Resources

What is the 2026 wildfire outlook for insurers? 

The 2026 wildfire season is forcing carriers to reassess wildfire exposure at the property level. ZestyAI’s analysis of 2025 fire perimeters shows that structural loss, risk concentration, and emerging drought patterns are shifting the underwriting conversation beyond traditional wildfire maps. For underwriters, actuaries, and product leaders, the priority is clear: identify which properties are most likely to be reached, damaged, or destroyed before the next fire season. 

For the full property-level analysis, state risk profiles, 2025 case studies, and regulatory overview, download the complete Wildfire Season Preview 2026. 

Key Findings

1. The 2025 wildfire season burned 5.1 million acres — below the five-year average — but destroyed 18,385 structures, more than 2.5 times the average, with nearly 90% of structural loss driven by the January Los Angeles fires.
2. Properties in ZestyAI's highest wildfire risk tier were roughly 500 times more likely to fall within a wildfire perimeter in 2025 than properties in the lowest tier.
3. Across three 2025 case studies in Oregon, Utah, and North Carolina, properties inside fire perimeters were concentrated in High or Very High risk classifications at 11 to 15 times the statewide average.
4. In 2025, California accounted for 92% of all U.S. residential properties that fell inside wildfire perimeters, even though only about 11% of its properties score High Risk.
5. Colorado's HB25-1182, effective July 1, 2026, requires insurers using wildfire models to explain scores, account for mitigation, and resolve policyholder appeals within 30 days.
6. The 2026 drought outlook shows improving conditions in California but extreme and exceptional drought developing across the central Rockies and parts of the U.S. South — expanding the wildfire risk footprint beyond traditional Western exposure zones.

Why 2025 Changed the Wildfire Risk Conversation for Carriers 

The 2025 wildfire season exposed how quickly insured loss can concentrate when fire reaches dense residential areas. Nationwide, 18,385 structures were destroyed, more than 2.5 times the 2021–2025 average. Nearly 90% of that structural loss came from the January Los Angeles fires, where the Palisades and Eaton fires destroyed nearly 16,000 structures.

Those losses showed how wildfire exposure is changing for P&C carriers. Ember-driven fire can move through dense neighborhoods far from the traditional wildland-urban interface, creating severe property loss in areas that may not look extreme through conventional wildfire maps.

For underwriting, actuarial, and product teams, the implication is direct. Wildfire risk has to be evaluated at the property level, using factors such as defensible space, roof materials, vegetation, structure characteristics, surrounding fuels, topography, and local fire behavior. Two homes in the same neighborhood can carry materially different risk based on the conditions around the individual property.

Carriers preparing for the 2026 wildfire season need visibility into which properties are most likely to be reached by fire, which are most likely to be damaged or destroyed, and where risk is concentrating across the portfolio before the next major event occurs.

ZestyAI's Wildfire Season Preview 2026 draws on property-level analysis of every residential property that fell within 2025 wildfire perimeters, three regional case studies validating model performance across distinct fire environments, and an overview of the emerging regulatory landscape reshaping how insurers assess and communicate wildfire risk.

How Wildfire Exposure Varies Across U.S. Properties

ZestyAI analyzed every residential property across the U.S. using Z-FIRE, then analyzed which properties fell within 2025 wildfire perimeters to understand where exposure actually occurred. The results show a highly uneven risk landscape, with significant variation across states, regions, and individual properties.

Nationally, roughly 91% of properties score Low Risk, 6% score Medium Risk, and 3% score High Risk under ZestyAI's Z-FIRE model. But in 2025, properties in the High Risk tier were far more likely to fall within wildfire perimeters than those in the Low Risk tier — by a factor of approximately 500. 

That separation matters for carriers because wildfire exposure is not evenly distributed across a book of business. A small share of properties can represent a disproportionate share of exposure when fire activity overlaps with dense development, vulnerable property conditions, and elevated local fire potential. 

State-level exposure adds another layer of complexity. The share of High Risk properties varies significantly by state. Montana, for instance, carries a much larger proportion of High Risk properties than the national average. But high risk-tier concentration does not alone explain where losses occur — actual exposure also depends on where fires ignite, how they spread, and how many properties sit in their path.

