The challenge
Underwriters and claims staff at NasClaims manually re-keyed data from ACORD forms, loss runs, FNOL reports and medical records arriving in inconsistent formats. Every document meant hand entry, and errors crept in at every step.
Underwriting and claims decisions were waiting on data entry rather than judgment. The extracted data had to land in their existing systems — not a new tool the team would have to learn.
What we built
A document pipeline that classifies each incoming document type automatically — ACORD forms, loss runs, FNOL reports, medical records. It extracts and validates fields against business rules, routes low-confidence items to a human reviewer with the original shown alongside, and writes clean data directly into the policy and claims systems through their existing APIs.
Every extraction is logged so decisions stay fully auditable. The team reviews exceptions, not every document.
The result
- 75% less manual data entry per document.
- Intake cut from days to minutes per submission.
- 95%+ of structured fields accepted without correction.
- Staff time moved from re-keying to review and decisions.