Wildfire risk in the United States is no longer confined to the edges of forests or traditionally high-risk zones. New analysis using ZestyAI’s property-level wildfire models shows that millions of homes classified as low or no wildfire risk under federal assessments face elevated wildfire danger when evaluated at the property level.

This analysis was recently featured in Vox, which examined how wildfire behavior is evolving — and why broad, backward-looking risk maps are increasingly misaligned with how fires spread today.

👉 Read the full article on Vox → https://www.vox.com/climate/476932/california-wildfire-los-angeles-risk-ai-housing-climate

Wildfire risk is closer — and more granular — than most maps show

Many homes damaged or destroyed in the 2025 Los Angeles wildfires were still classified as “low risk” under federal wildfire assessments. ZestyAI’s property-level analysis provides a different perspective.

By evaluating individual structures — including vegetation proximity, defensible space, building characteristics, and neighborhood-level fire dynamics — ZestyAI identified more than 3,000 properties worth approximately $2.4 billion in areas impacted by the Palisades and Eaton fires that showed elevated wildfire risk despite being classified as low or no risk under FEMA’s census-level assessments.

Across California, the classification gap is even broader. Approximately 1.2 million properties, representing roughly $940 billion in residential property value, are designated as low or no wildfire risk under federal maps, despite AI-driven property-level models indicating elevated wildfire danger.

Why census-level wildfire maps fall short

Wildfires do not spread evenly across census tracts or counties. Ember-driven ignition, structure-to-structure spread, wind conditions, and localized vegetation patterns create uneven outcomes, where one home survives and the next is destroyed.

Federal wildfire assessments are designed to provide a baseline view of community-level risk. FEMA has noted that its National Risk Index is not intended to serve as a property-specific risk assessment. When risk is evaluated at the individual property level, meaningful differences emerge that aggregated maps are not designed to capture.

What more granular wildfire risk intelligence enables

More detailed wildfire risk data can support:

• Targeted mitigation efforts at the property and neighborhood level
• More informed rebuilding and land-use decisions
• Clearer, more defensible underwriting and portfolio strategies
• Improved dialogue between insurers, regulators, and communities

A shift in how wildfire risk is understood

Wildfire risk is evolving faster than the systems built to measure it. Homes are no longer just adjacent to wildfire hazards; they increasingly influence how fires ignite, spread, and intensify, even in dense urban environments.

Property-level risk intelligence does not remove hard decisions. But without it, those decisions are made using an incomplete picture of where wildfire risk truly exists.

Read the full Vox article here.

In property insurance, roof age is one of the most critical factors in assessing risk. Yet too often, carriers rely on self-reported or agent-supplied data that is incomplete or inaccurate.

ZestyAI’s recent analysis of 500,000+ properties revealed widespread discrepancies in reported roof age. The result? Mispriced policies, unexpected losses, and operational inefficiencies that impact the bottom line.

As climate volatility grows and reinsurance pressure intensifies, overlooking the true condition and age of a home’s largest, most exposed surface is a risk no carrier can afford.

What’s Inside

• Uncover the biggest myths and blind spots in roof age records.
• Understand why traditional data sources, like claims systems and permits, fall short in providing accurate roof age.
• Learn how a multi-source verification strategy, combining aerial imagery, permits, tax records, and AI, offers a blueprint for improvement and 97% national coverage.
• Explore why roof age is a predictor of losses across multiple perils, not just wind and hail.
• Discover the one-two punch of verified roof age and real-time condition insights, delivering a complete view of risk, even for young roofs with hidden problems.
• Align your roof age data with growing regulatory expectations, particularly in states like Florida.

Access the Guide.

New research from ZestyAI reveals that 62% of U.S. homeowners are deferring critical home maintenance, adding up to $317 billion in potential claims exposure for insurers.

These findings come as Severe Convective Storms (SCS) caused an estimated $58 billion in insured losses in 2024, surpassing hurricane-related losses and marking the second-costliest SCS year on record.

Tornadoes, hail, and wind events now account for over 60% of all U.S. catastrophe claims, and research from the Insurance Institute for Business & Home Safety (IBHS) shows that roof damage accounts for up to 90% of residential catastrophe losses.

