Amica Mutual Insurance Company lost a lot of money in the 2017 Tubbs Fire. So the company did what any rational institution does after a catastrophe: it examined the models that had told it those houses were safe. CoreLogic, the risk analytics firm Amica had been using, had flagged some of the destroyed properties as low risk. Amica switched to ZestyAI's Z-FIRE model, trained on more than 1,400 wildfire events across two decades. Amica ran the numbers backward: if Z-FIRE had been used to re-underwrite Amica's entire California portfolio in 2020, it would have prevented 95% of the insurer's wildfire losses in the state. Ninety-five percent.
That's a staggering claim for a scoring model. It's also the kind of claim that should make every homeowner in a fire-prone area deeply uncomfortable, because while the AI can now identify exactly which properties will burn, the building codes that could make them survive haven't caught up.
The machines see everything
ZestyAI's Z-FIRE is now trusted by more than a third of California's insurance market, including the CA FAIR Plan, the state's insurer of last resort and the place you end up when nobody else will write you a policy. Z-FIRE analyzes topography, historical climate data, and high-resolution aerial and satellite imagery to assess building materials, defensible space clearance, and vegetation density across multiple buffer zones around each structure. It produces a property-level risk score, not a ZIP-code average, not a county-wide estimate, but a reading of your house, your roof, your overgrown juniper.
Z-FIRE isn't alone in this space. Nearmap uses AI trained on aerial imagery to identify roof condition, vegetation encroachment, and even trampolines and solar panels that alter risk profiles. Property Guardian, which partnered with EigenRisk in June 2026, goes further with a time-aware Wildfire Recurrence Risk Score that accounts for how recently an area burned and how fuel conditions have evolved since. During the January 2025 Los Angeles wildfires, Property Guardian's scores identified the areas of highest destruction two to three days before the fire arrived.
Then there's Stand Insurance, a startup that builds aerospace-grade digital twins of individual homes and parcels. It simulates wildfire physics across your specific terrain (wind speed, ember flow, radiant heat) and then modifies the digital twin to show you what happens if you swap your cedar shake roof for Class A clay tile or clear the brush on your south slope. Dan Preston, Stand's CEO, puts it bluntly: "The last 10 years of data about wildfire risk is telling us very little about the next 10 years because it's changing very rapidly."
These aren't experimental tools. They're in rate filings. Farmers Insurance used Z-FIRE to win California regulatory approval for AI-based underwriting, opening standard coverage to an estimated 30,000 homeowners who would otherwise have been declined. Kingstone Companies announced in April 2026 that it would enter the California homeowners market on an excess and surplus lines basis, using Z-FIRE for property-level rating and underwriting. Commercially, the case is clear. AI has moved from research lab to rate filing.
The codes see almost nothing
California's Wildland-Urban Interface Code, Chapter 7A (now Part 7 of Title 24 as of January 1, 2026), is the most comprehensive wildfire building standard in the country. It specifies ember-resistant vents, fire-rated wallboard, tempered windows, Class A roofing, noncombustible Zone 0 landscaping within five feet of the structure. It works, and the research backing it is extensive: IBHS full-scale burn tests show homes built to standard surviving wildfire conditions that destroy conventionally built structures within 30 minutes. Thirty minutes.
But Chapter 7A only applies in areas the state has mapped as "high" and "very high" fire hazard severity zones, and in January 2025, when wildfires destroyed more than 16,000 structures across Los Angeles County, only 60% of the roughly 11,500 single-family homes lost fell within zones that would require Chapter 7A compliance if rebuilt. Another 4,600 homes, roughly 40%, had no wildfire building code requirement at all.
That gap is both technical and devastating in its consequences, because California's hazard maps calculate the probability of fire spreading from wildland areas into developed neighborhoods but do not account for what happens when fire spreads from house to house, the urban conflagration dynamic that actually leveled Altadena. As Los Angeles County Deputy Fire Chief Albert Yanagisawa acknowledged, the very high fire severity zones "haven't reached deep, deep down into Altadena." The AI knew those homes were at risk. The code didn't. Nobody fixed them.
Outside California, the picture gets considerably worse. Colorado's Senate Bill 23-166 required all local governments to implement wildfire resiliency regulations by April 2026, but adoption is fragmented: Lakewood's code took effect July 1, 2026; Loveland's on April 21. Oregon's Section R327 wildfire hazard mitigation is municipal opt-in, meaning each city decides for itself, and seven states enacted some form of wildfire building legislation in 2025. Colorado, Montana, New Hampshire, and Texas passed laws, while Maine, Hawaii, and Maryland created study commissions. That's seven out of fifty. Fifty. And three of them only agreed to study the question.
The math nobody's required to do
A joint study by Headwaters Economics and IBHS, published in fall 2025, examined the actual cost of wildfire-resistant construction for a representative 1,750-square-foot single-family home in Southern California with a baseline construction cost of $500,000. Its findings undermine every argument that fire-resistant building is prohibitively expensive. It isn't.
