Scientific claim evaluationat production speed.
Crash photos in. Delta-V, severity triage, fraud signals, and defense-ready documentation out. Five minutes at FNOL, not six weeks and a $10,000 expert report.
Our founders have consulted for
Claims run on narratives. Adjusters need physics at intake.
An adjuster opens a file at 7:14 AM. A bumper with a scuff. A $75,000 demand for two herniated discs and chronic pain. Six months later, if it goes to mediation, someone pays for a $10,000 reconstruction report that confirms what the physics told you at FNOL. The conversation is already over.
Objective severity. Documented for defense.
Every file gets the same analytical treatment, at the same speed, by the same validated method. What used to require an outside expert lives inside the claim note.
Delta-V & Crash Severity
Range estimate from photos. The single variable most predictive of injury.
Damage-Injury Mismatch
Automatic flag when claimed injuries exceed what crash physics can produce.
Severity Triage Score
0–100 index that drives reserves, routing, and specialty handler referral.
AIS Injury Probability
Likelihood by body region with biomechanical confidence scores.
SIU Fraud Signals
Documented physics basis for SIU referral. Better referrals, fewer false positives.
Defense-Ready Documentation
Daubert-documented methodology and exhibits, ready for counsel handoff.
Deterministic science. Not generative opinion.
FNOL to severity score, in one workflow.
Three steps. No separate tool to launch, no uploads to staging, no delays. Runs at intake, at triage, or at escalation — whenever your claims lifecycle needs it.
Photos land in the file
The adjuster uploads FNOL crash photos, or your SDK pulls them from the policyholder app automatically. No staging folder, no file conversions.
Analysis runs in parallel
Delta-V, PDOF, occupant kinematics, AIS probability. Validated physics against NHTSA and IIHS. Deterministic, auditable, and consistent file-to-file.
Results drop into the claim note
Severity score, reserves guidance, SIU flag status, and defense-ready documentation, all attached to the file with a permanent audit trail.
From intake to SIU, one analytical layer.
Claims leaders, adjusters, SIU investigators, and defense counsel work from the same physics. Nobody re-litigates severity at the next handoff.
For Claims Operations
- Consistent severity triage across every BI file
- Reserve guidance anchored in physics, not estimation
- Cycle time cut from months to days on the files that benefit most
- Cleaner handoffs to supervisors and defense counsel
- Measurable impact on claim leakage and indemnity spend
For SIU & Defense
- Automatic damage-vs-injury mismatch flagging with documented rationale
- Better referral quality and fewer false positives
- Physics-based foundation for subrogation and recovery
- Daubert-documented baseline that defense counsel inherits cleanly
- Fraud patterns visible early, not after the demand letter
The physics used to live at month four of the file, after we'd already set reserves. Now it's there at FNOL. The same adjuster makes a different decision.
A carrier VP of Claims, on the operational shift
SDK for claims platforms and policy cores.
Drop Silent Witness into Guidewire, Duck Creek, Snapsheet, or your own stack. Trigger analysis at FNOL, at reserve-setting, or on supervisor escalation. Webhook in, structured JSON out, full audit trail preserved.
import { SilentWitness } from "@silent-witness/sdk";
const sw = new SilentWitness({ apiKey: process.env.SW_KEY });
// Webhook handler for FNOL event
const report = await sw.analyze({
claimId: "BI-2026-07-0142",
photos: await s3.fetch(fnol.photoKeys),
lossDate: fnol.lossDate,
});
await claimsCore.attach(fnol.claimId, {
deltaV: report.deltaV, // { min: 7.2, max: 9.1 }
severityScore: report.severity, // 34 of 100
siuFlag: report.mismatchFlag, // true | false
reserveGuidance: report.reserves, // structured recommendation
});