Who is in charge of this model?
When everyone owns a model, nobody does. Every model your organisation depends on needs one accountable name.
James A Lang
AI non-executive directorThe AI leader
at the planning table.
Your board is asking questions about AI it cannot answer. I sit beside you while you answer them.
Book a briefing
Boards in regulated industries, defence and government.
The questions
Three questions your board cannot yet answer.
Most AI programmes don't fail on technology. They fail on leadership.
What happens when it is wrong?
Governance frameworks go silent on loss. Resilience is what buys you the right to be wrong, and the board has to plan for it.
Where does judgement sit?
Leaders are responsible for AI decisions they did not build. Decide now which decisions stay human, and protect the space to make them.
The ten
Ten questions boards are asking me.
One sits above the rest: how do we balance innovation and responsibility? The other nine come in threes.
Direction
- 01What is our AI strategy for the next decade, not the next budget?
- 02How do we create an AI strategy our stakeholders will trust?
- 03How do we make money from AI without spending the trust that makes it worth having?
Risk
- 04Which AI risks should keep this board awake, and which should not?
- 05How do we prepare for regulation we cannot yet read?
- 06Who is in charge of each model, and what happens when it is wrong?
People
- 07How do we deploy AI responsibly, in practice rather than in policy?
- 08How do we manage agents that act on our behalf?
- 09How do we train our people to use AI wisely, and what happens to the ones who don't?
Download the one-pager
A page a chair can forward. Printable.
The person
What I bring to the table.
I spent 22 years in the British Army. I have built AI for the Ministry of Defence and audited it for boards. I know what command, delivery and scrutiny each demand.
Operational command
22 years of British Army service. Stood up the Army's Cyber Protection Team and took the first AI cyber security systems to the edge, from Estonia to an aircraft carrier.
AI in regulated sectors
The Ministry of Defence's first Cybersecurity AI Product Manager, leading its big data AI platform across 65 engineers. Adviser to boards in regulated industries, defence and government.
Audit and assurance
Oxford Artificial Intelligence Audit Programme, Saïd Business School. CISSP. Chartered Manager. Founder of Velinor, an AI governance, strategy and audit consultancy.
The record
Built with the Ministry of Defence, the Home Office and UK Fusion Energy.
The frameworks I bring to a board were built inside organisations where AI failure is not an option. Three engagements, anonymised as required.
Governance at the no-fail threshold
AI deployed where failure is unacceptable, because governance was treated as resilience infrastructure rather than paperwork. The board got one accountable name per model and a tested route to stop it.
Making mission data trustworthy for AI
AI scale fails when the data underneath has no owner, no consistent controls and no auditable lineage. We defined what trusted data for AI means in practice, so leaders could use it and account for it.
Board-grade AI incident readiness
Most organisations deploy AI with no plan for when it goes wrong. This one left with an incident capability leadership could trust: who acts, what they do, how fast, written for execution rather than the shelf.
The fit
Boards where AI can change the company.
Where it will decide what the business makes, how it competes, or what it can be held accountable for, and where the board has to answer for systems it did not build.
I know three sectors from the inside: critical national infrastructure, defence and security, and government and regulated services.
Non-executive director
A seat, for boards that will have AI on the risk register for the next decade and want one director who has built it, governed it and audited it.
Discuss a seatAdvisory board
The same judgement without the fiduciary role: a standing adviser to the chair, the audit and risk committee, or the executive, on the questions AI puts to them.
Talk it throughThe briefing
Thirty minutes. Three questions. No obligation. You leave knowing what your board needs to decide this year, and whether I'm the right person to help.
Book a briefingCISSP. Chartered Manager. Oxford Artificial Intelligence Audit Programme, Saïd Business School. MSc Leadership and Management. Founder, Velinor.
Availability
Up to two board seats, 15 to 20 days a year each, including committee work.
Independence
Velinor does not work for any organisation on whose board I sit. I declare my minor shareholding in Kaze on appointment.
