet enforce.

Paterson built and sold one UK threat intelligence company. With Harmonic he has moved early into AI governance, a category security teams are still defining, and the Rising in Cyber listing suggests CISOs are paying attention.