ZestyAI submitted filings for its Z-WATER model across multiple states, including Maryland, Nevada, Georgia, North Carolina, New Jersey, Oregon, and South Dakota.

Z-WATER is a property-level risk model designed to assess non-weather water exposure for homeowners insurance. The model uses machine learning and combines property-specific and location-based data to predict both the likelihood and severity of water-related claims.

The model produces two key scores for each property:

• A claim frequency score indicating the relative likelihood of a non-weather water loss

• A claim severity score estimating expected loss as a percentage of Coverage A

Unlike traditional catastrophe models, Z-WATER is based on actual claims data rather than simulated events, allowing for more granular, property-specific risk insights.

ZestyAI is positioning the model as a tool for carriers to integrate into both underwriting and rating. Insurers can use the scores to refine eligibility, introduce new underwriting rules, or develop rate-neutral rating factors, with implementation handled through separate carrier filings.

The filings are structured as rate and rule submissions rather than standalone insurance products, signaling that Z-WATER is intended to sit alongside existing homeowners programs as an input to pricing and risk selection rather than a product itself.

South Dakota has already marked the filing as effective, while others remain under review, pointing to a broader multi-state rollout strategy.