Three states illustrate how differently the same risk tiers translate into actual exposure:

California had only about 11% of properties in the High Risk tier in 2025, yet accounted for 92% of all U.S. properties that fell inside wildfire perimeters. The scale of fire activity, combined with dense development near wildlands and major fires moving through populated neighborhoods, turned a limited High Risk segment into a dominant share of national exposure.

Texas is weighted heavily toward the lower end of the risk spectrum, with approximately 94% of properties classified Low Risk and 2% High Risk. Fire incidence within risk tiers reflects that lower baseline exposure.

Colorado shows a more distributed profile — about 70% Low Risk, 17% Medium Risk, and 12% High Risk — with fire incidence rates that track its more balanced risk distribution.

These state profiles reinforce a core finding: wildfire exposure cannot be assessed through a single national average or a fixed risk threshold. Local fire behavior, development patterns, and property density all determine how risk translates into actual loss.

Where Wildfire Risk Is Shifting in 2026 

The 2026 wildfire outlook points to a broader and more complex risk footprint for P&C carriers.

Drought conditions are improving in parts of California, but elevated dryness is developing across the central Rockies and parts of the U.S. South. That shift matters because drought can dry fuels, stress vegetation, and make fire behavior more responsive to wind, heat, and ignition sources.

For carriers, the underwriting concern is not limited to where wildfire has historically been most severe. It is where changing fuel conditions, property density, and local weather patterns are creating new pockets of exposure.

The 2025 season already showed signs of this expansion. Major wildfire events occurred in Arizona and Oklahoma, and the Black Cove Fire in North Carolina demonstrated how quickly wildfire can become relevant in markets where it has not traditionally been treated as a primary underwriting peril. That fire was driven by dry conditions, wind, and heavy storm-damaged fuels left by Hurricane Helene.

The 2026 outlook reinforces the need for carriers to reassess wildfire exposure beyond legacy high-risk geographies. Underwriting, actuarial, and product teams should evaluate where wildfire risk is rising across the portfolio, which properties are most exposed, and whether existing pricing, eligibility, and renewal strategies reflect the current risk environment.

A static view of wildfire territory is no longer enough. Carriers need a property-level view of exposure that can adapt as drought conditions, vegetation, development patterns, and fire activity shift across regions.

How Z-FIRE Performed Across 2025 Wildfire Events 

ZestyAI evaluated Z-FIRE performance across three 2025 fires representing distinct regional settings and fire environments: a Pacific Northwest WUI fire, a Mountain West fire in rugged lower-density terrain, and a Southeast fire in a region where wildfire has not historically been a primary underwriting concern. 

Across all three events, properties inside fire perimeters were concentrated in High or Very High classifications at 11 to 15 times the statewide average. That consistency matters for carriers validating wildfire models across different geographies, fire environments, and portfolio segments. 

Flat Fire, Oregon (August 2025)

The Flat Fire was a Central Oregon wildland-urban interface fire that grew to more than 23,000 acres in under a week. Of the 107 Z-FIRE-scored properties that fell inside the perimeter, 95% had been classified as High or Very High before the fire. None were Low or Very Low.

Statewide, roughly 8 in 100 Oregon properties carry a High or Very High wildfire risk score. Inside the Flat Fire perimeter, more than 95 in 100 did — approximately 11 times the statewide average. For carriers with Central Oregon exposure, the implication is direct: the fire did not reach a random cross-section of properties. It reached a concentrated group that had already been identified as materially more exposed.

Monroe Canyon Fire, Utah (July–September 2025)

The Monroe Canyon Fire burned through rugged, lower-density terrain in south-central Utah, growing to more than 73,000 acres over 54 days. Every scored property inside the perimeter had been classified as High or Very High before the fire. 80% were Very High; 20% were High. None were Medium, Low, or Very Low.

Statewide, only 9% of Utah properties are classified as High or Very High. Inside the Monroe Canyon perimeter, that share was 100%. The Very High concentration was even sharper: only 2% of Utah properties fall in the Very High classification statewide, compared with 80% inside the perimeter.

Monroe Canyon illustrates a specific underwriting challenge: in lower-density Mountain West terrain, the affected population may be small — cabins, second homes, rural recreational properties — but the properties that fall inside a large fire perimeter are not random. They are the ones already scored highest. Geographic and ZIP-code views can show where exposure is located; property-level scores show which specific properties carry the most risk.

