The system surfaces relevance gaps by querying models with binary recommendation probes across every tracked entity and query to identify which topics the AI consistently fails to recommend an owned brand.