We do use the privacy APIs (when available) for requests, but we also extract schemas + data points from each silo for PII scanning and data mapping.