Black Cove Fire, North Carolina (March 2025)

The Black Cove Fire started in Polk County, North Carolina after a downed power line ignited dry vegetation in the Green River Gorge. Hurricane Helene had deposited heavy storm-damaged fuels in the area six months earlier. Drought and wind helped the fire become one of the most active early-season events in the eastern United States in 2025.

Inside the Black Cove perimeter, 84% of properties had been classified as High or Very High before the fire arrived — 64% High and 19% Very High. None were Low or Very Low. Statewide, only 5% of North Carolina properties are classified as High or Very High. Inside the perimeter, that share was 84%, more than 15 times the statewide average.

Black Cove is the most important case study for carriers reconsidering their exposure in non-traditional wildfire markets. The fire occurred in the Southeast, in a state where wildfire has not historically been treated as a primary underwriting peril. The signal was there before the fire. Carriers that treat wildfire as a Western-only concern are operating without visibility into a growing share of their exposure.

How Wildfire Modeling Regulations Are Changing in 2026 

Wildfire regulation is moving from model permission to model accountability. Carriers increasingly need to explain how wildfire scores are used, recognize mitigation, support policyholder appeals, and document model governance. 

Across the West, four states are pursuing that standard through different paths.

California has tied wildfire model adoption to coverage obligations through its Sustainable Insurance Strategy. Carriers using approved wildfire catastrophe models must write a proportional share of homes in high-risk areas relative to their statewide market presence. A carrier with 10% of the California market must write at least 8.5% of homes in high-risk zones. Meeting that obligation requires property-level risk assessment — not just catastrophe model outputs — to distinguish individual homes worth writing from those that are not.

Colorado's HB25-1182, enacted in 2025 and effective July 1, 2026, requires insurers that use wildfire risk models to explain how models are used, account for property-specific and community-level mitigation, provide annual notices about wildfire risk scores and available discounts, and create an appeal process for disputed scores. Appeals must be acknowledged within 10 days and resolved within 30 days. The law moves wildfire modeling from a permissioning question to an accountability standard.

Washington's Senate Bill 5928 would require insurers to disclose wildfire risk scores when they are used in coverage or pricing decisions, explain the factors driving the score, and provide plain-language steps homeowners can take to improve it. Wildfire scores can no longer operate as a black box.

Oregon repealed its statewide wildfire hazard map in 2025 after public pushback against classifications that were too broad to reflect property-specific characteristics. The state is now moving toward a more granular, property-level approach. For insurers, the Oregon experience illustrates the risk of relying on broad geographic classifications: when scores cannot distinguish vulnerable homes from better-protected ones, or when policyholders cannot understand what drives their rating, the regulatory and consumer response can be swift.

Beyond state-level rules, the NAIC Model Bulletin on the Use of Artificial Intelligence Systems by Insurers, adopted in December 2023, establishes national governance expectations for AI-based underwriting tools. As more states adopt the bulletin's guidance, wildfire models will increasingly be evaluated on how they are governed, tested, explained, and monitored — not only on predictive performance.

What Carriers Should Prioritize for Wildfire Underwriting in 2026 

Wildfire underwriting in 2026 requires more than identifying high-risk geographies. Carriers need a sharper operating model for how wildfire risk is segmented, monitored, validated, and explained across the policy lifecycle. 

1. Segment wildfire risk at the property level

Broad geographic proxies can miss the variation that matters most for underwriting and pricing. ZIP codes, static wildfire maps, and community-level classifications may show where exposure exists, but they cannot reliably distinguish the individual properties most likely to be reached, damaged, or destroyed.

Property-level risk segmentation gives underwriting, actuarial, and product teams a more precise view of exposure. It can support new business eligibility, renewal decisions, pricing refinement, mitigation targeting, portfolio concentration management, and reinsurance planning.

The 2025 wildfire season reinforced the need for that precision. Properties inside fire perimeters were not a random cross-section of the market. They were disproportionately concentrated in higher-risk classifications before the fires occurred.

2. Monitor mitigation and property conditions continuously

Wildfire risk changes between policy cycles. Defensible space can be cleared or allowed to regrow. Roof materials can be replaced or deteriorate. Vegetation conditions shift with drought, storms, land use, and seasonal growth.

For carriers, mitigation recognition cannot be a one-time underwriting event. It requires current data on the property and its surroundings, along with the ability to explain how mitigation affects eligibility, pricing, or renewal decisions.

