Boards do not need a technology lecture on AI. They need clear answers to three questions: what will this deliver for the business, what could go wrong and how is it controlled, and how will we know whether it is working? Structure your board conversation around those three questions and you will be far more persuasive than with a slide full of AI terminology.
Before the meeting: understand the board's lens
Board members are thinking about strategy, risk, return and reputation. Some will be enthusiastic about AI, some sceptical, and many unsure. Speak to the chair or a key member beforehand to understand their concerns. This lets you address them directly.
Part 1: What will it deliver?
- Link AI to existing strategic priorities, such as growth, margin, customer experience or resilience.
- Be specific: "reduce proposal turnaround time" rather than "leverage AI for efficiency".
- Show early evidence from pilots in your own organisation, if you have them.
- Separate quick wins from longer-term transformation so expectations are realistic.
Part 2: What could go wrong, and how is it controlled?
Boards are responsible for risk oversight, so address this proactively:
- Data protection and UK GDPR, with approved tools and clear rules.
- Accuracy: human review of AI output where it matters.
- Reputation: transparency with customers where appropriate.
- People: how you will handle changes to roles.
- Wasted investment: staged funding linked to results.
A one-page risk summary with owners and controls reassures boards far more than verbal assurances.
Part 3: How will we know it is working?
Propose a short set of measures the board will see regularly:
- Adoption in priority teams.
- Time saved on key workflows against baselines.
- Quality indicators.
- One or two business outcome measures.
- Risk incidents.
Commit to a review rhythm, for example quarterly.
The ask
Make your request specific and staged:
- Phase 1: a defined, modest investment for pilots and training, with success criteria.
- Phase 2: a larger investment released if Phase 1 criteria are met.
Staged requests are easier for boards to approve because the downside is limited.
Questions boards often ask
- What are our competitors doing?
- What happens if we do nothing?
- Who is accountable?
- How are we protecting customer data?
- What does this mean for headcount?
- How confident are you in the numbers?
Prepare honest, concise answers. "We do not know yet, and here is how we will find out" is a perfectly good answer when it is true.
What to leave out
- Technical detail about how models work.
- Vendor marketing claims.
- Exaggerated projections you cannot support.
- Long lists of possible use cases without priorities.
For a deeper guide to presenting AI strategy, see how to explain AI strategy to your board, and for winning over individual executives, how to get executive buy-in for AI.
Our Leading AI Innovation & Organisational Transformation course ends with you building and presenting an AI plan for your own organisation, which is excellent preparation for a board conversation.
Frequently asked questions
How long should an AI board presentation be?
Short. A few slides or a two-page paper, with time for questions, is usually more effective than a long presentation.
Should we bring an external expert to the board?
It can help with education, but boards usually find evidence from your own organisation more persuasive than external opinion.
How often should the board hear about AI?
Quarterly updates against agreed measures work well for most organisations, with more frequent reporting if major risks arise.