Last year at InsureTech Connect we outlined six amazing technologies already reshaping insurance: drones / robotics, conversational artificial intelligence (AI), computer vision, sensors, blockchain, and genomics. So what’s happened in that time in these areas, and what’s new and upcoming?
Over the next few posts let’s revisit these themes and see where things are, and where they could go. One important lesson: innovation in insurtech doesn’t take place in isolation. Breakthroughs in one technology often have carry-over effects into others. Take drones: robotic technology plus computer vision, enabled by AI and machine learning, turns into powerful tools to reshape risk assessment and claims.
Drones are making their way into everyday life. They track storms, make deliveries, film sporting events, monitor air pollution levels and increasingly perform insurance inspections. For instance, with these small-unmanned aircraft, insurers can access property damage faster and in some cases more accurately than human adjusters. Drones can also reach areas difficult for people to access from sunbaked rooftops to flooded-out buildings. But it’s not just about taking pictures. To get real value from drones insurers need to be able to quickly and accurately make sense of the data they collect, data captured in dozens of images.
The story of Betterview, one of the first companies to use drones for insurance underwriting and claims adjudications, illustrates this point well. The company, which was founded in 2014 and now employs 4,500 pilots, quickly learned that it not only had to analyze drone footage for its clients, but also had to put the images into a larger context by providing additional information, like historical weather patterns and previous building repairs.
“We’re driving towards providing something like a credit score for buildings and rooftops,” said David Tobias co-founder and chief operating officer in an interview at last year’s Oliver Wyman Insuretech Connect (see the full interview with David here). The key, he said, is to turn dozens of images of a single property into actionable information for its clients, who are mostly insurers, inspection companies and claims management firms.
Tobias began his career working at an insurance inspection company founded by his father, Research Specialists Inc. Its customers often wanted information about rooftops, but accessing and inspecting the roof could be treacherous. In an audio interview with the website Insurance Nerd, Tobias recalled insurance adjustors in Arizona recoiling at the idea of doing a roof inspection in a 125 degree dry desert heat.
Drones seemed like the obvious solution. The drones could fly to the top of a building and look for damage. But when Tobias shared the drone’s footage with one of Betterview’s first clients, the clients wanted to know what the images meant. Basically the client wanted Betterview to analyze and interpret the pictures, he told Insurance Nerd. Tobias and his co founder David Lyman soon realized technology could help them and their clients make better sense of the images, much quicker. Betterview began using a new technology called computer vision. With computer vision, experts label certain elements of a picture, like missing shingles, pooling water or hail damage and then over time the computer learns to identify these features on its own via artificial intelligence.
This enables the company to quickly identify trouble spots– lose debris or objects blocking gutters or drains- and provide its customers with reports about existing and possibly future damage. Betterview also incorporates outside information into its analysis.
For instance, it can identify hail damage and then provide pertinent information about historical weather patterns, from average hail size to typical wind gusts and monthly perception statistics, extending back over the past five years. It also gives relevant building information, like previous repairs.
The company hopes to incorporate more outside data into its platform, Tobias said during the aforementioned interview at Oliver Wyman Insuretech conference. He used manned aircraft images as an example. These images, he said, could be used as a diagnostic tool since they are cheaper to obtain than drone footage.
“If you see the building has a poor roof [via the manned aircraft], then task the drone to collect the rest of the data,” he said during the Oliver Wyman conference interview. As the quote implies drones, may not always be best way to collect digital images, every single time.
For example, drones work well for accessing property damage in the aftermath of hurricanes. They can be launched into flooded areas from a dry spot several hundred feet away. However, drones were the wrong tool for property insurers in Napa Valley, who wanted to quickly get a handle on their potential losses after last year’s wildfires.
Drones are difficult to operate in smoke. Plus these clients were not looking for minute details, like missing shingles or hail damage, they were more interested in the big picture – was it a total loss, partial damage or had the property been spared, Tobias told Insurance Nerd, adding that for these questions, satellite imagery is often the best, and least expensive, option.
Where to now with drones? The trick, as David and others are learning, comes in the implementation. One key implication lies in regulatory oversight that can enable or hem in this implementation. BRINK news just published a piece by Richard Levick, Chairman and CEO of LEVICK, “Can Drones Escape the Coming Legal Minefields Unscathed?” that provides a helpful summary of the issues involved.