Key Findings from ZestyAI’s Homeowner Survey

According to ZestyAI’s nationally representative survey, 62% of homeowners have delayed essential repairs due to budget constraints, representing nearly 59 million U.S. homes with unaddressed vulnerabilities. Forty percent said they would rely on an insurance claim to cover major repairs like roof replacement, adding up to an estimated $317 billion in potential exposure for carriers.

Alarmingly, 63% of homeowners who weren’t living in their home at the time of the last roof replacement don’t know how old their roof is, making it even harder to detect aging systems before they fail. Meanwhile, 12% admitted they would delay repairs indefinitely, further increasing their risk of property damage.

Severe Convective Storms: The Growing Catastrophe Risk

This blind spot compounds known risks: prior ZestyAI analysis has identified over 12.6 million U.S. properties at high risk for hail-related roof damage, representing $189.5 billion in potential roof replacement costs.

“Deferred maintenance has long been a known risk factor, but today the stakes are higher than ever,” said Kumar Dhuvur, Co-Founder and Chief Product Officer of ZestyAI. "With claim severity rising and storm losses compounding, insurers need more than hazard maps to navigate this landscape."

"Property-level insights allow carriers to proactively address known vulnerabilities, improve underwriting precision, and work with homeowners to reduce losses before they happen.”

ZestyAI’s findings support a growing push toward data-driven, preventative underwriting strategies, especially as carriers face rising claim severity and pressure to improve combined ratios across storm-prone states.

With insured losses projected to exceed $30 billion, the recent Los Angeles wildfires rank among the costliest in U.S. history—reshaping how insurers think about risk, resilience, and readiness.

Watch the Full Webinar → LA Fires in Focus: What Insurers Need to Know

In this on-demand webinar, experts from the Insurance Institute for Business & Home Safety (IBHS), the Western Fire Chiefs Association, Cal Poly’s WUI Fire Institute, and ZestyAI unpack what really happened—from frontline response to lab-based research and model performance—and share critical strategies insurers can use to prepare for what’s next.

Watch this session if you’re a Product Managers, Underwriters, Actuaries, and Risk & Innovation leaders looking to make informed decisions in an increasingly volatile wildfire landscape.

What You’ll Learn

• Key takeaways from the Los Angeles wildfires
• Research on structure-to-structure fire spread and resilience factors
• How wildfire risk models performed—what we got right (and wrong)
• Practical strategies to reduce exposure and strengthen resilience

Meet the Experts

• Anne Cope, Chief Engineer, IBHS
• Bob Roper, CEO, Western Fire Chiefs Association
• Frank Frievalt, Director, WUI Fire Institute at Cal Poly
• Kumar Duhvur, Co-Founder & CPO, ZestyAI

With insured losses projected to exceed $30 billion, the recent Los Angeles wildfires rank among the costliest in U.S. history—reshaping how insurers think about risk, resilience, and readiness.

Watch the Full Webinar → LA Fires in Focus: What Insurers Need to Know

In this on-demand webinar, experts from the Insurance Institute for Business & Home Safety (IBHS), the Western Fire Chiefs Association, Cal Poly’s WUI Fire Institute, and ZestyAI unpack what really happened—from frontline response to lab-based research and model performance—and share critical strategies insurers can use to prepare for what’s next.

Watch this session if you’re a Product Managers, Underwriters, Actuaries, and Risk & Innovation leaders looking to make informed decisions in an increasingly volatile wildfire landscape.

What You’ll Learn

• Key takeaways from the Los Angeles wildfires
• Research on structure-to-structure fire spread and resilience factors
• How wildfire risk models performed—what we got right (and wrong)
• Practical strategies to reduce exposure and strengthen resilience

Meet the Experts

• Anne Cope, Chief Engineer, IBHS
• Bob Roper, CEO, Western Fire Chiefs Association
• Frank Frievalt, Director, WUI Fire Institute at Cal Poly
• Kumar Duhvur, Co-Founder & CPO, ZestyAI

Wildfire Risk Is Rising Nationwide‍

Wildfire seasons are getting longer, more destructive, and harder to predict—and they’re no longer just a Western U.S. concern. From the Southeast to the Midwest, wildfire risk is emerging in places many insurers haven’t traditionally watched.

