Most executive AI briefings are a tour of the technology. Ninety minutes later the room is impressed and no better equipped to make a decision. The briefing that works does something narrower.
The room does not need to understand transformers
Leadership teams are asked to approve budget, accept risk and sponsor change. None of those decisions require knowing how attention works. They require a reliable sense of three things: what this class of system is genuinely good at, where it fails, and what it costs to find out.
Time spent on architecture is time not spent on those.
One framework, four questions
We give the room a single instrument they can apply to any proposal that reaches them afterwards.
- What decision or task does this change? If the answer is a capability rather than a task, it is not yet a proposal.
- What does being wrong cost? Drafting a reply an officer reviews is cheap to get wrong. Determining eligibility is not. This sets how much human review the design must carry.
- What is the baseline? The current number, measured, before anything is built.
- Who owns the result? A named person in the operational line whose own performance moves with it.
Four questions, applied consistently, filter out most of what wastes an organisation's first two years with AI.
Then they use it
The second half is hands-on, on the organisation's own material — a real circular, a real inbox, a real report. Two things reliably happen. Someone is surprised by how well it handles a task they assumed needed a specialist. Someone else is surprised by how confidently it gets something wrong.
Both matter. A leadership team that has personally watched the system fail makes far better decisions about where to deploy it than one that has only seen the demo.
Calibration, not enthusiasm, is the point of the ninety minutes.
What we leave behind
- The four-question framework, on one page, in Arabic and English.
- A shortlist of candidate use cases the room generated themselves, roughly scored on value and risk.
- An honest note on which of them we think are not worth doing.
That last item is usually what people remember.