A board AI report template in eight slides: position, opportunity, risk, investment, progress, KPIs, decision and question. By AI strategy advisor Yüce Zerey.
On a Tuesday last quarter I sat in a boardroom as an independent advisor while an 87-slide AI presentation was loaded. By slide 14, nobody was reading. By slide 32, the phones were out. By slide 51, the CFO was taking notes on something else entirely. A board AI report of that length produces a briefing; it never produces a decision.
This guide sets out the alternative I use in board-level advisory work: a board AI report of exactly eight slides, in a fixed order, built around the decisions a board exists to take. The template, the logic behind the sequence and the discipline rules that keep it intact follow below, drawn from Yüce Zerey's advisory casework.
During that 87-slide meeting I wrote one line in my notebook: good strategy fits on an A4 page; poor strategy spreads across 87 slides.
Boards currently sit at two extremes with AI. Some watch marathon decks and learn nothing usable. Others see no reporting at all and stay blind on the most consequential capital question on their agenda. Both extremes fail the company. The eight-slide board AI report draws the middle line: eight slides exactly.
Why eight? A board member gives the AI agenda 30 to 45 minutes. Attention follows a fatigue curve across any meeting: the first 20 minutes are sharp, the rest fragments. Eight slides at roughly four minutes each makes 32 minutes, which leaves the remainder for questions. Eight is also the minimum structure needed to tell the company's AI story in full.
The governance gap is real and documented: McKinsey's State of AI research shows that CEO-level ownership of AI governance is one of the elements most strongly correlated with bottom-line impact, yet only a minority of organisations report AI metrics to leadership on a regular rhythm. Until 30 to 45 protected minutes exist on the agenda, that gap stays open.
A case from practice: a FTSE 250-listed manufacturer, quarterly board meeting. The first quarterly AI report ran to 23 slides; in a one-hour slot the board discussed two of them. The following quarter the deck was cut to eight and the sequence discipline applied. In the same hour the board discussed six slides, took three decisions and parked two precise questions for the next meeting. The CTO's verdict afterwards: "The board finally understood me."
The opening slide of the board AI report answers a single question: where are we on AI this quarter? It carries four charts and one line of text.
The four charts: number of active AI use cases, weekly AI usage hours per person, pilot success rate, and AI's share of the total IT budget. Each is plotted across the last four quarters, so the trend is visible without commentary.
The one line of text follows a fixed grammar: "Focus this quarter is X. Change against last quarter is Y. Target next quarter is Z."
This slide answers the board's "where do we stand" question in 60 seconds. Sixty seconds is all it needs.
The opportunity slide of the board AI report shows the value pool the organisation can realistically reach with AI, built on scenarios.
Three scenarios: conservative (50 per cent likely), expected (30 per cent), ambitious (20 per cent). Each carries one combined figure: annual savings plus annual revenue.
For a mid-to-large UK group, the typical range I see in advisory work runs from £4-5 million a year in the conservative case, £8-10 million expected, and an ambitious band around £15 million. Every figure is then sense-checked against published deployment evidence such as Stanford's study of 51 successful enterprise deployments.
A single number misleads a board. Three numbers open a decision range, and "which scenario are we aiming for" becomes a strategic call the board makes for itself.
A downloadable PPTX version of the eight-slide template is available on request: contact the Speaker Agency team.
Opportunity and risk travel together. Showing one without the other distorts the board's picture, so the risk slide follows the opportunity slide every quarter.
Five risks, listed: data breach (with UK GDPR and Data Protection Act 2018 exposure), copyright infringement, employee resistance, loss of customer trust, and competitor moves. Next to each risk sit three scores: likelihood (1-5), impact (1-5) and current control level (1-5). Where likelihood multiplied by impact runs high and control runs low, the line is flagged red.
Klarna's 2024-2025 experience is the best teaching case for this page. The company deployed an AI service assistant announced as covering the work of roughly 700 agents, and watched satisfaction fall on complex interactions. By 2025, CEO Sebastian Siemiatkowski conceded the firm had focused too hard on efficiency and cost while quality dropped, and began bringing humans back. Was the risk foreseeable? Yes: "what happens to customer NPS on complex interactions after deployment" belonged on this slide.
The widely reported Hong Kong deepfake case is the second lesson: an employee transferred roughly $25 million during a fake, AI-generated video meeting. Any risk register without an identity verification protocol leaves the organisation open to the same attack.
This slide answers "what are we facing". Its job is to show reality as measured and let the board argue about it. Omit it, and the board AI report loses its credibility.
The money page, and reliably the CFO's favourite slide in the board AI report.
Two columns: investment to date this year, and proposed investment for next year. Both are split across four categories: infrastructure, licences and tools, training, and consultancy plus people.
PwC's Global CEO Survey keeps recording the same frustration at the top: a majority of CEOs have yet to see meaningful return from AI investment. In my casework the cause is usually allocation. The recurring pattern: roughly 75 per cent of budget goes to licences while training and process design are neglected.
Stanford Digital Economy Lab's Enterprise AI Playbook (Pereira, Graylin and Brynjolfsson) carries a blunt warning from its 51 successful deployments: 77 per cent of the hardest obstacles sat outside the technology, in change management, data quality and process redesign; the invisible costs. An organisation that puts three quarters of its money into licences and almost nothing into those three lines sees no return.
This slide answers "where does the money go, and where will it come back". When the category split is wrong, every other number in the deck becomes suspect.