What the Latest Data Reveals About Wildfire Exposure

Drawing from the latest national datasets and insights from ZestyAI’s Z-FIRE™ model, this visual guide to wildfire risk in the U.S. shows:

• New wildfire hotspots: Discover where risk is rising fastest.
• Mitigation gaps: Learn how a lack of defensible space is putting thousands of homes in danger across the country.
• Top risk drivers: See how features like overhanging trees and wooden roofs are fueling destruction in high-risk areas.

Download Free Infographic

BONUS: You’ll also get access to our latest online event with IBHS and Western Fire Chiefs Association, The LA Fires in Focus: What Worked, What Didn’t, What’s Next for Insurers.

For property insurers, roof age is more than just a data field — it’s a critical underwriting decision point that directly impacts pricing, risk selection, and loss costs. But what happens when that data is wrong two-thirds of the time?

A large U.S. carrier with over $500 million in direct written premium recently found out. Relying on self-reported and agent-estimated roof ages, they were systematically underpricing risky properties while overpricing safer ones. The result: adverse selection, elevated loss ratios, and underwriting decisions built on shaky foundations.

The Scale of the Problem: Two-Thirds of Roof Age Data Is Wrong

ZestyAI’s research shows that 67% of self-reported roof ages are inaccurate:

• 43% underestimate roof age — meaning roofs are older and riskier than reported
• 24% overestimate roof age — leading to overpricing or turning away good business

This isn’t just a pricing issue. Analysis also found that 78% of carriers in key U.S. regions use age-based triggers for ACV roof endorsements, with some starting as early as 8 years old. When roof age is wrong, policies can be misclassified across underwriting, eligibility, and coverage terms — creating compounding risk across the insurance lifecycle.

From Estimates to Evidence: How ZestyAI Verifies Roof Age

To replace guesswork with ground truth, the carrier deployed ZestyAI Roof Age, which analyzes building permits, more than 20 years of aerial imagery, and regional climatology using advanced machine learning. Each assessment is paired with a transparent confidence score.

Unlike traditional approaches that rely on policyholder memory or limited inspections, ZestyAI Roof Age:

• Anchors assessments in the property timeline to prevent false positives
• Cross-validates imagery with permits and climatological patterns
• Provides confidence scores to distinguish high-certainty predictions from cases requiring inspection
• Delivers explainable, auditable results that underwriters and actuaries can trust

The difference was immediate.

Real-World Examples from the Carrier’s Portfolio

In one Denver property, the agent reported an 8-year-old roof. ZestyAI identified it as 10 years old, confirmed by aerial imagery showing the replacement event.

In a Baltimore case, what was reported as a 5-year-old roof was actually 21 years old — verified through imagery and permitting history.

These weren’t edge cases. They reflected a systemic pattern across the portfolio.

The Impact: A 1.71% Improvement in Combined Ratio

By integrating verified roof age into underwriting and pricing workflows, the carrier achieved a 1.71% reduction in combined ratio. The improvement came from three measurable levers:

• Loss Cost Controls (-1.08%)
  Accurate age enabled appropriate use of deductibles and ACV endorsements, lowering claims severity.
• Better Risk Selection (-0.38%)
  More precise pricing attracted lower-risk properties while deterring higher-risk ones.
• Inspection Optimization (-0.25%)
  Confidence scores guided inspections to properties that truly needed them, reducing wasted expense.

Beyond loss ratios, better roof age data improved portfolio transparency, supported expansion into previously restricted markets, and strengthened actuarial and underwriting decision-making.

What’s Next: Expanding the Foundation of Property Intelligence

After proving the value of accurate roof age, the carrier is now building on that foundation. They are incorporating additional property attributes — including roof complexity, roof quality, and parcel-level features — through ZestyAI’s Z-PROPERTY™ platform.

By standardizing and elevating property data quality at scale, the carrier expects to unlock similar gains across quoting, underwriting, renewals, and even reinsurance discussions.

The takeaway is clear: in an industry built on precision, even a single data point — when made accurate — can deliver outsized impact.

Read the full Roof Age Accuracy case study to see how verified roof age drives measurable underwriting and pricing gains → From Self-Reported to Verified: Roof Age Accuracy That Pays Off

ZestyAI today announced that Augusta Mutual has selected ZestyAI’s Roof Age and Z-PROPERTY™ to enhance underwriting and rating accuracy, target inspections more effectively, and support sustainable growth across Virginia.

