Weekly Audit of Unanswerable Policy Questions and Coverage Gaps

Closes the loop on your grounded Q&A system by surfacing what it could not answer. On a schedule, it pulls logged questions that retrieved low-confidence or empty matches, groups them by topic, and produces a ranked list of missing or thin policy areas your team should document in Confluence.

Use it when you run a contract clause Q&A assistant and want to systematically improve coverage instead of guessing what to write next. Ideal for legal-ops owners who maintain the policy library and need evidence-based priorities.

1. 1A weekly schedule triggers the flow.
2. 2It queries Supabase for the past week's Q&A logs where retrieval confidence fell below threshold or returned nothing.
3. 3OpenAI clusters the failed questions into themes and drafts a suggested policy topic for each gap.
4. 4A logic step ranks clusters by frequency and recency to set priority.
5. 5The ranked gap report posts to a Slack channel for the policy owner to triage into new Confluence pages.