“We are proving that the models that we have built for the property and casualty insurance use case are widely applicable for the $12 trillion U.S. mortgage servicing market as well,” said Attila Toth, founder and CEO of ZestyAI, in an interview.
The technology aims to help servicers size up risks to the collateral securing loans in their portfolios using various data sources, including information gathered about properties remotely through images.
“If they have an aerial image, or if they have a building permit, they cannot use that as an input. It’s an unstructured data source. So we are using machine learning to tease out structured data,’ said Toth.
Eventually, Zesty.AI plans to offer mortgage companies the analytics that allow insurers to size up risks to properties from natural disasters like Hurricane Ian or wildfires.
The analytics will, for example, calculate the annualized probability that a property will be in a wildfire perimeter based on location. It can assess the likelihood it might burn if so, based on the characteristics of the structure involved, including the pitch of the roof, the materials it’s made of and whether the home has overhanging vegetation.
“The first step is providing [mortgage companies] with property valuation and property characteristics, and then we will take it to the next step when we are going to also provide climate risk metrics,” said Toth.