This is becoming both an underwriting priority and a regulatory expectation. As states move toward greater transparency around wildfire scores, carriers need defensible ways to identify mitigation, reflect it in risk decisions, and communicate the impact to policyholders.

3. Validate wildfire models across multiple fire environments

Wildfire exposure is no longer confined to one familiar pattern. Carriers may face risk in dense California neighborhoods, Western wildland-urban interface zones, rugged Mountain West terrain, rural recreational areas, and emerging non-traditional markets in the Southeast.

Model validation should reflect that diversity. A wildfire model that performs well in one region or fire environment may not provide the same signal across an entire book of business.

Carriers should require evidence that wildfire models separate risk across the geographies and property types represented in their portfolios. That means evaluating performance against observed fire outcomes, not only relying on theoretical risk assumptions or broad hazard classifications.

4. Prepare for explainability, governance, and appeals 

Wildfire modeling is moving into a more accountable phase. Carriers increasingly need to explain how scores are used, identify the factors driving risk, recognize property-level and community-level mitigation, and support policyholder appeals.

This changes what underwriting and actuarial teams need from wildfire models. Predictive performance still matters, but it is not enough on its own. Carriers also need transparency, documentation, governance, and operational workflows that can stand up to regulatory and policyholder scrutiny.

For 2026, the strongest wildfire strategies will combine accurate risk segmentation with explainable decision-making. Carriers that can see risk clearly, act on it consistently, and explain it credibly will be better positioned for underwriting discipline, regulatory readiness, and profitable growth.

Frequently Asked Questions

What is the 2026 wildfire outlook for P&C insurers?

The 2026 wildfire outlook points to a broader and more complex risk environment for P&C insurers. Drought conditions are improving in parts of California, while elevated dryness is developing across the central Rockies and parts of the U.S. South. For carriers, the key issue is not only where fires may occur, but which properties are most likely to be reached, damaged, or destroyed when wildfire moves into populated areas.

Why do insurers need property-level wildfire risk models?

Insurers need property-level wildfire risk models because wildfire exposure can vary sharply from one home to the next, even within the same neighborhood. Defensible space, roof materials, vegetation, topography, structure characteristics, surrounding fuels, and local fire behavior can all affect whether a property is reached, damaged, or destroyed. ZIP codes, static maps, and broad geographic classifications cannot capture that variation with enough precision for underwriting, pricing, renewal, and portfolio management.

How many structures were destroyed in the 2025 wildfire season?

18,385 structures were destroyed in the 2025 wildfire season — more than 2.5 times the 2021–2025 average. Nearly 90% of that structural loss came from the January 2025 Los Angeles fires (Palisades and Eaton), which together destroyed nearly 16,000 structures and generated an estimated $40 billion in insured losses. Total acres burned in 2025 came in below the five-year average, making the divergence between acreage and structural loss the defining characteristic of the season.

How can carriers use wildfire risk scores in underwriting and pricing?

Carriers can use wildfire risk scores to segment new business, evaluate renewals, refine pricing, identify mitigation opportunities, manage portfolio concentration, and support reinsurance planning. The most useful wildfire risk scores are property-specific, explainable, and validated against observed fire outcomes, so underwriting and actuarial teams can understand both the risk level and the drivers behind it.

How much does wildfire risk vary by property?

Wildfire risk varies significantly at the property level. In ZestyAI’s analysis of 2025 wildfire perimeters, properties in the highest wildfire risk tier were roughly 500 times more likely to fall within a wildfire perimeter than properties in the lowest tier. Across case studies in Oregon, Utah, and North Carolina, properties inside fire perimeters were concentrated in High or Very High risk classifications at 11 to 15 times the statewide average.

Is wildfire risk growing outside the Western United States?

Yes. Wildfire remains a major concern in the West, but 2025 showed that wildfire exposure is also relevant in non-traditional markets. The Black Cove Fire in North Carolina occurred in a state where wildfire has not historically been treated as a primary underwriting peril. Inside the Black Cove perimeter, 84% of properties had been classified as High or Very High Risk before the fire, more than 15 times the statewide average.

Which states face elevated wildfire exposure?

Wildfire exposure depends on both the share of properties in higher-risk tiers and the location of actual fire activity. California accounted for 92% of all U.S. residential properties that fell inside wildfire perimeters in 2025, even though only about 11% of its properties score High Risk. Colorado has a more distributed profile, with approximately 12% High Risk and 17% Medium Risk. Montana carries a higher share of High Risk properties relative to the national average.

