AI, machine learning help generate vivid property data details
Arturo has long generated property-focused AI-driven analytics for insurers. But the company – spun out of American Family Insurance in 2018 – is turning to the mortgage industry as it continues to expand.
“There’s great interest from major lenders around how they can utilize that same information we’ve historically and currently are providing to the residential P&C insurers,” Arturo CEO John-Isaac “JC” Clark (pictured) told Mortgage Professional America.
Clark said that Arturo is close to consummating a few mortgage-related deals in Australia and is in discussions with several major players in the United States.
“We are effectively taking the exact same product we sell our leading US and Australia insurers today… to those lending customers,” Clark explained.
Read more: Best zip codes to invest in property revealed
To be sure, the Chicago based company has had some early traction in insurance. It has raised $33 million from its combined Series A and B rounds, supporting its use of machine learning and artificial intelligence models to analyze property images. So far, the focus has been primarily on residential properties but Clark noted the company has expanded to provide analysis for commercial structures too. Images collected for analysis come from aerial imaging technologies but can also include satellite, stratospheric balloon captures, drones, or sometimes from ground level.
“Fundamentally, we’re using AI to analyze those images and extract out property characteristics – what is on the property as well as physical dimensions, conditions, materials, and risk factors to help our customers make better decisions about an economic transaction conducted on that property,” Clark said.
That approach works just as well for the mortgage industry, Clark explained.
“Both are regulated industries. They both have a fairly established lifecycle, like of an insurance policy or a mortgage loan. They both have to be quoted. They both have to be underwritten,” Clark noted. “You consistently pay on your homeowners insurance policy [and] you consistently pay on your mortgage, and both are subject to risk… which has an impact on the lender and the insurer.”
Clark said that mortgage presented itself as an attractive option because of those correlations. Mortgage was also in some ways an original part of Arturo’s DNA anyway.
“One of the original theses behind the [Arturo] spin out was that this technology incepted in American Family could have great applicability beyond just the insurance sector,” Clark said. “Where we began to explore it more in earnest is when multiple customers who are businesses – banks who can do lending on property – approached us and said ‘we’re very interested in this data.’”
As a result, Clark said, some early exploratory partnerships are starting to bear fruit.
“Over the course of 2021 we partnered to really explore the utility of the product… and are now very soon going to concentrate on business relationships based on their evaluation of its utility,” Clark said.
Same integrations, different industry
Clark expects integration processes for mortgage will be the same as they have been for Arturo’s insurance customers. The company provides three options.
“The first is our largest and oldest customers have integration directly with Arturo via API, an application programming interface where they’re directly talking to our machine learning infrastructure, and they’re pulling out [data] on a per property basis whatever they need, be that a quote [or] automated underwriting, pulling out millions of properties to do a portfolio analysis,” Clark said. “That’s the most sophisticated, although relatively simple integration.”
For this option, customers send Arturo a property address and what attributes they want to look at, and the company then does a quick property analysis, with 3D digital surface models around the properties in question.
Read next: CoreLogic collaborates with Google to offer 'game-changing services'
The second option involves a larger scale analysis where the client sends Arturo a list of properties, which Arturo analyzes and returns to the customer. Customers define the reach, though internal APIs will be incorporated into the process later this year.
For option three, Arturo provides web-based tools that let customers handle much of the process themselves – logging into a web application and typing in an address they want to analyze, after which the address generates an analysis. This option, Clark said, is known internally as “Day Zero ROI” because customers just need a list of properties in the system or to type in a property to immediately access analyses without any integrations.
For all, once contracts are signed, customers can sync or integrate with Arturo’s platform within two to four weeks, Clark said.
After a year of research and focus, Clark said Arturo is ready for mortgage and excited about the potential the market offers.
“Your general homeowner’s policy in the US, the average is around $1,700 per year in annual premium, but the average single family home mortgage is hovering around $250,000,” Clark said. “I see mortgages as a huge additional vertical for Arturo to bring value to.”