Based in Staunton, Virginia, Augusta Mutual is a single-state carrier serving Virginia since 1870 with a longstanding reputation for personalized service and local expertise. By upgrading from traditional imagery and inspection approaches to ZestyAI’s computer vision and machine learning technology, the insurer gains broader, more consistent property coverage and a more comprehensive, AI-driven view of property risk—unlocking property-level insights such as verified roof age, roof condition, vegetation overhang, and debris accumulation that directly influence claim frequency and severity.

“ZestyAI’s solutions bring a new level of precision to our underwriting process,” said Gretchen H. Collins, Vice President of Underwriting at Augusta Mutual.

“We moved from legacy property risk tools to gain broader, verified property coverage, helping us make faster, more consistent, and more confident decisions for our policyholders across Virginia.”

ZestyAI’s Roof Age delivers verified roof age by cross-validating building permit records with over 20 years of aerial imagery, detecting roof replacement events and assigning confidence scores across 97% of U.S. properties. Z-PROPERTY™ further enhances this insight by assessing roof complexity, materials, and condition, along with other parcel-level attributes that influence loss potential.

ZestyAI works closely with regulators to ensure transparency, validation, and continuous monitoring of its AI-driven models. Its portfolio of risk models has secured nearly 100 approvals from regulators nationwide, giving insurers confidence they can be deployed immediately with the accuracy and transparency regulators demand.

By Attila Toth, Founder & CEO of ZestyAI

The U.S. P&C industry enters 2026 with stronger balance sheets, renewed underwriting profitability, and a sense that the hardest part of the cycle may be behind it. But beneath the surface, the risk environment is moving in the opposite direction. Climate-driven loss volatility, localized catastrophe patterns, and structural property vulnerabilities are accelerating — even as markets begin to soften.

The result is a widening gap between carriers chasing growth and those wiring discipline deeper into how risk is selected, priced, and managed. Here are three dynamics that will define P&C performance in 2026.

1 — A Softer Market Meets a Hard Climate Reality

The industry enters 2026 from a position of renewed financial strength: the last couple of years produced the best U.S. P&C underwriting results in more than a decade, with combined ratios improving into the mid‑90s and a clear swing back to underwriting profit. 

Capital has rebuilt, competition is intensifying in many property segments, and some markets are now seeing flat or slightly negative renewals, encouraging carriers to cautiously re‑enter territories that were pulled back during the hard market. 

The risk environment, however, has not softened; insured catastrophe losses have exceeded USD 100 billion for multiple consecutive years, and recent nat‑cat studies now describe annual insured losses approaching USD 150 billion as the emerging “new normal,” driven disproportionately by severe convective storms, wildfire, localized flooding, and non‑weather water losses rather than a single headline hurricane season.

In 2026, carriers will not move in lockstep. Some will quietly relax property underwriting and broaden appetite to chase top‑line volume in what feels like a more forgiving market, even as U.S. SCS losses alone have entered a period where annual insured losses now consistently exceed USD 40 billion, while others will double down on discipline by wiring property‑level climate and vulnerability metrics into day‑to‑day decisions. 

Early in the year, the visible story may favor the volume‑chasers as premium growth accelerates, but by late 2026 the more revealing story will be in loss ratios, with hail‑, SCS‑, wildfire‑adjacent, and water‑heavy portfolios that were loosely underwritten posting the most uncomfortable deterioration.

2 — Hyperlocal Exposure Management Becomes a Core Profit Lever (and Reinsurers Will Expect It)

Even with some rate relief on better risks, carriers face a structural problem going into 2026: loss volatility is increasingly driven by frequent, highly local events and structural property issues rather than a single major catastrophe. 

A two‑block hail cluster, an ember‑exposed hillside parcel at the wildland–urban interface, or aging roofs can generate thousands of mid‑sized claims that erode margin even when headline cat activity looks “average.” 

When property‑level secondary modifiers are missing or stale, catastrophe models and capital providers default to conservative assumptions, inflating modeled losses, uncertainty loads, and reinsurance costs; reinsurers are responding by demanding clearer visibility into roofs, vegetation, defensible space, elevation, and mitigation before offering the most favorable terms.

