SIGNALRISK INTELLIGENCE PLATFORMWe read 4.2M claims in 11 seconds.Your loss ratio dropped 23%.
Satellite imagery. IoT sensor feeds. Claims language in real time. Signal surfaces patterns human actuaries miss by years — at a fraction of the latency.
$47B
Mispriced US Risk
94.6%
Confidence Interval
3.5K+
Data Sources
94.6%
CONFIDENCE
4.2M
CLAIMS PARSED
11s
LATENCY
-23%
LOSS Δ
$47B in risk priced
on dead data.
The US property insurance market isn't broken because carriers lack data — it's broken because the data pipeline from incident to model update takes 18 months. Signal compresses that to real time. The following figures are drawn from NAIC filings, Lloyd's of London aggregate reports, and Signal's proprietary loss database.
MISPRICED US PROPERTY RISK · 2025
$47B
Annual aggregate exposure priced on outdated actuarial tables with no real-time data ingestion.
CARRIERS STILL USING PRE-2010 MODELS
73%
Three-quarters of the US P&C market runs pricing engines built before climate acceleration data existed.
AVERAGE LAG: INCIDENT → RATE ADJUSTMENT
18mo
Legacy actuarial cycles mean carriers absorb 18 months of mispriced risk before any correction.
COMBINED RATIO DRAG FROM PRICING ERROR
6.2pts
The gap between what carriers charge and what they should — measured in lost combined ratio points.
The incumbents built on sand.
COBOL-era pricing engines were architected for a world where risk was stable and data was scarce. Neither is true anymore. Coastal flood frequency has shifted by 34% since the models were calibrated. The models don't know.
Human actuaries are the bottleneck.
The average P&C actuarial cycle takes 14 months from data collection to rate filing. In that window, a single hurricane season can invalidate every assumption. Signal closes that gap to 11 seconds.
CLAIMS PROCESSED TODAY
12,847,293
RISK EVENTS FLAGGED
1,429
LIVE COUNTER · RESETS AT 00:00 UTC
Four layers.
One verdict.
Signal's architecture eliminates every point where human latency compounds pricing error. Each layer feeds the next in under 3 seconds. The output is a confidence-weighted rate recommendation your system can act on immediately.
PIPELINE ARCHITECTURE
INGESTION LAYER
Satellite (30cm resolution), IoT sensors (wind/flood/fire), NOAA feeds, carrier claims filings
NLP PARSE ENGINE
Transformer model fine-tuned on 47M historical claims. Extracts causal chains, liability language, fraud signals
RISK SCORING MODEL
Ensemble: gradient boosting + neural attention. Outputs probability distribution per peril per location
RATE RECOMMENDATION
Confidence-weighted pricing delta. Integrates with carrier rating engines via REST or batch export
SIGNAL vs. INCUMBENT PRICING ENGINES
METHODOLOGY NOTE
Benchmarks derived from blind holdout test on 2.1M claims, Q1–Q4 2024. Legacy engine composite = average of three top-10 US P&C carrier internal models. Signal had no access to incumbent training data.
████████ Insurance
dropped 23 points.
Carrier name redacted per NDA. Loss ratio improvement curve is real. The carrier has since expanded Signal's model to their entire Gulf Coast and Southeast book — $4.2B in premium.
INTERNAL DOCUMENT — SIGNAL PROPRIETARY
CLEARANCE: RESTRICTED
Client: ████████ Mutual Insurance Group
Book: Gulf Coast residential property, ████ policies
Deployment: Q2 2024 · Full integration in 6 weeks
Pricing engine replaced: ██████ v3.1 (2009 vintage)
−23pt
LOSS RATIO
vs prior 12mo
6wk
INTEGRATION
REST API
$4.2B
BOOK EXPANDED
after pilot
REQUEST ACCESS
Request the
Full Dataset
Get Signal's complete risk pricing dataset: 47M claims, 3.5K data sources, full benchmark methodology. Carriers use it to validate the model before procurement. VCs use it to underwrite the moat.
Risk Pricing Brief — PDF
12-page methodology overview. Email only.
Three ways
in.
Signal is a seed-stage company. We're selective about who we work with because the model gets better with every deployment. Start with the research brief. Escalate when you're ready.
RESEARCH ACCESS
For analysts, investors, and actuaries who need to evaluate the model before any commercial conversation.
- Risk Pricing Brief (12pp PDF)
- Benchmark methodology summary
- Sample heatmap data (anonymized)
- One-page model architecture
PILOT PROGRAM
Live model integration on a defined book of business. Signal provides the model; you provide the claims data. Loss ratio improvement guaranteed by contract.
- Full dataset access (47M claims)
- REST API integration (6-week setup)
- Dedicated ML engineer during onboarding
- Benchmark report vs your current engine
- Contractual performance guarantee
ENTERPRISE LICENSE
Full model licensing for carriers and reinsurers. Includes white-label output, custom peril configurations, and SLA-backed uptime for rate-filing workflows.
- Unlimited claims processing
- Custom peril model configuration
- White-label rate recommendations
- SOC 2 Type II compliance docs
- Dedicated infrastructure node
- Quarterly model refresh cycle
SIGNAL IS SOC 2 TYPE II COMPLIANT · GDPR READY · NAIC REPORTING COMPATIBLE