Every downstream product — the valuation, the matching, the listing — depends on one thing being right: the agency's data. If the inputs are wrong, the valuation is wrong and the matches are wrong, no matter how good the models are. The ingestion pipeline is the layer that gets the data right. It pulls structured records from the agency's management system, reconciles them, and quality-gates them before anything else runs.
From the system of record.
Agencies run on agency-management systems — Hawksoft, Vertafore, and others — that hold the book's structured history: policies, premiums, carriers, lines of business, retention, and the dates that define the book's trajectory. The pipeline pulls from that system of record directly, rather than asking the owner to fill in a questionnaire from memory. The difference matters: a valuation built on the actual book is defensible in diligence; one built on recollection is not.
One shape.
Each management system stores its data differently — different fields, different conventions, different ways of recording the same fact. Ingestion maps all of them to one normalized model, so the valuation and matching engines consume a single consistent shape regardless of which system the data came from. This is what lets the platform treat a Hawksoft agency and a Vertafore agency on equal terms, and it is why the same valuation discipline applies uniformly across the marketplace. The normalized data feeds the matching and appetite-scoring engine directly.
What earns publication.
Two checks stand between raw data and a live listing. Carrier-statement reconciliation cross-checks the management-system figures against carrier records, catching the discrepancies that would otherwise surface — embarrassingly — in a buyer's diligence. Then data-quality gates decide whether the listing publishes or sits in review: a book with clean, reconciled, complete data publishes; a book with gaps is held until they are resolved.
- Pull structured records from the agency-management system — Hawksoft, Vertafore, and others.
- Normalize every source to one consistent data model.
- Reconcile management-system figures against carrier statements.
- Apply data-quality gates — publish clean books, hold incomplete ones in review.
The result is a listing a buyer can trust on sight, because the data behind it was reconciled and gated before publication. The scored outputs the pipeline produces are covered in the matching engine page, and the reporting cadence in the KPI orchestration page. The whole pipeline sits within the platform's intelligence layer.