Want this template adapted for your organisation? Yüce Zerey delivers it as a keynote, workshop or masterclass: request a corporate AI briefing.
What happened since last quarter: one summary, three metrics.
The three metrics: use cases completed, measured savings in pounds, and employee adoption percentage.
This page of the board AI report answers "did we use the time well". When there is no movement, the slide says so. "Progress slowed this quarter because of X" is acceptable to any board; a padded figure is the one thing this page must never carry, because a single discovered embellishment poisons every future report.
One page per quarter. Plain. Where there is movement, show the evidence; where there is none, state it and explain it.
A seven-metric dashboard. The seven metrics from AI ROI: The 7 Metrics That Actually Measure It appear on this single page of the board AI report: cycle time, error rate, capacity released, revenue uplift, churn effect, employee adoption and optionality. Each metric shows three values: pre-AI baseline, current, target.
Seven rows, four columns, one page. Colour code: green (on target), amber (halfway), red (behind). A non-executive director scans this page in 30 seconds and sees instantly which metric is in trouble and which is moving.
Sector benchmarks sit in small print beneath the table; the rework share of AI-saved time is a useful one, because a healthy speed metric can hide it. The footnote lets the board ask the right comparative question: is the company ahead of the published base or behind it?
The most important page in the board AI report. It answers the question every director is silently asking: what am I being asked to decide?
Three options, each with three attributes: what it means, what it makes possible, what risk it carries.
|
Option |
Move |
Outcome |
|
A |
Hold current investment; open no new use cases |
Slow scaling, low risk |
|
B |
Increase investment by 30 per cent; open two new use cases |
Medium scaling, medium risk |
|
C |
Double investment; open five new use cases |
Fast scaling, high risk |
The board decides between the three. The decision matrix is open, and each option's consequence is explicit.
This slide is the true centre of the meeting. The previous six pages are preparation for it; the decision is taken here.
The closing slide. The CEO or CDO puts a direct question to the board, in the format "here is what we need from you".
Example: "We recommend Option B. Do we have your approval? Do we lock the investment mandate for this quarter? If the answer is no, under what condition would it become yes?"
The question converts the board from passive audience into active decision maker. Without it, the board AI report stays a monologue, the meeting stays a slide show, and the executive team leaves with nothing it can act on.
Three questions at most, each a single line, each answerable with yes, no or a condition. Anything that needs deeper discussion is parked for a second session or referred to the relevant committee, with detailed risk questions typically routed to the audit committee.
Five Rules That Keep the Board AI Report at Eight Slides
These five rules create discipline. Discipline protects the slide count, the slide count protects focus, and focus is what turns a board AI report into decisions.
Conclusion: Decision First, Deck Second
A board AI report earns its place on the agenda by producing decisions. Until the right sequence, the right owner and the right measurement frame are in place, no board AI report repays the time and money behind it. The eight-slide template exists to force those three things into a page-per-question discipline: position before opportunity, risk beside opportunity, investment before progress, and a decision before the meeting ends.
The template is deliberately rigid because board attention is the scarcest resource it manages. Fifty words per slide, sourced numbers, constant colours and a closing question will feel restrictive in the first quarter and natural by the third. This guide distils the discipline that has emerged from Yüce Zerey's advisory casework into a structure any executive team can apply at its next quarterly meeting.
Planning an AI briefing for your board? Contact the Speaker Agency team to arrange it.
Yüce Zerey speaks on AI strategy and corporate transformation for boards and executive teams. His most requested topics include the 100-day AI roadmap, corporate AI literacy, autonomous AI strategy and board-level briefings, delivered as keynotes, workshops, masterclasses, webinars and advisory engagements for CEOs, CTOs, CDOs and boards.
Yüce Zerey is an AI strategy and transformation advisor with 25+ years of corporate leadership experience across Turkish and European enterprises. As Speaker Agency's AI keynote speaker, he leads literacy programmes, board-level briefings and 100-day transformation roadmaps for UK and EU organisations. His content is built on concrete decision matrices and measurable ROI frameworks.
Which slide is hardest the first time you use the template?
Usually slide 7, the next decision. Most companies are used to presenting the board a single option in a 'here is our plan, please approve' format, and offering three costed options takes practice for the CEO and the executive team. The first month is hard; by the third quarter it is routine.
How do you get from an 87-slide deck to eight?
Cut without sentiment. Ask three questions of every slide: does it touch a board decision, can it merge into another slide, can it move to the appendix? If none of the three earns a yes, delete it; a first pass typically takes 87 down to 24, a second to 12, a third to 8.
Does the investment split on slide 4 vary from company to company?
Yes. Stanford's Enterprise AI Playbook warns that 77 per cent of the hardest obstacles in successful deployments sat outside the technology: change management, data quality and process redesign. A licence-and-infrastructure-heavy budget therefore underperforms, and a data-mature organisation shifts weight from infrastructure towards training and process design; what matters is that the split is deliberate.
Could the KPI dashboard run five metrics instead of seven?
It could, though I advise against it. Five metrics usually drop optionality and capacity released, and losing both misleads the board: the first hides strategic value, the second hides where the freed hours went. The seven-metric page still fits one slide and 30 seconds of scanning.
How often should the board AI report be refreshed?
Quarterly. More often fatigues the board; less often falls behind a fast-moving field and weakens oversight. The quarterly rhythm also matches how most boards already cycle risk and investment reviews.
After Dinner Speakers
Contents