In this environment, hyperlocal exposure management is becoming a core profit lever rather than a niche analytics exercise. Leading carriers are using verified parcel‑level attributes to identify frequency‑prone parcels inside ZIP codes that look stable in aggregate, to counter overly conservative model assumptions with auditable evidence, and to walk into reinsurance renewals with property‑level documentation rather than broad averages. 

They are steering appetite, pricing, inspections, and mitigation strategies on a near‑real‑time basis instead of waiting for annual rate cycles, effectively trading unmanaged volatility for intentional, data‑driven control. The net result is that 2026 will reward carriers that can prove property‑level truth to reinsurers, regulators, and their own underwriting teams, replacing assumptions with evidence and episodic adjustments with continuous portfolio management.

3 — Agentic AI Becomes Insurance’s Next Operating System

2026 is shaping up as the year agentic AI shifts from experimental to essential in P&C, as carriers discover that the binding constraint is no longer access to data but the speed, consistency, and defensibility of decisions across underwriting, filings, compliance, and product change. Risk conditions are moving materially faster than traditional annual guideline refreshes can accommodate, supervisors and rating agencies are sharpening expectations around explainability and consistency, and decades of underwriting and regulatory expertise are retiring faster than they can be replaced.

Across the market, early adopters are already using agent‑like systems to flag likely regulatory objections before filings go in, compress filing and approval timelines from months to weeks, and synthesize competitive and regulatory intelligence with strong safeguards and human‑in‑the‑loop review. These systems are also starting to refresh underwriting and pricing playbooks using live property‑risk signals instead of static territorial assumptions, closing the loop between climate data, filings, and front‑line decisions. For many carriers, 2026 will be remembered as the year AI stopped being primarily predictive and became operational infrastructure — software that can understand intent, reason through complex rules, coordinate multi‑step workflows, and take auditable action alongside human teams.

How Leading Carriers Are Responding

The most forward-positioned carriers entering 2026 are already using parcel-level intelligence to refine appetite, pricing, inspections, and mitigation in high-hazard and water-exposed regions, treating hyperlocal data as a core underwriting input rather than an afterthought.

They are refreshing eligibility criteria and underwriting guidelines based on property-specific hazard, vulnerability, and mitigation features, and preparing regulator-ready and reinsurer-ready documentation on defensible space, roof condition, and other secondary modifiers.

They are steering portfolios continuously, adjusting aggregates, concentrations, and mitigation incentives throughout the year instead of relying solely on renewal season to reset course. Together, these behaviors signal a broader shift away from episodic, once-a-year recalibration toward continuous, property-level risk management supported by AI-enabled operating systems.

Insurance availability has become a constraint on the housing market.

That’s the central argument Ross Martin, VP of Risk Analytics at ZestyAI, makes in ATTOM’s newly released Q4 2025 Housing News Report—and it’s one that will increasingly shape affordability, underwriting, and buyer behavior heading into 2026.

Housing discussions still focus on mortgage rates and inventory. But in many markets—especially catastrophe-exposed ones—insurance is becoming a gate in the transaction. If a property can’t get insured, or coverage is uncertain, deals stall. And when insurance costs spike unexpectedly, affordability breaks even when the mortgage penciled out.

Ross’s point isn’t simply that insurance is getting more expensive. It’s that availability and predictability now matter as much as price—and the market would function better with clearer, property-level risk signals.

Today, homes in similar locations can carry meaningfully different risk based on factors like roof condition and materials, defensible space and vegetation management, yard and debris conditions, and documented improvements captured in permits or listing data. When those distinctions aren’t consistently reflected in underwriting or pricing, mitigation efforts go unrewarded—and higher-risk properties don’t get early, property-specific signals to improve.

For insurers, this lack of granularity creates real portfolio risk. When individual properties aren’t differentiated clearly enough, volatility increases, adverse selection becomes harder to avoid, and long-term participation in catastrophe-exposed markets becomes less sustainable. Property-level, mitigation-aware models help address this by improving segmentation and enabling insurers to stay in market with more confidence.

