The Fact & Gap Checker automates verification by comparing newly discovered live-research findings against a brand's stored parametric memory claims using structured JSON decisions. When fresh findings validate or dispute model recall, it updates the claims table with correct or incorrect verdicts and saves supporting citations to claim_sources. Unrecalled findings are minted as new gap claims with unverified status to highlight missing brand knowledge without grading research against itself. Throughout this reconciliation, existing human rulings are preserved, ensuring automated audit passes never overwrite manual editorial judgements.