A city AI strategy office is ranking three candidate AI pilots, permit-review triage, park-maintenance scheduling, and a resident chatbot, ahead of a budget cycle. The city's published strategic plan names reducing the permit-approval backlog as a top priority, but the resident chatbot vendor delivered by far the most polished live demonstration to city leadership. Which ranking approach correctly reflects sound prioritization practice?

Sound AI prioritization starts by mapping each candidate initiative against the organization's own documented objectives, and here only the permit-review triage pilot connects directly to a goal the city has already published and committed to, reducing the permit-approval backlog, which gives it a clear, defensible link between investment and mission outcome. Ranking by demonstration quality substitutes a vendor's presentation skill for an actual strategic case, and a polished demo says more about interface design and staged scenarios than about whether the underlying capability solves a problem the city has prioritized. Treating all three pilots as equal until after rollout defers a decision the strategic plan has already partly answered, and spending pilot resources evenly across unranked initiatives wastes scarce budget on lower-priority work instead of directing it toward the documented goal. Choosing based on implementation speed elevates convenience over impact, since a pilot that ships quickly but does not map to a stated objective still leaves the actual priority backlog unaddressed. A reasonable scope caveat is that a technically dazzling but strategically unaligned pilot might still be worth a small, low-cost trial for future learning, just not the office's top budget priority. A concrete check here is pulling the strategic plan's goal list and confirming, in writing, which specific goal each candidate pilot maps to before the ranking is finalized.