What does Colorado HB25-1182 require from insurers?

Colorado HB25-1182, effective July 1, 2026, applies to insurers that use wildfire risk models, catastrophe models, or scoring methods to assign property risk. The law requires insurers to explain how models are used, account for property-specific and community-level mitigation, provide annual notices about wildfire risk scores and mitigation discounts, and establish an appeal process for disputed scores or classifications. Appeals must be acknowledged within 10 days and resolved within 30 days.

How are wildfire modeling regulations changing for insurers?

Wildfire regulation is moving toward greater transparency, mitigation recognition, model governance, and consumer explainability. Insurers increasingly need to show how wildfire scores are used in underwriting and pricing, which factors influence those scores, how mitigation is reflected, and how policyholders can dispute or improve their classification. This shifts wildfire modeling from a purely analytical capability to an operational and regulatory requirement.

What is Z-FIRE?

Z-FIRE is ZestyAI’s property-level wildfire risk model. It produces scores designed to estimate both the likelihood that a property will fall within a wildfire perimeter and the likelihood that the property will be damaged or destroyed if exposed. The model incorporates property and environmental factors such as defensible space, roof materials, structure characteristics, vegetation, topography, and surrounding fuels.

How did Z-FIRE perform in 2025 wildfire events?

ZestyAI evaluated Z-FIRE across three 2025 wildfire events in Oregon, Utah, and North Carolina. In each case, properties inside the fire perimeter were heavily concentrated in High or Very High risk classifications before the fire occurred. Across the three case studies, that concentration was 11 to 15 times the statewide average, showing strong risk separation across different fire environments.

How many structures were destroyed in the 2025 wildfire season?

The 2025 wildfire season destroyed 18,385 structures nationwide, more than 2.5 times the 2021–2025 average. Nearly 90% of that structural loss came from the January Los Angeles fires, where the Palisades and Eaton fires destroyed nearly 16,000 structures.

What should carriers do to prepare for wildfire risk in 2026?

Carriers should reassess wildfire exposure at the property level, monitor mitigation and property conditions continuously, validate wildfire models across multiple fire environments, and prepare for greater explainability, governance, and appeals requirements. The goal is to understand which properties carry the highest wildfire exposure before the next major event occurs, then use that insight to improve underwriting, pricing, renewal, and portfolio decisions.

For the full property-level analysis, state risk profiles, case study data, and regulatory overview, download the complete Wildfire Season Preview 2026.

Engineered for the Scale Carriers Demand

Holding sub-second response times at this scale, across billions of data points spanning property, imagery, and risk sources, requires infrastructure purpose-built for speed and reliability.

For our clients, faster response times aren't a nice-to-have; they're necessary to support the workflows their teams rely on every day. And as underwriting becomes more automated, carriers need data products that can keep pace with real-time decisioning.

Sub-second response times are the latest step in a multi-year effort to push the boundaries of what carriers can expect from a property intelligence API. In late 2024, we redesigned our ML inference platform, halving API response times. A year later, we cut latency in half again, bringing the typical call below one second. 

A Smoother Customer Journey

Faster responses also improve the experience at the point of decision. When property intelligence returns in under a second, quotes can move forward without making underwriters, agents, or policyholders wait on data behind the scenes.

Policyholders now bring Amazon-shaped expectations to every digital purchase: answers in seconds, not days, with zero tolerance for a spinning loader—and insurance is no exception. When property intelligence returns in under a second, carriers can deliver that experience without the data layer holding them back, and shoppers don’t drift to a competitor mid-quote. Customer experience also drives business performance: McKinsey found that P&C insurers ranking among customer-experience leaders outperformed peers in total shareholder return by 65 percentage points over a five-year period (McKinsey & Company, 2023).

This is the gap ZestyAI closes. Property intelligence that responds instantly helps keep the customer journey moving, giving carriers a faster, smoother path from quote to decision.

Workflow Integration Without the Friction

Sub-second responses also make it far easier to embed property intelligence right where your teams already work: your rating systems, underwriting workbenches, and internal dashboards.

As a property intelligence partner, ZestyAI delivers insights fast enough to support the workflows carriers rely on every day. For technical teams, that means a more reliable foundation to build on. For business users, it means the data is simply there when they need it, with no extra steps.

