All case studies

Insurance · A regional insurance carrier

Insurance: claims intake that no longer waits in a queue

The situation

Every claim arrived as a pile of documents — intake forms, photos’ descriptions, adjuster notes, policy language — and a person read the pile before work could start. The queue grew with every storm season, and the backlog priced itself in policyholder churn.

A general-purpose AI assistant misread the carrier’s policy terminology and coverage codes often enough that adjusters double-checked everything, which erased the time it was supposed to save.

What moved to Run BiOS

The carrier fine-tuned a model on its own claims history and policy documents through Run BiOS, running on dedicated GPUs billed per second, with the trained weights owned outright. Intake submissions now arrive pre-summarized in the carrier’s own terminology, flagged with the coverage codes that matter.

Adjusters review instead of read. The model handles the queue at storm-season volume without a premium-model bill growing underneath it.

Managed fine-tuningDedicated GPUsCustomer-owned weights

The outcome

  • AI spend reduced significantly against the general-purpose assistant it replaced
  • Claims cycle time reduced significantly — adjusters start at review, not at reading
  • Accuracy improved noticeably once the model spoke the carrier’s own policy language

Before

Every claim read end to end by a person; the queue set the pace

With Run BiOS

Pre-summarized intake in the carrier’s own terminology, queue absorbed

Illustrative, not measured.

Facing the same bill?

Estimate your workload on the rate card, or talk to us about the deployment pattern that fits.

More case studies