S.07CLAIM TRIAGE

claim triage.

Read the claim. Classify the peril. Assess severity. Route to the right adjuster before the paperwork piles up.

pick a fixture_

paste your own_

turnstile · invisible

waiting for verification token...

~/demo/claim-triage.log

idle · waiting for input

model_ —

pick a fixture on the left, then press [ run demo ].

~/claims/triage

classifying…

assigned: ──────
tier: ──────

proposed · no side effects in demo mode

── log ──

no log entries yet

── how this works ──

01— ingest

first-notice-of-loss from call centers, web forms, or agent portals — unstructured narratives.

02— reason

classify peril, assess severity, scan for fraud indicators, check coverage.

03— act

assign to the right adjuster tier, queue next actions, escalate suspicious claims to SIU.

This demo runs on free-tier AI against fake data. In production, the same workflow writes to your real tools, logs every run, and is reversible — the reasoning step is the smallest part of the stack.

[ deploy one like this → ]