Most AI proposals reaching a board today are framed around opportunity. Fewer are framed around readiness. Both matter. A board that only hears the opportunity case is not being given enough information to approve the investment responsibly.
Here are five questions worth asking before signing off on any significant AI rollout.
1. Who owns this once it is live
AI tools tend to arrive through whichever function moved fastest. That is rarely the same function that should own the risk, the budget and the ongoing performance of the tool once it is embedded in a business process. Before approval, the board should be able to name the accountable owner and confirm that the owner has the authority and the budget to act on that responsibility.
2. What data is this system touching
Every AI use case has a data profile behind it: what goes in, where it is stored, who can see it and whether it includes anything sensitive, confidential or personal. A proposal that cannot answer this clearly is not ready for approval, regardless of how strong the productivity case looks.
3. What happens when it gets something wrong
Every deployed AI system will produce a wrong answer at some point. The question is not whether that happens. It is whether a human is positioned to catch it before it causes harm. It is also whether the process around the tool assumes perfection or plans for error.
4. Do we understand how this will be priced as it scales
Many AI tools are priced on consumption rather than seats. A pilot that costs very little can become expensive once it is rolled out across a whole team or a whole customer base. Before approval, the board should see a cost model that reflects the scaled version of the proposal, not just the pilot.
5. Is this solving a business problem or following a trend
The strongest AI proposals start with a specific, named problem: a process that is too slow, a decision that takes too long, information that is too hard to find. The weakest start with the technology and go looking for a use case afterward. A board should be able to state the business problem in one sentence before it approves the solution.
Why this checklist matters
None of these questions are designed to slow AI adoption down for its own sake. Good governance should give an organisation the confidence to move faster, not the excuse to stand still. The organisations that get the most from AI tend to be the ones that put simple, proportionate structure around it early, rather than the ones that adopt the most tools the fastest.
If your board is weighing an AI investment and wants an independent view before it commits, get in touch or read more about AI deployment and board advisory support.