The Story of Tensorflight: Building the Future of AI-Powered Property Insurance
In 2016, when drones were the next big thing in tech, a team of engineers saw an opportunity to apply computer vision and machine learning to analyze aerial imagery. Today, that company – Tensorflight – works with three of the five largest commercial property insurers in the US. But the path between these two points reveals how sometimes the best companies emerge from following the technology rather than the initial vision.
From Drones to Data at Scale
In a recent episode of Category Visionaries, Tensorflight co-founder Robert Kozikowski shared how the company’s journey began amid the drone industry boom. “We started by analyzing drone imagery… our first customer was QB, but on the side of drone analytics. Right. So our kind of series, a lead investor and customer QB, we started with drone analytics of some agricultural orchards, different farms, we providing metrics using machine learning, computer vision.”
However, the team quickly discovered a fundamental challenge with their drone-based approach. As Robert explains, “If you apply AI, you’re kind of making money on scale. And the problem with drones is that there were just too few drone flights, too much challenges with drone flights to really make it that computer vision startups would succeed.”
Finding the Right Market for the Technology
Rather than abandoning their core technology, Tensorflight pivoted to where their computer vision expertise could have the greatest impact. “We moved to satellites and also working with insurance, a few other industries. And over time from a few industries we’re working within satellite analytics, we kind of cut down to just insurance.”
This transition wasn’t just about changing data sources – it was about finding the right problem to solve. The insurance industry had been struggling with property assessment challenges for years, but the complexity of buildings and underwriting made it difficult to solve at scale.
Building Trust Through Deep Industry Focus
Unlike many startups that try to serve multiple verticals, Tensorflight doubled down on the insurance industry. Their approach to product development reflected this focus, with Robert noting, “We’re joking that our product management department is our customer underwriter… we kind of were, because what we’re solving is a problem that existed in the industry for years.”
This dedication to solving industry-specific problems helped them differentiate from competitors. As Robert explains, “We differ within this category by firstly, we focus more on commercial properties than just residential. We also focus on whole structure, not just roof… we’re also more global.”
The Future of Property Insurance
Looking ahead, Tensorflight’s vision extends beyond just providing building attributes. Robert outlines their ambitious plans: “In case of medium term roadmap, we think in two areas, kind of T shaped focus, I would say, and one is just going deeper into insurance and from just providing just building attributes, we provide additional risk scores, additional more in depth analytics than building attributes.”
The ultimate goal? “The long term vision in that area is just fully automated insurance pricing… That’s kind of dream for many insurance companies that you can get accurately priced insurance without manual tweaking, without judging anything by feeling of underwriters.”
Beyond insurance, the company sees opportunities to become “the most detailed building information database on the market with up to date accurate building information” serving industries from real estate to climate risk assessment.
Tensorflight’s journey from drone analytics startup to enterprise insurtech provider demonstrates how finding the right application for advanced technology often matters more than the initial vision. By focusing on where their computer vision expertise could solve real industry problems at scale, they’ve positioned themselves to transform how the insurance industry evaluates and prices property risk.