Satellite imagery, fire science, and machine learning - pointed at the one peril most carriers would rather not price.
It is fire season in California, and an insurance agent in the foothills is doing something that, a few years ago, would have been impossible: writing a homeowners policy on a house with chaparral creeping up the back slope. The carrier that covered it last year is gone. The state's insurer of last resort is overloaded. And yet the quote comes back - bindable, A-rated, real.
The company behind that quote is Delos Insurance Solutions. It is a managing general agent, which is the insurance industry's polite term for the people who decide who gets covered and at what price. Delos does that decision with satellites. It pulls NASA Earth-observation imagery, climate datasets, high-resolution wind models, and a few hundred other parameters, and asks a narrower question than the rest of the market bothers with: not "is this a fire zone?" but "is this house likely to burn?"
The answer, more often than the industry assumes, is no.
For decades, the insurance answer to wildfire was geographic. Draw a circle around the risky counties, raise the rates, or stop writing entirely. It is a tidy strategy if you are an actuary and a catastrophe if you own a house inside the circle. As California's fires grew, carriers retreated, non-renewals piled up, and hundreds of thousands of homeowners were pushed toward the FAIR Plan - more expensive, less generous, and never meant to carry that much weight.
Here is the inconvenient detail the retreat ignored. Wildfire risk is not evenly spread across a zip code. Two houses on the same street can carry wildly different odds depending on slope, vegetation, wind corridors, roofing, and defensible space. Treating them identically is simpler. It is also, in Delos's telling, just wrong - and the kind of wrong you can measure.
Delos likes to put its message to retreating insurers plainly: send the homes to us.
Kevin Stein and Shanna McIntyre did not come from insurance, which turned out to be the point. Kevin studied engineering at Berkeley and Stanford, then spent years designing spacecraft and selling satellites. Shanna grew up in the Mount Diablo foothills - fire country she would later model for a living - studied physics at Berkeley, and did predictive modeling at Lockheed Martin.
They noticed that a new generation of satellite imagery could model wildfire at far higher spatial and temporal resolution than the insurance industry was using. The bet was simple to state and hard to build: if you can measure wildfire risk precisely enough, the homes everyone calls "uninsurable" turn out to be a business, not a charity.
Berkeley and Stanford engineer who designed spacecraft and managed $50M+ in satellite sales before deciding insurance underwriting was the more interesting modeling problem.
Berkeley physicist and former Lockheed Martin predictive-modeling engineer who grew up in the Mount Diablo foothills - the exact terrain her models now grade.
What Delos sells looks ordinary - a California homeowners policy, backed by A-rated carriers and Lloyd's of London capacity, sold through roughly 11,000 distribution partners. What is unusual sits underneath it: a patented wildfire model that runs hundreds of parameters per property, blending NASA's LANDFIRE datasets, current climate science, and high-resolution wind data into a per-address risk score.
Customers also get something the FAIR Plan does not hand out - personalized home-hardening advice, the dynamic to-do list of clearing brush and upgrading vents that can move a house from "decline" to "covered." It is risk modeling that occasionally tells you how to lower your own premium, which is a strange and welcome thing for an insurance product to do.
Homeowners and dwelling-fire policies for wildfire-exposed California properties, placed through A-rated carriers and Lloyd's capacity.
Patented geospatial underwriting blending NASA imagery, LANDFIRE data, climate science, and wind models down to the parcel.
Personalized, dynamic recommendations that help homeowners reduce risk - and qualify for a policy.
Quoting and policy tools for ~11,000 California-based agents and brokers, running on Guidewire InsuranceNow.
Two aerospace engineers start modeling wildfire at satellite resolution for an industry still using broad zones.
Delos makes its public pitch to carriers shedding wildfire risk, positioning as the place declined homes can land.
IA Capital Group leads, with Blue Bear Capital, Gallatin Point, Generation Space, and Avanta Ventures.
Insurance Insider US names Delos MGA/MGU of the Year; CB Insights lists it among the Top 100 Global Fintech.
HSBC Asset Management leads an oversubscribed round to fund expansion and new products.
A refined wind-aware model opens coverage across five Southern California counties - an 11% jump.
A wildfire underwriter only earns trust by being right about fires. Delos says its model predicted the extent of every major California fire over the past seven years - then priced policies on that basis, with what its investors describe as market-leading loss ratios. The scale chart below is the receipt.
Delos's tagline is "Home insurance powered by science," which sounds like marketing until you notice the company means it almost literally. The mission is not to be generous with fire-zone homeowners. It is to be accurate about them - on the theory that accuracy, not retreat, is what keeps insurance available when the climate gets harder to underwrite.
CEO Kevin Stein frames it modestly: "We are committed to providing affordable and accessible insurance for as many Californians as possible." The ambition, though, is bigger than California. The Series A was raised explicitly to fund expansion into new states and new products, because wildfire is not the only peril a satellite can see coming.
Return to that agent in the foothills, policy in hand, the chaparral still creeping up the back slope. Nothing about the fire risk changed. The hill is as steep, the season as dry, the wind as willing. What changed is that someone finally measured the house instead of the map.
That is the whole bet, and it is not a sentimental one. As climate risk spreads, the insurers that survive will be the ones who can price precisely enough to keep saying yes. Delos started with the hardest peril, in the hardest state, and built the instrument first. If it works at scale, the house on the slope stays insurable - and so does the next one, in the next fire-prone place, wherever the satellites point next.
The market's old answer to wildfire was a circle on a map. Delos's answer is a measurement. One of those scales with a warming planet. The other already left.
How the wildfire model and agent portal work.
Find on YouTube →The CEO on satellites, fire science, and California's insurance crisis.
Find on YouTube →Insurance Journal on Delos's pitch to retreating carriers.
Read the feature →