Recent advances in property-specific data and modeling now make this differentiation possible at scale. Insurers can assess dozens of attributes—including roof age and materials, defensible space, vegetation conditions, building permits, occupancy type, and hazard-specific science—to build a clearer view of a structure’s vulnerability. Just as importantly, these models can recognize mitigation actions—like roof replacements, defensible space creation, and debris removal—and incorporate them more consistently into underwriting and pricing.

When mitigation is visible and rewarded:

• Homeowners and investors gain more control over premiums
• Insurers can maintain more stable portfolios, even in high-risk regions
• Housing markets get clearer signals—making insurance availability and long-term cost less of a guessing game for buyers and lenders

Regulators are paying attention as well. In several states, regulators are examining how property-level data, mitigation, and modern risk modeling approaches can be incorporated more consistently into rate structures, with transparency as a common objective.

The takeaway is straightforward: insurance is shifting from a background cost to an active constraint—and clearer, property-level risk signals are key to easing that constraint. As 2026 approaches, the ability to differentiate risk at the individual property level will play a growing role in restoring predictability, supporting availability, and shaping housing market outcomes.

Read the full article, “The Insurance Shift Reshaping the 2026 Property Market,” in ATTOM’s Q4 2025 Housing News Report.

‍

By Abdul Mohammed, Director of Product Marketing, ZestyAI

Every year, carriers and filers submit hundreds of thousands of rate, rule, and form transactions to state insurance departments (DOIs), many through SERFF. In 2023 alone, SERFF processed 517,571 transactions (NAIC SERFF, reported 2025). These filings are the DNA of the insurance market: the definitive record of how competitors set rates, where they plan to expand, and where they get approval for new ideas or pull back.

Even though much of this information is publicly available in many jurisdictions, with different rules for access and confidentiality, most carriers still miss out on the competitive signals hiding in plain sight. 

Key takeaways

• Regulatory filings are a strategic dataset, not just compliance paperwork. They reveal competitor intent and market shifts.
• “Public” does not mean “easy to analyze.” Filings are often massive, cross-referenced, and inconsistently structured, making manual review at scale impossible.
• The winning approach is a filing intelligence stack. Success requires filing data tracked across amendments and effective dates, structured parsing, deterministic calculation, and precise citations.

How Filings Got So Complicated

From 1945 to The Modern Filing Ecosystem

The story begins with the McCarran-Ferguson Act of 1945, which gave states the power to regulate insurance rather than the federal government. As a result, requirements differ by state, making rates, rules, and forms complicated across jurisdictions. 

In 1998, the National Association of Insurance Commissioners (NAIC) introduced SERFF (System for Electronic Rate and Form Filings), developed in collaboration with regulators and industry to digitize rate and form submissions. While it successfully moved filings online, the underlying complexity of the content remained.

Since then, the volume and complexity of filings have exploded. Our analysis of SERFF filing packages shows that the largest homeowners’ rate filing now exceeds 300,000 pages, including attachments, exhibits, and correspondence. In our data, the biggest package grew from 38,102 pages in 2009 to 346,064 pages in 2024.

‍

Why Public Filings Remain Inaccessible for Analysis

A public filing is not always easy to access or understand. Carriers know that regulators, consumers, and competitors will review their filings, so the resulting documents are often dense, cross-referenced, and hard to piece together. If you have ever tried to reverse engineer a competitor rate change, you have probably faced these challenges:

The Trade Secret Exception 

Some carriers request confidential treatment for specific elements of a filing that may qualify as trade secrets under state rules, such as granular territorial data or specific model inputs. When those sections are redacted or withheld, you lose visibility into important details even though the overall filing is public, and the level of protection varies by jurisdiction.

"When a Product Filing contains Trade Secret information, the Product Filer may identify those portions of the Product Filing, including correspondence with the Compact Office, that contain Trade Secret and seek to protect their disclosure." 

Interstate Insurance Product Regulation Commission, FIN 2024-1

The Reference Maze 

Sometimes, instead of putting all rate information in one document, a filing will reference several other filings across multiple years. To fully understand the change, an analyst has to track down and reconcile multiple historical filings, creating a confusing trail of "breadcrumbs"  that is almost impossible to follow manually.

The PDF Image Trap 

Many carriers submit rate tables as scanned images in PDFs instead of machine-readable text. While human eyes can read these tables, most data tools and basic OCR software treat them as pictures, so the information cannot be easily searched, filtered, or analyzed at scale.