Faster Batch, Faster Books

Quicker responses also matter at the portfolio level. When each API call returns faster, large jobs like portfolio reviews, renewal runs, and book-level analysis can be completed more efficiently. That helps carriers refresh their view of risk more often and act on it sooner.

What Faster Response Times Mean for Carriers

Faster property intelligence pays off in two compounding ways:

• Higher workflow capacity: With each API call returning in under a second, teams can move from risk evaluation to action sooner and process more work with less delay.
• Higher conversion potential: Decisions that keep pace with the customer reduce the delay between quote and bind.

ZestyAI's recent infrastructure investments help carriers access property intelligence with the speed and scale their workflows require, while keeping the reliability carriers depend on. This combination of speed, scale, and reliability is why leading carriers use ZestyAI to support critical risk decisions.

Want to experience it firsthand? See how ZestyAI's risk models assess property-level risk across the major perils driving insured losses. Contact our team for a tailored demo built around your line of business.

GuardianPointe Insurance Company has selected ZestyAI's Roof Age and Z-PROPERTY™ solutions to bring greater accuracy and confidence to how property risk is understood across its personal and commercial lines portfolios.

For insurers, portfolio performance depends on having a complete and accurate view of exposure. Yet most portfolios are built on inconsistent or outdated property data, introducing uncertainty into underwriting, pricing, and portfolio management decisions. 

As a new entrant, GuardianPointe is establishing a data-driven foundation from the outset—prioritizing precise, property-level insight to support disciplined underwriting and portfolio management.

ZestyAI addresses this gap by delivering verified, property-level intelligence at scale. Its Roof Age solution uses building permit records, historical aerial imagery, and advanced AI to detect roof replacement events and assign accurate roof age. Z-PROPERTY extends this with a comprehensive view of each structure and parcel, including characteristics that materially impact loss. 

Together, these solutions give GuardianPointe a consistent, data-driven foundation for understanding the properties it insures and how risk accumulates across its portfolio.

Rick Espino, CEO of GuardianPointe Insurance Company, said:

“ZestyAI gives us a clearer, property-level understanding of the risks across our portfolio. That visibility helps reduce uncertainty and strengthens how we evaluate and manage exposure across our book.”

“The industry is shifting toward a more exact understanding of risk—grounded in what’s actually present at each property,” said Attila Toth, Founder and CEO of ZestyAI.

“GuardianPointe is building that clarity into how its portfolio is understood, giving it a stronger foundation for long-term performance.”

ZestyAI has secured more than 200 regulatory approvals nationwide, giving insurers a trusted, property-level foundation to underwrite, price, and manage risk with precision. 

ZestyAI today announced that Adaptive Insurance has integrated Z-STORM™ into its underwriting and pricing for its wind and hail programs.

Severe convective storms are now one of the most persistent drivers of insured loss in the U.S., with annual losses exceeding $50 billion for three consecutive years. As losses become more localized and volatile, traditional territory- and ZIP code-based models are increasingly misaligned with how risk actually behaves.

Z-STORM delivers property-level risk scores that differentiate storm risk across individual properties, even within the same neighborhood. Z-STORM is trained on verified carrier claims data and evaluates how local climatology interacts with the specific characteristics of each structure. 

Mike Gulla, CEO and Co-Founder of Adaptive Insurance, said:

"What sets ZestyAI apart is their ability to split risk at the individual property level. Integrating the precision that Z-STORM offers will allow us to price and underwrite more confidently and offer coverage that truly reflects how storm risk behaves."

Z-STORM predicts the expected frequency and severity of severe convective storm losses, including hail and wind, by combining climatology with detailed property-specific characteristics. The model delivers clear explanations of the factors behind each property's risk score, supporting more transparent and defensible rate-making. Key capabilities include:

• Pinpointing hail risk using property-level drivers, such as roof geometry, accumulated damage, and local climatology, to identify buildings most likely to sustain hail damage, even within the same geographic area.
• Assessing wind vulnerability using AI-driven analysis of roof condition, complexity, and potential failure points, combined with localized wind climatology, to determine which structures are most susceptible to wind damage.
• Predicting total storm losses by examining the interaction between climatology and the unique characteristics of every structure and roof to forecast claim frequency and severity.

