1. Purpose of the gate
Use case accepted is the first funding commitment the organisation makes to an AI-assisted step. Use case proposed, at the start of the stage, only gave a small team the time to look; this gate decides whether what they found is worth an evaluation. It is not a decision to buy a tool or to change the flow. It is a decision that one named flow step is a sensible place to look, that the organisation knows how that step performs today well enough to measure an improvement, and that someone with authority will pay for an evaluation plan. Everything later in the lifecycle inherits the choice made here, so the review is short in time but should be careful in substance.
The gate tests four things, which are its criteria: a use case proposal that states the flow step, the expected benefit and the consequence of an undetected error; a flow baseline that records the current method, its effort and its quality measures; a classification by sensitivity of the design data the step would touch; and a named sponsor and owner. A proposal can be enthusiastic about benefit and still pass, but it cannot pass if it is vague about what a wrong output would cost or about which data the tool would see.
The most useful thing reviewers can do at this gate is to slow down the jump from a general interest in AI to a particular tool. A proposal that names a supplier before it has described the step and its baseline is a tool looking for a use case, and the review should send it back.
10 more sections for registered members.