When you first buy a house, your insurance company doesn’t know very much about it or how much insuring it will cost. That’s because it first has to send out an inspector to look at the exterior of your house, take measurements, and check out your roof to see what kind of shape it’s in.
Cape Analytics, a Mountain View-based data analytics startup, aims to change all that. Its API-pipeline can feed an insurance company information about a house’s exterior square footage, roof type, roof condition, changes in a home and more – all thanks to the use of machine learning to analyze aerial imagery.
The company announced Monday that it’s launching data coverage for the entirety of the continental United States – over 70 million American homes – though it’s already been providing information to insurance customers like national reinsurer XL Catlin and the Florida-based Security First Insurance.
This deep-dive into a particular kind of data set is what CEO and cofounder Ryan Kottenstette says distinguishes his company from other imagery analytics firms like such as Orbital Insight.
“What makes us unique is that we provide property intelligence for enterprise customers that are interested in that, particularly insurance companies,” he said. “We are laser-focused on a vertical rollout as opposed to more of a platform approach.”
Kottenstette told me the company gets data from a number of sources, including drone data and satellite imagery from companies like DigitalGlobe. But the bulk of its information, he says, comes from aerial imagery taken from airplanes. In particular, Cape Analytics has a partnership with Google to utilize the images that it takes with its Earth observations. The company then feeds those images into its machine learning algorithms to develop information about particular homes.
Cape Analytics trains its machine learning algorithms to determine roof conditions
To process all that information and develop its algorithms, Cape Analytics employs a team of people to handle computer vision, machine learning and data science in order to get what data that insurance companies are looking for. This entails training its algorithms to identify things like the condition and type of roof a home has, what the building footprint is, and whether the home has a new addition, pool, solar panels, or other new construction.
“We’ve put together a world-class team of computer vision and AI people as well as data scientists,” said Kottenstette. “We do all that work in-house.”
That team includes the company’s cofounder and CTO Suat Gedikli, who has a doctorate in Computer Science and was previously a researcher at Willow Garage, where he developed computer vision models for robots. The company’s Vice President of Data Analytics, Olivier Collingnon, has worked on a number of data science projects, including developing the first catastrophe risk models for RMS.
Kottenstette says that instead of necessarily sending out an inspector, an insurance company can use Cape Analytics’ API to get most of the data that an inspector would gather nearly instantly, making it easier for insurance companies to determine the best price to insure a home.
“Our principal value proposition is improving insurance data,” he said. “It’s too expensive to inspect every home and homes change over time. So the data those companies have doesn’t reflect reality. When they work with us, we fix all that so they can price more accurately.”