Instead of sending a person to walk through a cornfield after a hailstorm, Verdex uses satellites to instantly see what happened and verify the farmer's claim.

Verdex ingests high-resolution satellite imagery (likely from providers like Planet Labs, Maxar, or Sentinel-2) and applies computer vision models trained on labeled crop damage datasets to classify field-level conditions — healthy, stressed, partially damaged, or destroyed. These classifications are cross-referenced against submitted insurance claims to flag discrepancies, confirm legitimate losses, and prioritize cases that require human review. The platform currently covers over 100 million acres for a major US crop insurer, enabling near-real-time audits at a scale that would require thousands of field adjusters to replicate manually. By automating the verification pipeline, Verdex dramatically reduces the cost per claim audit while improving consistency and reducing fraud exposure.

It's like replacing a building inspector who drives to every house with a drone that photographs the entire neighborhood in five minutes and highlights exactly which roofs have damage.

Verdex compares satellite photos of a farm before and after a reported disaster to catch claims that don't match what actually happened on the ground.

Crop insurance fraud costs the US federal crop insurance program hundreds of millions annually, often involving exaggerated damage reports or claims filed for fields that were never planted. Verdex likely builds temporal analysis pipelines that compare multi-date satellite imagery — capturing vegetation indices (e.g., NDVI), soil moisture, and canopy cover — before and after a reported loss event. Machine learning models trained on historical fraud cases can identify statistical anomalies: fields that show no change despite a claimed total loss, damage patterns inconsistent with the reported peril (e.g., hail vs. drought), or planting activity that contradicts policy declarations. By automating this forensic analysis across millions of acres simultaneously, Verdex provides insurers with a scalable fraud detection layer that would be impossible to replicate with manual spot-checks alone.

It's like having a security camera that recorded the parking lot before and after someone claims their car was damaged — except the parking lot is 100 million acres of farmland.

Verdex uses years of satellite data to tell insurers which specific fields are riskier to insure before they ever write a policy — like a credit score, but for farmland.

Beyond claim verification, Verdex's satellite imagery pipeline and ML infrastructure position it to offer predictive underwriting intelligence. By analyzing multi-year historical satellite archives — capturing crop rotation patterns, soil health trends, drainage characteristics, historical yield variability, and microclimate exposure — Verdex can generate field-level risk scores that far exceed the granularity of traditional county-level actuarial tables. Machine learning models (likely gradient-boosted ensembles or transformer-based architectures) trained on historical loss data correlated with geospatial features can predict which fields are most likely to file claims in a given season. This enables insurers to price policies more accurately, identify adverse selection, and allocate reserves more efficiently. As Verdex expands into property and energy insurance, this same predictive framework could assess wildfire exposure, flood risk, or solar panel degradation — all from satellite-observable features.

It's like a real estate appraiser who can instantly evaluate every house in the country from space instead of driving to each one with a clipboard.