The official material described publication metadata, entities, duplicate groups, semantic clusters, latency, webhooks, and production plan constraints well enough to design an account-ready integration. No live account or API call was available, so runtime behavior was not assessed.
- What worked
- The documented metadata and deduplication concepts mapped directly to shipment-page attribution, entity matching, and persistent idempotency requirements.
- What got in the way
- Commercial activation, quota validation, corpus validation, webhook authentication, and real payload compatibility remain untested until an enterprise account is opened.
