Takeaways:
- Near Space Labs produces higher quality images at scale
- Product supports underwriting in risky areas
- AI flawed at evaluating visual information
Aerial imagery company Near Space Labs, founded in 2017, works to improve the information that will feed AI assessments of property risks for insurers.
This article is excerpted from a longer interview and edited for clarity.
How does Near Space Labs' data and information improve on what AI can do for insurers?
We're removing the trade-offs that existed in the earth imaging space with satellite, airplane and drone imagery. Satellites show global coverage. Airplanes and drones fly very low, being able to see detail, but they don't scale.
We've seen so many headlines where people would be denied coverage using imagery coming from a satellite or an airplane that was old. We're able to provide the resolution that you would anticipate from a drone or an airplane, at the frequency and scale of a satellite. We're combining these two worlds in one single platform and doing it cost effectively, too.
We can capture data frequently, so when an insurer is looking at an image to underwrite or make parcel-level decisions, the data is not two years old. It's a quarter old or two quarters old, depending on the state and the risk level. Consumers will know they are being underwritten based on reality and not historical, inaccurate information. Instead of spending weeks or months on claims, insurers can process them very quickly, within days, and provide payouts. We unlock fast processing of claims and fast payouts.
What impact does this have for insurers?
They're able to confidently underwrite in places and in states where the risk is high, like California and Colorado, using granular recent imagery that represents reality. We can step in and support insurers on secondary or smaller perils, which consistently drive the biggest chunks of losses.
We're unlocking new business for insurance companies. The data coming from airplane-based solutions is often based on densely populated areas, where there are many policies to underwrite. Because of our cost-effective approach, we can cover smaller towns and rural areas and help insurers underwrite in those areas effectively, instead of just flying over the top 30 metropolitan areas, which is what happens with the status quo.
What insights does your technology yield that nothing else does?
Anything that requires frequent assessment or seasonal assessments. For example, with wildfire risk, vegetation-related attributes are huge in California. Tree overhangs, the health of trees — those are very, very important. A variety of different trees in your backyard can either be a protective layer against wildfires, or they can actually be increasing the risk of wildfire for your home.
We can effectively assess vegetation for its health in every season, multiple times a year, which airplanes alone cannot do. The Los Angeles wildfires are a prime example. So many housing units were built in woody areas and wildland-urban interfaces. There hasn't been any scalable, repeatable way to assess the health of the vegetation around those very expensive properties, so that's a very easy and very important example where Near Space is uniquely positioned to help insurance companies.
Designating a swimming pool
Look, AI is going to get there. I don't think anybody doubts that, but today visual data and AI don't really play that well. There's a lot of push around what comes after large language model. It's really an interesting wild wild west in terms of physical world models and bringing visual information into AI. That will really unlock some exciting performances, but we probably cannot get there without an AI step function workflow to perform tasks with multiple steps.
