Evidence before assertion
Every output from an AI system should carry the basis it rests on. If an analyst can't see whether an answer comes from six observations or three hundred, they can't oversee it.
Non-Executive Co-Chair, Stealth Labs
AI that can be audited, overseen and trusted.
I've spent a decade across the civil service, big tech and defence, working where AI decisions have to be defensible. In 2024 I co-founded Stealth Labs. Today I co-chair its board, and I advise on AI assurance, procurement and public sector technology.
Now
Non-Executive Co-Chair, Stealth Labs
Background
A decade across the civil service, big tech, defence, national security and government
Thinking about
AI assurance, government procurement, predictive intelligence, AI in defence and quantum
Open to
Non-executive, advisory and forum roles in AI, cyber and public sector technology
About
I'm drawn to the gap between what technology makes possible and what organisations actually do with it, and to the question of how you know an AI system is fit to be relied on.
I started in the civil service, running teams and budgets, and learned first-hand how hard it is for large institutions to adopt new technology. From there I moved into big tech, leading AI for regulated industries and working alongside defence, national security and government teams as they took their first real steps with AI.
In 2024 I co-founded Stealth Labs in Kent, an AI studio that builds production systems for UK government and enterprise. I've since stepped back from the day to day. As Non-Executive Co-Chair, I focus on long-term direction, risk and governance.
I also write about the distance between government AI ambition and procurement reality, and I've submitted written evidence to the House of Lords Science and Technology Committee.
I'm a Member of the Institution of Engineering and Technology (MIET).
I'm open to non-executive, advisory and forum roles in AI, cyber and public sector technology.
Principles
How I think about AI, oversight and evidence.
Let pilots fail
If every pilot succeeds, you aren't being ambitious enough. Fail cheaply, early and visibly, and report it honestly.
Procurement is part of the system
Good AI companies get locked out by how government buys. Fixing the buying process matters as much as fixing the technology.
Build the thing you oversee
Boards make better decisions when someone in the room has built, shipped and operated what is being discussed.
Stealth Labs
AI engineering for UK government and enterprise.
Born in Kent. Grown nationally. Kent is where it began, and where its roots remain.
Stealth Labs is an AI studio that builds production systems for UK government and enterprise, including multi-agent systems, agentic workflows and intelligent automation. It started in Kent in 2024 and now works with organisations right across the UK.
Visit stealthlabs.uk- Founded
- 2024, in Kent
- Reach
- National, across the UK
- My role
- Non-Executive Co-Chair since September 2026
- Previously
- Co-Founder and CEO, 2024 to September 2026
- Focus
- Long-term direction, risk and governance
Yochi
Predictive location intelligence with the evidence attached.
I designed Yochi to answer a hard question for defence users: where will an entity be, and how sure are we? It learns an entity's habitual pattern from its own history and returns a calibrated probability, with the observation counts behind every figure, so the answer can be audited under oversight. It ships as a single container and runs fully air-gapped.
Visit yochi.ioWriting
On government AI procurement, assurance and emerging technology.
The Government AI Gap
Writing on government AI procurement and why so much of it fails.
Written evidence: Innovation in the NHS, personalised medicine and AI
Why the lessons from cross-government AI procurement should carry over to NHS innovation, and how framework design keeps innovators locked out.
Most Government AI Pilots Should Fail
Government AI pilots are designed to succeed on paper while delivering nothing in practice. If your pilot portfolio has a 100% success rate, you're not innovating; you're paying for theatre.
I Used to Be the Buyer
Most AI vendors pitch fear, jargon and vague promises to government buyers. Having sat on the other side of the table, here's what actually gets a meeting, and what doesn't.
How Government Could Actually Buy Good AI
The UK procurement system locks out the best AI companies and rewards the slowest movers. How other countries do it better, and what reform could look like.
AI and National Security: Unlocking Innovation Through Smarter Support
How AI is transforming service delivery, predictive analytics and cyber defence, and why the barriers holding back startups and SMEs have to come down.
The Arrival of Quantum Technology
The UK's position in the global quantum race, and the case for aggressive funding, talent strategy and a serious commitment to leading it.
Get in touch.
For non-executive, advisory or speaking enquiries.