GeoX AI US Property Database (USDB)

Property underwriting in the US often rests on self-reported applications, tax records and inspections that are outdated or missing, leaving roof condition, vegetation risk and other exposures unverified. The GeoX AI US Property Database (USDB) addresses this for property and casualty insurers with a pre-processed dataset covering 186.2 million commercial and residential properties across all 50 US states (97% nationwide coverage, according to the provider).

GeoX AI derives the attributes from high-resolution aerial and satellite imagery using proprietary computer vision models; the provider does not name its imagery suppliers. Each property carries 64+ attributes: building data (footprint area, year built, occupancy), 3D data (building height, number of stories, commercial square footage), roof data (material, type, rust, ponding, discoloration, tarp evidence, solar panel area), vegetation and defensible-space measures, distances to coast, water bodies and fire stations, and parcel features such as pools and trampolines. Roof condition is scored from 0 to 100 on a six-label scale, and each score is traceable to visual evidence and carries its image date and model run date. The database is refreshed quarterly, and multiple imagery dates per property support change detection at renewal.

Because every record is processed in advance, the USDB is delivered through a RESTful API with sub-second response, as CSV or GeoPackage batch files for portfolio analysis and catastrophe models, or through the GeoX AI web platform. Its attributes are also available inside Cytora's risk processing platform. SC Farm Bureau's regional underwriting manager reports that testing GeoX AI against another vendor showed a clear accuracy advantage.