The Fact & Gap Checker automates brand fact verification by extracting live findings through a grounded research loop and benchmarking them directly against the property's claims ledger.

Using a structured JSON schema, a Google Gemini model evaluates each discovered fact against known claims, categorizing it as confirming, contradicting, or identifying a new gap in the model's memory. Confirmed and contradicted items update the corresponding claim's verdict and attach verifiable source URLs as evidence, while manual human rulings remain permanently protected from automated overwrites.