Skip to content
Financial Services 2025 Project plus staffing

Claims, straight through. Seven days to two.

A regional property and casualty insurer handled first notice of loss through email, fax, and a portal from 2011. Adjusters spent their first hour of every case transcribing. The claims SVP wanted touchless claims and a better cycle time. We rebuilt intake, applied document AI with proper evaluation, and moved 80% of first notices to straight-through processing.

  • Claims cycle time
    7 to 2 days

    Median cycle from first notice to first payment decision.

  • Straight-through rate
    80%

    Of first notices now processed without manual transcription.

  • Adjuster capacity
    +22%

    Additional claims handled per adjuster per week.

01 / Challenge

What was hard.

First notice of loss arrived in five different formats. Adjusters transcribed every claim into the policy admin system before they could work it. Document accuracy mattered: a wrong policy number or date meant rework downstream. The claims SVP had tried an AI vendor pilot that failed evaluation in the first 100 claims. She wanted a system with accuracy gates a claims lead would sign, not a vendor promise.
02 / Approach

How we worked.

  1. 01

    Evaluation-first design

    We built the evaluation harness before we built the extraction pipeline. Held-out set of 500 historical FNOLs with ground truth. Accuracy thresholds agreed with the claims lead before development started.

  2. 02

    Document understanding pipeline

    Intake across email, fax, and portal into one normalized document store. Extraction with an LLM pipeline, validated against the evaluation harness on every model change.

  3. 03

    Policy admin integration

    Idempotent write path into the policy admin system. Audit trail for every claim touch. Human-in-the-loop review for anything that fell below the accuracy threshold.

  4. 04

    Adjuster workflow redesign

    Adjuster screen redesigned around the automated draft. Three clicks to accept, one click to escalate, a clear reason code for each.

  5. 05

    Model governance

    Model registry, versioning, and quarterly review. Drift monitoring on document types and extracted field accuracy. Runbooks for downgrade and rollback.

03 / Outcome

What we shipped.

The claims cycle time dropped from 7 days median to 2 days median across the pilot portfolio and held. Eighty percent of first notices processed without manual transcription. Adjuster capacity increased 22% without headcount change. The evaluation harness became the basis for quarterly model reviews with the claims lead signing each one.
  • 80% straight-through first notice rate inside the pilot portfolio.
  • Claims cycle time cut from 7 days to 2 days, median.
  • Adjuster productivity up 22% with no quality regression.
  • Quarterly model review signed by the claims lead, not by engineering.
Team shape

Who shipped it.

Duration
12 weeks
Team
4 people
Staffed role
ML engineer embedded post-launch
Ongoing
Quarterly model review

Start here

Tell us what you want to move.

Projects, managed services, or staffing. Start with a short note. We reply inside one business day with a clear read on scope, team, and timeline.

Talk to us

Reply within one business day