A utility's executives selected a model vendor before anyone named a use case, a data owner, or a responsible AI review. What should the architect put in place first?
Select an answer to reveal the explanation.
Short Explanation
Picking the truck before the cargo is known is how the wrong vehicle shows up. You start with use cases, data governance, and responsible AI before the vendor choice is treated as settled. Buying the model first does not name the work.
Full Explanation
Cloud Adoption Framework strategy for AI starts from the work, the data, and the responsible AI duties, not from a vendor logo. This utility named a model vendor before a use case, a data owner, or a responsible AI review, so those three are the first work, and the vendor choice stays unsettled until they exist. Purchasing the model and writing use cases after production locks the spend to a tool no scenario has earned. A separate AI landing zone for that vendor builds platform ahead of strategy and splits off from the zone the utility would otherwise extend. A proof in every department on the chosen model, with no use case and no data owner, multiplies pilots that have no success criteria and no one accountable for the data. A vendor can be selected once there is something to fit the vendor to. Exam caveat: strategy is not a ban on talking to vendors. It is a ban on treating the purchase as the decision while the use cases are still blank. Operational check: publish the first named use case, the data owner for its grounding, and the responsible AI review, and hold the vendor commitment until those three are on the page.