Compliance with California's CWUIC Part 7 adds approximately $13,070 to construction costs, or 2.6% of the total. IBHS's Wildfire Prepared Home Base standard, which addresses the most common ignition pathway (windblown embers), adds $8,876, just 1.8%, and actually saves $4,194 compared to the California code because it doesn't require gutter guards, allows open eave configurations, and doesn't mandate tempered windows. Its enhanced Plus designation, which addresses both embers and radiant heat exposure, adds $15,242, bringing the total premium to 3.0%.
To restate: building a home that can survive a wildfire costs between $9,000 and $15,000 more than building one that can't. In practice, it means fiber cement siding instead of wood lap, clay or concrete tile instead of asphalt composition, ember-resistant vents instead of standard mesh, tempered glass, metal gutters, and gravel instead of bark mulch within five feet of the foundation. A Headwaters Economics price check against current Home Depot inventory found that actual material costs run $2,418 below the RSMeans national average the study assumed.
Now compare that to what's happening on the insurance side of the equation. Homeowners in California's wildfire-exposed areas are paying $10,000 to $20,000 or more annually for coverage, when they can get it at all. IBHS's Wildfire Prepared Home designation is now linked to California Insurance Commissioner Ricardo Lara's "Safer from Wildfires" framework, which requires participating insurers to offer premium discounts for verified mitigation. If hardening reduces your annual premium by $2,000 to $5,000, the $9,000 to $15,000 construction investment pays for itself in three to five years. On a 30-year mortgage, that's $45,000 to $135,000 in savings before accounting for the part where your house doesn't burn down.
Rebuilding all 11,500 homes destroyed in the LA fires to Chapter 7A standards would cost approximately $345 million, while total insured losses from the January 2025 fires exceeded $30 billion. That ratio is roughly 1:87, meaning that for every dollar spent on wildfire-resistant construction, the insurance industry and homeowners would have avoided $87 in losses.
The asymmetry that matters
Both sides of the equation solved themselves. Building codes are the holdout. AI scoring is here, deployed, and working at property-level granularity. Construction solutions are here, proven in full-scale burn tests, costing a fraction of what people assume. What's broken is the regulatory layer between them, the building codes and hazard mapping apparatus, which operates on a fundamentally different geography than either one.
AI wildfire risk models evaluate your home as an individual physical object in a specific landscape, reading your cedar fence connecting your deck to your neighbor's wood siding, while building codes evaluate your home as a point inside a mapped zone derived from wildland fire behavior models that were never designed to capture urban conflagration. Code sees a zone boundary that stops three blocks short. Nobody mapped the gap.
IBHS expanded its Wildfire Prepared program in mid-2026 with new neighborhood and multifamily standards, precisely because single-home hardening fails when your neighbor's wood-sided house 15 feet away ignites and radiates heat across the gap. The program is voluntary, though, and Commissioner Lara's insurance incentive framework is the closest thing to a market-based mandate. It only works for homeowners who can still get policies from participating insurers. All 190,196 households on the FAIR Plan are out of luck.
IBHS is now available in 14 states, which is fourteen out of fifty, and "available" means you can pursue the designation voluntarily and hope your insurer recognizes it, not that your building department requires it.
What it means for anyone building
If you're building or buying in any of the 14 states where the IBHS Wildfire Prepared designation is available (Arizona, California, Colorado, Florida, Idaho, Montana, Nevada, New Mexico, Oklahoma, Oregon, Texas, Utah, Washington, Wyoming), pursue the designation whether the code requires it or not. Construction premiums run 2–3%, and the insurance implications are enormous and growing.
If you're a builder working in Colorado, confirm which of your jurisdictions has adopted wildfire resiliency codes under SB 23-166, because compliance timelines vary by municipality, and getting caught building to last year's code in a jurisdiction that updated in April is an expensive mistake that no amount of AI scoring can fix after the fact.
If you're rebuilding in LA County, know that 40% of the fire zone had no Chapter 7A requirement before the fires. Whether the county expands its hazard mapping to close that gap is an open question. Don't wait for the answer. Build to survive. Materials cost the same whether the code requires them or not. Same price. Same shingles.
And understand this: the AI scoring your risk doesn't care whether your local building code caught up, and ZestyAI, Property Guardian, Nearmap, Stand, and their competitors are analyzing your property from satellite and aircraft imagery right now, cataloging your roof material, your defensible space clearance, and the fuel load on your uphill slope with a precision that no building inspector has ever matched. When your insurer reprices your policy or declines to renew it, the decision will be based on what the AI sees, not on what the code required when you built.
Limitations
This analysis relies primarily on the IBHS/Headwaters Economics cost study, which models a single representative home type in Southern California. Costs for two-story homes, larger footprints, complex rooflines, and homes in regions with different baseline material preferences (wood siding dominant in the Pacific Northwest, brick in parts of the South) will differ. That 2–3% construction premium applies to new builds; retrofit costs for existing homes are substantially higher. Headwaters Economics estimates $18,000 to $27,000 for the most stringent "Optimum" standard in Southern California, and post-disaster labor markets inflate all costs further. Insurance premium savings estimates ($2,000–$5,000 annually) are inferred from market reporting on CA WUI zone premiums and Lara's mitigation discount framework; actual savings will vary by carrier, location, and policy terms. We were unable to obtain specific premium reduction percentages from any insurer tied to IBHS Wildfire Prepared status.