450 findings on 60,762 real NHTSA crashes.
60,762 crashes from the NHTSA Fatality Analysis Reporting System, run through our P&C integrity pack: VIN repeat detection, geographic crash clustering, make/model fatal-rate outliers. 450 deterministic findings, sealed CREB™ on each. Real federal data; the same engine that surfaces these patterns in the public register surfaces them in your claims queue.
JIL is not a claims management system, not an SIU case manager, not a recovery vendor, not a reinsurance analytics platform. It is a payment-integrity layer between the carrier's claims platform and the payment rail - producing court-admissible evidence packages at the moment of detection, admissible in civil subrogation, NICB referrals, SIU investigations, state fraud bureau filings, DOJ referrals, and reinsurance recovery. Guidewire, Duck Creek, Insurity, and ISO ClaimSearch don't produce court-admissible packages. That is the wedge.
Here's what we surface for you.
If you're a P&C insurance carrier (auto, home, commercial property, general liability), an SIU lead, a claims VP, or a state DOI fraud-bureau investigator, this is what JIL detects in P&C claim flows. JIL plugs in between your claim adjudication system and your payment-release function, runs 500+ checks per claim (148 attestation catalog + 27 P&C-specific), and produces sealed evidence per finding. Loss-ratio impact tied to the specific finding pattern.
Staged-loss ring
Multiple "victims" filing claims linked by shared bank fingerprint, shared attorney, shared body shop, shared medical provider, or shared address cluster. JIL surfaces the network pattern across your active claim queue - the kind of ring that takes a SIU investigator months to map by hand surfaces in seconds with the entity-graph evidence.
Inflated repair / vehicle-valuation fraud
Claim amount is 4x the median for the same vehicle / damage type / zip code. Repair invoice lists parts at 2x manufacturer-published pricing. JIL surfaces the outlier with the comparison data your adjuster can act on - faster than re-running the appraiser.
Ghost contractor / phantom service provider
Repair shop registered 60 days ago, no commercial-utility records, no prior tax filings, beneficial owner has 3 other entities that received claim payments and dissolved within a year. JIL flags the ghost-contractor pattern before the disbursement.
Faked injury claim with clinic kickbacks
Same chiropractor + same attorney + same diagnosis pattern across 47 claims in 6 months. Or a clinic with a beneficial owner who shares a bank fingerprint with the attorney's firm and the body shop. JIL surfaces the kickback ring with entity-graph evidence.
Out-of-state policy mismatch
Claim from a vehicle registered in a state where the named insured doesn't actually live (premium-evasion via out-of-state garaging). Or commercial property claim from a property the insured doesn't own. JIL cross-checks against state DMV records, state real-estate records, and policy data.
Sealed evidence per blocked claim
Every claim JIL holds gets a sealed evidence record showing exactly why. Useful for SIU investigation, defending the hold to the policyholder or the regulator, and (if it escalates) civil litigation or criminal referral. NICB referral-ready.
How to use it: integrate at the claims system (Guidewire, Duck Creek, Insurity). Per-claim pricing, with volume tiers for carrier-scale processing. See the live POC with 450 findings against synthetic P&C claim batches.
60,762 real crashes from NHTSA FARS audited.
On day one, pc-engine ingested 60,762 real fatal-crash records from NHTSA's FARS National CSV (calendar year 2022) and ran three deterministic checks: VIN-repeat detection, geographic crash clustering at 500m radius, and make/model fatal-rate outlier scoring. It surfaced 450 Tier 1 findings on real federal data - VIN-repeat clusters that suggest cloning or salvage-title fraud, geographic clusters that match staged-accident-ring patterns, and make-model cohorts with fatal rates 2x peer mean.
What we uncovered (top 5)
| Subject | Category | Finding |
|---|---|---|
| VIN-repeat clusters | Same VIN in multiple distinct crash records | VIN cloning / salvage-title fraud signal |
| 500m geographic clusters | 5+ crashes within 500m radius | Staged-accident-ring location pattern |
| Make/model fatal-rate outliers | >=2x peer-cohort fatal-crash rate | NHTSA defects-investigation precursor |
| Cross-state VIN appearances | Same VIN in 3+ states | Critical-severity escalation |
| Real NHTSA federal data | FARS 2022 National CSV (60,762 vehicle records) | Replayable bit-for-bit |
NHTSA FARS (Fatality Analysis Reporting System)
Annual federal CSV (FARS{year}NationalCSV.zip). No subscription, no DUA, no per-record licensing. The full ingest manifest is replayable bit-identically.
Deterministic pass
Tier 1 ran a single deterministic pass against the ingested public dataset. SQL aggregates only - no stochastic LLM in the verdict path. Ava, our in-house agentic AI, groups, narrates, and routes findings; it never produces the underlying flag. Same kernel that ships the other 7 verticals.
What this means for your business
For a P&C carrier's SIU, an NAIC anti-fraud bureau investigator, or NHTSA defects engineer, this is the VIN + geo-cluster + make-model-cohort queue that the public FARS dataset implies but does not surface. Each finding ships with a sealed CREB™ suitable for NAIC referral, NHTSA TREAD Act reporting, or carrier-internal claim-cohort exclusion from rate-making.
What Tier 2 unlocks. Customer engagement adds carrier-supplied claim records, repair estimates, telematics, photos, and the named-actor (claimant + provider + body-shop) cross-reference. Tier 2 confirms or rules out the staged-accident-ring or VIN-cloning interpretation and produces a sealed CREB™ suitable for state insurance fraud bureau referral or carrier-internal subrogation.
P&C Insurance Integrity at a glance.
