Playbook · 6 minute read
How to Present AI to the Board Without Losing the Room
A board presentation on AI should state the decision being asked for, describe current exposure honestly, show what is already happening in the organisation, and commit to what will be reported next time. Presentations that explain the technology lose the room and answer nothing.
Board presentations on AI fail most often by explaining the technology. Boards need exposure, accountability, and a decision. This playbook covers structuring a presentation that provides them, drawing on FISTA Solutions' AI enablement work.
When is this worth doing?
When a decision is genuinely needed from the board, when regulatory or governance obligations require oversight to be evidenced, or when management should be telling them something before they find it out elsewhere.
Presenting for information without a decision is defensible once. Doing it repeatedly trains the board to treat AI as a standing item nobody acts on.
What does the sequence look like?
| Step | Purpose |
|---|---|
| 1. Name the decision | Or say plainly there is none |
| 2. Show what exists | Inventory, including unsanctioned use |
| 3. Present exposure | Regulatory, operational, reputational |
| 4. Name accountability | A person per area |
| 5. Ask for the decision | With options and a recommendation |
| 6. Commit to reporting | Few metrics, reported consistently |
Step 1 — Name the decision in the first minute
Open with what you are asking the board to decide: approve an investment, accept a risk position, endorse a policy, or note a position with a specific concern.
Boards allocate attention by importance, and a presentation without a decision competes poorly. If there genuinely is no decision, say so at the start and keep it short.
This also disciplines the preparation. A presentation built backwards from the decision is shorter and clearer than one built forwards from what the team has been doing.
Step 2 — Show what already exists
Present the inventory: systems in production, what decisions they influence, who owns each, and what regulatory obligations attach.
Include unsanctioned use. Staff are using AI tools regardless, and a board that discovers this through an incident has been failed by management. Presenting it with a plan is uncomfortable once; the alternative is worse and lasts longer.
Keep it to a single page. Boards do not need the full register, they need the shape and the outliers. See what is an ai inventory.
Step 3 — Present exposure honestly
Cover regulatory obligations that apply, operational risks, and reputational exposure, with what controls exist for each and what gaps remain.
The temptation is to present a controlled picture. Resist it: boards include people who read the same regulatory commentary you do, and a presentation that omits an obvious obligation invites the question you least want.
Named gaps with owners and dates are a sign of competent management. Absent gaps are a sign of an incomplete review.
Step 4 — Name who is accountable
For each area — systems, data, regulatory position, incidents — name a person.
Boards are increasingly expected to oversee this, and in some regimes accountability extends to them personally. A presentation that cannot name who owns what invites the reasonable inference that nobody does.
Committees are not accountability. Name individuals.
Step 5 — Ask for the decision with options
Present two or three options with the trade-offs and a recommendation, rather than a single path.
Boards make better decisions with alternatives, and presenting only one invites the question of what else was considered. It also makes the recommendation stronger, because it shows the analysis behind it.
Be explicit about what happens if they decline. A decision with no stated consequence for inaction is easy to defer.
Step 6 — Commit to what you will report
Pick three or four metrics you can report consistently: systems in production, incidents and their severity, adoption, and cost.
Consistency matters more than comprehensiveness. A board that sees the same four numbers each time can see a trend and ask better questions; one that sees a different comprehensive picture each time sees only the current presentation.
Report them next time including the ones that moved the wrong way. Boards trust management that reports bad numbers without being asked.
How long should it be?
Shorter than you think. Ten to fifteen minutes of presentation with the rest for discussion, supported by a short written paper circulated in advance.
The paper does the explaining; the meeting does the deciding. Presentations that read the paper aloud consume the discussion time that makes board oversight worth having.
What questions should you expect?
What could go wrong and what stops it. Whether we are exposed regulatorily. What competitors are doing. Whether staff are using it unsanctioned. What it costs and what it has delivered.
Prepare specific answers with evidence. A board member who receives a vague answer to a direct question will ask a sharper one, and that dynamic is hard to recover from in the room.
Who needs to be involved?
The executive accountable for the area, supported by whoever holds the detail. Presenting through a technology specialist with no accountability weakens the message.
The risk or compliance lead should be present where exposure is on the agenda, because the board will ask them regardless.
How long does it take?
Two to three weeks of preparation, most of it assembling the inventory and exposure picture honestly. The presentation itself should take an hour to build once the substance exists.
What are the common failure modes?
Explaining the technology. No decision. Sanitised exposure. Committees instead of named owners. A different metric set each time. And a live demonstration.
How do you know it worked?
A decision made, the board able to describe the organisation's exposure, the same metrics reported again next time, and no surprise arriving from outside the reporting.
What does it cost?
Mostly people's time rather than tooling. The expensive version is the one that stalls halfway and leaves the organisation with neither the old state nor the new one, which is why a narrow first pass beats a comprehensive plan nobody finishes.
Budget the work as an operated change rather than a project with an end date, because most of these need a maintenance tail. See AI total cost of ownership.
What should you do first?
Build the one-page inventory, including what staff are using unsanctioned. That page is the presentation's foundation and the hardest part to produce honestly.
How FISTA Solutions helps
FISTA Solutions runs this work alongside client teams rather than around them: inventories and exposure assembled honestly including unsanctioned use, a small consistent metric set established for repeat reporting, evidence produced as the work proceeds, and handover that leaves your people able to continue without us. Delivery runs through AI agents, AI enablement, and forward deployed engineers. The record is 150+ projects for 50+ companies across 12+ countries, with 47% average efficiency gains where measured.
To run this with support, message FISTA on WhatsApp, or read how to run an AI governance review.
Share-ready article cover
Download the generated social format.
Clear answers
Questions raised by this field note.
Straightforward guidance for evaluating scope, fit, and the next step.
01What do boards actually need?
To understand the organisation's exposure, to know who is accountable, and to make whatever decision is being asked of them. They do not need an explanation of how models work, and providing one consumes the time available for the rest.
02How should exposure be presented?
Honestly and specifically: what systems exist, what decisions they influence, what regulatory obligations apply, what could go wrong, and what controls exist. Sanitised presentations get found out, usually by a non-executive who read something.
03Should unsanctioned use be disclosed?
Yes. Staff are using AI tools regardless, and a board that learns this from an incident rather than from management has been poorly served. Presenting it alongside the plan to address it is far better than omitting it.
04What metrics should be committed to?
Three or four that can be reported consistently: systems in production, incidents, adoption, and cost. Reporting the same small set each time lets the board see a trend, which is more useful than a comprehensive one-off picture.
05Should you demonstrate the technology?
Rarely. Demonstrations consume time, invite questions about the tool rather than the governance, and risk failing live. A one-line description of what a system does is usually more useful than showing it.
Continue exploring
Related capabilities
Start with the hard problem
Need the outcome owned, not merely analyzed?
Tell us where delivery is constrained. We’ll map the fastest credible path from intent to verified production.