The Objection and Response Trail

Often, the most valuable intelligence isn't in the initial filing, but in the "Objection and Response" exchanges between the carrier and the state regulator. These discussions can reveal rationale, supporting evidence, and the boundaries regulators will accept. However, this material is often spread across multiple attachments and correspondence, making it easy to miss critical insights without a structured way to collect and review them.

Non-Standard Nomenclature

There is no universal dictionary for insurance variables. One carrier might call the roof age factor “rf_yr_mod”, while another uses “const_age_rel”. Without a way to normalize these labels, mapping equivalent factors across carriers becomes manual work, making benchmarking slow, error-prone, and difficult to repeat.

Why General Purpose LLMs Often Struggle Here

We are in the age of Generative AI, so the natural question is: "Why not just upload these PDFs into a tool like ChatGPT or Gemini and ask questions?"

You can upload filing PDFs to ChatGPT or Gemini, and you may get a helpful summary. But insurance filing work is not “writing assistance.” It is a precision workflow in which small mistakes lead to incorrect conclusions. General-purpose LLMs are built to generate plausible text from the input you give them, not to reliably preserve filing structure, track versions, run exact calculations, and produce audit-ready citations. 

The following are the top reasons why general LLMs often struggle in the inusrance domain:

1. The Math Problem

Insurance filings require exact math and exact linkage across tables, factors, relativities, and formulas. LLMs are probabilistic; they predict likely answers rather than performing exact calculations. If you ask an LLM to calculate a 3.5% rate increase over three years using a specific table, it may give a confident answer that is still wrong. In insurance, even a 0.01% mistake can mean millions in lost premium. As actuarial researchers noted in a 2024 paper from Cambridge University Press, “while LLMs can explain concepts, they often provide inaccurate or incorrect mathematical facts, sometimes in subtle ways.”

2. Structure Blindness

LLMs are mostly trained on regular text, such as books and articles. They are not skilled at understanding tables, following footnotes, or applying formulas consistently across multi-part exhibits, especially when documents are scanned or formatted inconsistently. A standard LLM treats a table as plain text and often fails to understand how the cells are logically and mathematically connected.

3. Context Window Overload

State filings are often longer than what standard LLMs can handle. If you give a model a 2,000-page document, it may lose track of what appeared earlier in the filing and still try to answer questions as if it remembered everything. This can cause the AI to make up numbers (hallucinate) to fill in missing information. 

What a Modern Filing Intelligence Approach Looks Like

The industry does not need a better chatbot. It requires a filing intelligence stack that combines structured data, deterministic computation, and auditable reasoning.

Below are the key steps in building this intelligence stack:

Build a clean, versioned filing archive 

Ingest filings continuously and preserve them with consistent metadata such as state, carrier, line, status, effective date, and relationships to related submissions. This creates a single, reliable system of record for all filing history.

Parse filings into insurance native components

Break filings into rates, rules, forms, exhibits, objections, and responses. Store rating tables, factors, and hierarchies as structured data instead of plain text, so they can be queried and reused.

Pair language with deterministic calculation

Use deterministic engines for calculations and rate reconstruction, then use language models to explain the results, clarify their meaning, and support structured analysis. The math engine produces the numbers; the LLM explains what they mean.

Make everything traceable

Every conclusion should link back to the exact filing section it came from. This traceability is what turns AI output into something regulators, actuaries, and executives can trust and defend.

The Future Belongs to the Agile

The number of filings and their sizes keep growing. As climate risk reshapes markets, rate reviews and underwriting changes will happen more often. The carriers who succeed will be those who treat regulatory filings as a source of strategic insight rather than a compliance burden. 

Those who don’t will pay the price. Without visibility into how competitors are changing rates, rules, and eligibility in near-real time, carriers slip out of sync with the market, underwriting yesterday’s risk at today’s loss costs. That gap is where adverse selection takes hold.

Having access to filings is not enough to gain an edge. The real advantage comes from understanding them quickly and accurately, in a way that can be repeated across teams and product lines. ​​To do this well, you need systems built for the job, not just a general-purpose model. This is where a purpose-built filing intelligence stack changes what is possible.