Where the integrity layer sits, what it produces, and how the sealed CREB™ flows back to the buyer's existing systems.
Where this product earns its place.
The strategic case for P&C as a JIL line of business - what makes the wedge defensible, what makes it economically meaningful, and how it compounds with the rest of the platform.
Sealed at the moment of detection
Existing tools (Guidewire, Duck Creek, Insurity, ISO ClaimSearch) flag suspicious claims. None produce FRE 902(14) court-admissible evidence at the moment of detection. JIL ships a sealed CREB™ on every flagged transaction.
Marketplace listings on top three platforms
Pre-Settlement integrates into the claims-platform payment-authorization workflow via Guidewire Marketplace, Duck Creek Anywhere, and equivalent. ~40% of US carriers run Guidewire ClaimCenter; the marketplace path is the standard procurement motion.
~95-105 healthcare checks transfer
Identity, duplicate, velocity, network, geographic-temporal, document integrity, sanctions, beneficiary, settlement integrity, and behavioral signals all transfer with minor schema retuning. The wedge is the net-new P&C-specific checks.
Detection and proof, not recovery
Pricing carries over from the MCO playbook unchanged - flat-fee scans, subscriptions, per-claim CREB™ bundles. No percentage-of-recovery, no contingency. JIL is detection and proof; recovery sits with the carrier or its existing partners.
Net-new checks, sealed evidence.
A representative slice of the P&C-specific check pack. Each one runs in the same five-stage pipeline as the rest of the platform - intake, profile load, parallel checks, verdict, sealed CREB™ - and ships with a 14-of-20 BFT signature, a CourtChain™ L1 anchor, and a reproducibility manifest pinning the exact check-logic version.
VIN Integrity
Title washing, salvage rebrand, NMVTIS divergence detection across the title chain. Cross-references state DMV, NICB theft data, and commercial vehicle-history sources.
Staged-Accident Pattern
Claim clustering by VIN, address, attorney, clinic, and time window. Reuses provider-network primitives from healthcare for attorney-clinic ring detection.
Repair Estimate Anomaly
Line-item deviation from CCC ONE, Mitchell, and Audatex baselines for the year-make-model-region. Flags persistent over-estimation patterns by shop.
Damage Narrative Inconsistency
Claimed mechanism of loss versus photo-documented damage geometry. Photo and document forensics including EXIF integrity, prior-loss image reuse, and AI-generated image detection.
Weather and Event Correlation
NOAA verification of claimed hailstorm, hurricane, freeze, or wind event at the coordinate and time of loss. Catastrophe-event opportunism detection.
Provider-Mill Detection (BI)
Bodily-injury treatment-pattern analysis - clinic-attorney pairing, treatment duration versus injury severity, billing-code stacking. Reuses healthcare provider-fraud architecture.
Total-Loss Valuation Drift
Carrier valuation versus market comp from the live used-vehicle market. Surfaces undervaluation disputes alongside fraud-side valuation manipulation.
Premium-Fraud Signal
Garaging-address mismatch, undisclosed driver, vehicle-use misrepresentation. Surfaces the underwriting-side mirror of claim-side fraud.
Who runs this in production.
The buyer pattern for P&C - who carries the budget, who carries the regulatory exposure, and how the engagement starts. Most first engagements are a Retroactive Proof Audit on a defined lookback window; Pre-Settlement integration follows once the check pack is calibrated to the customer's profile.
Direct enterprise sales
Progressive, Allstate, State Farm, GEICO, Travelers, Liberty Mutual, Farmers, Nationwide, USAA. Direct enterprise sales into claims, SIU, and integrity teams.
Cross-carrier reach
Sedgwick, Crawford, Gallagher Bassett. Cross-carrier reach and standardized integration paths. TPA Pre-Settlement integration is a fast lane to scale.
Faster procurement cycle
Specialty lines, MGAs, captive carriers - smaller buyer base but faster procurement cycle than Tier 1 personal-lines.
Mandatory reporting
State insurance fraud bureaus (mandatory reporting in most states), state Attorneys General, NICB, DOJ Insurance Fraud Unit for federal mail/wire-fraud predicates, reinsurance carriers for treaty-claim documentation.
Four-SKU model. No percentage. No contingency.
Pricing carries over from the canonical four-SKU model unchanged - Retroactive Scan (flat fee), Retroactive Proof Audit (with credit-back against the next subscription tier), Pre-Settlement Subscription (tiered annual), and per-case CREB™ bundles (Tier 3 court-ready evidence). Asset Intelligence is the standard fifth SKU where the vertical needs it.
No percentage of recovery. No contingency. No success fees. JIL is detection and proof, not recovery. Recovery sits with the customer or its existing partners (subrogation counsel, recovery vendors, regulators). The structure is what allows JIL to operate as neutral integrity infrastructure across plaintiffs and defendants, payers and payees, regulators and the regulated, on the same case.
Ready to scope a P&C engagement?
Initial briefings are 60 minutes. Retroactive Proof Audit lookback windows, check-pack profile design, and integration runbook are available under NDA. We start where the buyer's procurement gate is shortest.
One kernel. Eight industries.
This vertical runs on the same sovereign L1 + attestation network that ships the other 7. Kernel age: 18+ months. Adding a vertical: ~1 week. Competitor moat: build the kernel first.
See also -- fraud-pillar counterparts
P&C and Workers Compensation Fraud Detection -- If you're running a SIU and need cross-carrier intelligence on the $334B P&C + WC fraud pool, the fraud-pillar engagement at /products/pc-insurance-fraud offers 12.5% flat + 10% contingency + 4yr retroactive scan.