Use Cases / Auto Claims

Auto claims investigation, fully automated

Auto fraud costs insurers $29 billion annually. Staged collisions, phantom passengers, and body shop collusion require investigation resources most carriers don't have. Hesper AI investigates every auto claim in 2-4 hours.

Investigation in progress

AUTO-2026-089237

Rear-end BI · $48,400 exposure

Running

12m elapsed

Progress20%

Investigation phases

✓

Document ingest & intake triage3m

VIN + NMVTIS title verification

Scene photo pixel-level forensics

Medical record cross-reference

Ring & provider network analysis

Evidence gathered2 items

PolicyCoverage verified - active since 2024-03-12

NMVTISVehicle has prior salvage title (2021)⚠

Risk score

Low signal

28

/ 100

[01] Fraud patterns

Common auto claims fraud schemes

Auto fraud ranges from individual exaggeration to sophisticated rings involving drivers, attorneys, medical providers, and body shops working in coordination. Hesper investigates every flagged claim against all four schemes simultaneously.

Est. loss impact

~$10B

Staged accidents

Organized rings orchestrate collisions with predetermined participants and pre-arranged witnesses. Hesper AI cross-references participant histories, analyzes accident scene photos for staging indicators, and maps connections between drivers, passengers, attorneys, and medical providers.

How Hesper detects it

Scene photo forensicsParticipant historyAttorney network mappingCross-claim matching

Est. loss impact

~$4B

Phantom passengers

Claimants add non-existent passengers to increase injury payouts. Hesper AI verifies passenger identities against DMV records, cross-references medical treatment timelines, and detects patterns of the same individuals appearing across multiple unrelated accidents.

How Hesper detects it

DMV identity checkMedical timeline analysisCross-claim matchingOSINT verification

Est. loss impact

~$9B

Inflated injury claims

Minor accidents with exaggerated medical treatment. Hesper AI analyzes medical records for altered billing codes, detects inconsistencies between reported injuries and accident severity, and flags treatment patterns that exceed norms for the type of collision.

How Hesper detects it

Medical record forensicsInjury-severity correlationBilling code analysisProvider benchmarking

Est. loss impact

~$6B

Repair shop collusion

Body shops inflate repair estimates or bill for work not performed. Hesper AI detects altered amounts in repair invoices, maps relationships between claimants and specific shops, and cross-references repair costs against regional benchmarks and OEM parts databases.

How Hesper detects it

Invoice alterationShop network analysisCost benchmarkingParts verification

[02] Timeline compression

Manual workflow vs. Hesper

Every phase compresses. The cumulative effect is the difference between a two-week investigator assignment and a same-day decision.

Investigation phase

Manual workflow

Hesper AI

Photo and doc forensics

Investigator review, 6-8 hrs

Automated, 3 min

Vehicle history lookup

Manual NMVTIS queries, 1 day wait

API real-time, 30 sec

Medical record analysis

Peer review vendor, 3-5 days

AI analysis, 14 min

Network / ring mapping

Investigator diagramming, 1-2 days

Auto-graph, 10 min

Report assembly

Investigator write-up, 4-8 hrs

Auto-generated, 0 min

Total time

14-21 days

2-4 hours

[03] Investigation flow

How Hesper AI investigates an auto claim

Every auto claim runs through a structured investigation pipeline. Phases run in parallel where dependencies allow - the full sequence completes inside a few hours.

Document and photo ingestion

~3 min

Accident photos, police reports, medical records, repair estimates, and rental receipts are ingested simultaneously. Each document is analyzed across 200+ fraud signals at the pixel level before any text extraction.

Pixel-level forensicsMetadata analysisEXIF cross-checkPDF layer inspection

Vehicle and identity verification

~8 min

Vehicle VINs are checked against NMVTIS for title history, salvage records, and prior damage. Driver and passenger identities are verified against DMV records and cross-referenced with prior claims in ISO ClaimSearch and NICB.

NMVTIS lookupDMV cross-referenceISO ClaimSearchNICB match

Medical treatment validation

~14 min

Medical records are analyzed for altered billing codes, fabricated treatment dates, and inflated charges. Treatment timelines are cross-referenced with the accident date and injury severity to detect patterns inconsistent with the reported impact.

Billing code analysisTimeline verificationTreatment-severity matchProvider benchmarking

Network and ring analysis

~10 min

Hesper AI maps connections between all parties - drivers, passengers, witnesses, attorneys, medical providers, body shops. Shared service providers across multiple claims are flagged as potential organized fraud indicators.

Entity resolutionNetwork graphRing pattern detectionProvider overlap

Investigation report and resolution

~12 min

A complete investigation report is generated with evidence citations, network diagrams, risk scores, and recommended actions. Denial justifications and referral packages are prepared for SIU review or law enforcement referral.

Cited evidenceNetwork diagramsRisk scoringSIU-ready output

[04] By the numbers

$0B

Annual auto insurance fraud in the US

Per Coalition Against Insurance Fraud

Of auto injury claims involve fraud

Higher than P&C average

2-4h

Hesper AI investigation time per claim

Vs. 14+ days manually

Fraud signals analyzed per document

Pixel-level plus text-level