A benefits-eligibility team uses a model to auto-split every epic into stories and then treats the output as the plan. What should the Team Coach insist on?
Select an answer to reveal the explanation.
Short Explanation
Auto-split is a first draft of the sandwich, not lunch. The benefits team still decides the slices and whether each story is ready. Treating the model output as the plan is how you eat the wrapper.
Full Explanation
When a benefits-eligibility team uses a model to auto-split every epic into stories and treats the output as the plan, the Team Coach must insist the model remain a split-suggestion aid while the team still slices and validates readiness. Auto-split is a first draft, not lunch: humans decide the thin slices residents will actually receive and whether each story is ready for development. Surrendering that validation invents false plans and false confidence for public benefits work.
Committing the model's story list as the PI plan so refinement can be skipped fails because unverified output is not a plan the team owns or can empirically defend. Having the RTE accept the auto-split as ART-level scope fails because train facilitation is not a substitute for team slicing and readiness conversations. Freezing epic splitting until a vendor certifies the model as backlog owner fails because no vendor becomes the team's backlog authority under SAFe.
Exam caveat: AI can save facilitation setup time; it cannot own empiricism, readiness, or commitment on a benefits ART. Operational check: use model suggestions only as input in refinement, have the benefits team rewrite and validate each slice, confirm readiness with the Product Owner, and refuse to baseline PI scope on unreviewed auto-splits.