Tell me how you would tell a customer that their favourite AI use case should not be built.

Strong answer

Acknowledge the goal behind the idea, then show the evidence: the scoring against value, feasibility and risk, the data gap or regulatory exposure that makes it a poor first project. Never just say no; offer the alternative that gets them closer to the same goal sooner, and the conditions under which their idea becomes viable later ("once the contract data is digitised, this becomes a strong candidate"). Put it in writing so the sponsor can share the reasoning upward. Customers trust FDEs who save them from expensive failures more than those who say yes to everything.

Weak-answer indicators

• Builds it anyway to keep the customer happy
• Says no without evidence or an alternative