"Adaptive Insurance joins a growing group of carriers bringing greater precision to storm risk underwriting," said Attila Toth, Founder and CEO of ZestyAI.

"With Z-STORM delivering a sharper, property-level view of hail and wind risk, Adaptive can optimize price and coverage even further to identify pockets of exposure earlier, before they translate into outsized losses."

The selection builds on ZestyAI's continued growth across the insurance market. Z-STORM has earned regulatory acceptance in 32 states, with more than 200 regulatory approvals across ZestyAI's full portfolio of AI-powered risk models nationwide. Adaptive Insurance joins a growing roster of carriers leveraging ZestyAI to strengthen underwriting and pricing decisions in an increasingly volatile climate environment.

ZestyAI, the Risk and Decision Intelligence Platform for the insurance industry, today announced that Windward Risk Managers is using Z-FIRE™ to power wildfire underwriting as it expands into the California insurance market.

Windward Risk Managers already utilizes Z-PROPERTY™ across its Florida homeowners portfolio. The platform applies computer vision and machine learning to analyze aerial imagery, building permit records, tax assessment data, and other verified sources, generating insights across more than 70 structural and parcel attributes for 150 million residential and commercial properties nationwide.

The expansion into California builds on that relationship, extending the insurer’s use of the ZestyAI platform to address wildfire risk in one of the country’s most complex catastrophe markets.

"Z-PROPERTY gave us accurate property insights and broad coverage across our Florida portfolio," said Gard Olbers, Chief Risk Officer at Windward Risk Managers.

"That experience made ZestyAI a natural partner as we expand into California, where Z-FIRE gives us an accurate, predictive view of wildfire risk as we evaluate and manage exposure."

In recent years, catastrophic wildfires have driven record losses and prompted several carriers to reduce their presence in the state. For insurers entering the market, accurately assessing wildfire exposure at the individual property level has become essential.

Z-FIRE predicts which properties are most likely to experience a wildfire and which ones will survive. The model applies computer vision and machine learning to analyze structural and environmental characteristics such as defensible space, vegetation density, building materials, and topography, and is trained on the industry’s largest wildfire loss dataset.

“Windward Risk Manager’s expansion into California reflects how insurers can use advanced analytics to operate more confidently in wildfire-exposed markets,” said Attila Toth, Founder and CEO of ZestyAI.

"Windward Risk Managers has been a valued partner for many years, and their continued expansion with ZestyAI reflects the trust insurers place in accurate risk models as they navigate increasingly complex markets."

Z-FIRE is approved across Western wildfire markets and was the first AI-based wildfire model approved as part of a carrier rate filing in California. ZestyAI’s broader portfolio of risk models has secured more than 200 regulatory approvals nationwide.

Columbia Lloyds Insurance Company, a Houston-based regional insurer serving homeowners across Texas, Oklahoma, and Arkansas, has selected ZestyAI's Risk and Decision Intelligence platform to sharpen underwriting accuracy, strengthen portfolio management, and improve visibility into property condition and risk exposure across its book of business.

Columbia Lloyds will use ZestyAI's Z-PROPERTY™ model to evaluate roof complexity, materials, condition, and surrounding risk factors that directly shape loss potential and underwriting outcomes. Paired with ZestyAI's Roof Age — which cross-validates building permit records against more than 20 years of aerial imagery to detect replacement events that traditional records miss and assigns a confidence score to every property — the carrier gains a verified, property-level view of exposure it can apply consistently across its homeowners book.

"We write homeowners business in some of the toughest weather territory in the country," said Sam Bana, Chief Operating Officer of Columbia Lloyds.

"ZestyAI gives us the verified property data we need to make better decisions on every risk we write and across the portfolio as a whole."

"The carriers succeeding in the most weather-exposed markets today are the ones operating on verified, property-level data rather than assumptions," said Attila Toth, Founder and CEO of ZestyAI.

"Columbia Lloyds is taking a disciplined approach to how they underwrite, price, and manage exposure, and we're proud to support how they're shaping a more resilient book across Texas, Oklahoma, and Arkansas."

ZestyAI's models are built with transparency, validation, and regulatory readiness at the forefront, giving insurers confidence to rely on them in portfolio decisions. With more than 200 regulatory approvals secured nationwide, the platform is used by leading insurers to improve underwriting accuracy, manage exposure, and reduce loss volatility across weather- and non-weather-driven perils.