AI Score raises £4M seed to govern enterprise AI agents before regulators do it for them

AI Score (London) has closed a £4 million (~€4.6M) seed round led by Fuel Ventures, with participation from GALLOS Technologies, the founding studio that incubated the company. The round comes with a notable advisory board: Sir Jeremy Fleming, the former director of GCHQ, the UK's signals intelligence and cybersecurity agency. AI Score was founded in 2025 by Alex Harland (CEO, previously on the founding team of the UK National Cyber Security Centre), Benita Tibb (co-founder, former City lawyer), and Jonathan Kewley (co-founder, tech-law specialist). The company employs 26 people.

The product is a real-time governance platform for enterprise AI agents: a monitoring layer that tracks which AI models are running inside an organisation, who is using them, at what cost, and with what behaviour — and produces the audit trails and guardrails that regulated-sector compliance functions require.

The governance gap AI Score targets

The proliferation of AI agents inside enterprises has moved faster than the governance frameworks designed to control them. A large professional services firm may now be running dozens of AI systems across document review, client communication, research, and billing — systems that operate autonomously, at volume, and often without the same oversight trails that human decision-making would require. Traditional governance — access controls, audit logs, periodic reviews — was not designed for processes that execute thousands of decisions per day without pausing for human sign-off.

AI Score's platform operates at the agent layer: monitoring what models are running, flagging behavioural anomalies, enforcing policy guardrails, and generating compliance audit records. Early clients include Magic Circle law firms and a FTSE 250 company — among the most governance-sensitive buyers in the UK market — alongside a leading UK fintech. At 26 employees and 18 months old, the client quality outpaces the company's size in a way that matters: in enterprise compliance tooling, the reference is everything.

Regulatory timing and the GCHQ signal

The EU AI Act's provisions governing general-purpose AI systems entered force in August 2026. For regulated industries — legal, financial services, insurance — the obligation to document, audit, and govern AI decision-making processes is now a compliance requirement, not a strategic preference. AI Score sits at the intersection of that legal pressure and the real-world absence of tooling to meet it.

"AI operates at a scale and level of automation that traditional governance simply wasn't built for," said Alex Harland, CEO and co-founder.

Sir Jeremy Fleming's advisory role is not incidental. Enterprise security and compliance buyers in government-adjacent sectors — defence primes, regulated financial institutions, major law firms — use team provenance as part of their vendor evaluation. A founding team whose previous roles included standing up the UK's national cybersecurity infrastructure is a credential that a VC-backed AI startup from a generic enterprise background cannot replicate.

The £4M seed is sized to establish the compliance-market beachhead, not to win the broader enterprise governance market. What the company needs over the next 18 months is depth in its existing reference accounts — integrations robust enough that replacing AI Score becomes operationally inconvenient — before the governance modules inside larger enterprise platforms catch up to what a dedicated tool can do.