The mayor’s office wants reviewers to see why the policy bot cited a particular ordinance section. What should the solution emphasize?

When the mayor’s office needs to see why a policy bot cited a particular ordinance section, the solution should emphasize explanation and instrumentation that show supporting sources for grounded civic policy answers. This works because grounded Foundry answers can retain retrieval metadata—document IDs, ordinance section numbers, passages, and scores—and expose that evidence in reviewer UI or logs. Explanation tooling makes citations challengeable and trustworthy, which is the transparency property leadership is asking for.

Groundedness scoring dashboards that rate factual alignment across grounded civic policy answers miss the point because rating overall factual alignment does not show which ordinance section grounded a particular citation, so reviewers still cannot see the evidentiary basis for the policy answer. Citation style controls that standardize ordinance references across grounded civic policy answers fail the mayor's office request because standardizing formatting is presentation, not evidence tying a specific answer to the exact ordinance section or passage the model actually used. Retrieval tuning workflows that improve search ranking for grounded civic policy answers fail because better search ranking does not expose the specific sources behind an individual cited ordinance section, leaving reviewers without the per-answer provenance they need.

Exam caveat: showing sources is not the same as guaranteeing correctness—reviewers still must validate that the cited section actually supports the claim. Operational check: persist citation metadata with each policy answer, render sources in the reviewer experience, sample weekly for citation faithfulness, and prohibit production configs that strip retrieval provenance from stored civic policy conversations.