In the legal profession, the work is only as strong as its support. A good argument isn't just persuasive, it's backed by citations. You can point to the contract clause, the case, the exhibit, and the chain of reasoning that got you to the conclusion. That's not academic formality. It's how legal work survives scrutiny in a court of law. 

Insurance is moving in the same direction.

For decades, insurance operations could lean on experience and muscle memory. A tenured underwriter knew what "this form usually covers." A claims leader knew the standard response posture. A broker knew which markets were flexible. That knowledge still matters, but the environment has changed. Regulators, legal teams, and procurement groups now expect decisions to be documented, explainable, and consistent over time.

"Trust me" is no longer a reasonable operating model.

Why legal workflows look the way they do

Legal work is predicated on the ultimate potential that it will end up in front of a judge. This fear shapes all the work lawyers do. 

A motion, an opinion letter, or a contract position might get scrutinized months or years later by a judge, with millions of dollars at stake. The only way to truly prepare for that situation is if the work product is structured to be audited. That's why legal workflows emphasize three things:

Citation. Show exactly where the claim comes from.

Reasoning. Make it possible to retrace the reasoning steps from source to conclusion.

Conclusion. Make it easy for another expert to validate or challenge the conclusion by having a very clear and articulate conclusion.

These practices aren't about slowing work down. They're how the legal industry moves quickly while staying defensible when the stakes rise.

Insurance is discovering the same truth, especially in claims and coverage interpretation.

Insurance is already under similar scrutiny

Insurance has always been regulated, but scrutiny is broader now and comes from more directions, including clients. Decisions can trigger omissions, bad-faith allegations, and liabilities that far exceed a coverage dispute.

State-by-state variability adds another layer. A defensible decision in one jurisdiction may be incomplete or risky in another.

At the same time, the work is deeply document-driven. Policies, endorsements, submissions, claim files, and correspondence are still stored in PDFs, scans, and formats that were created for manual review.

That means insurance decisions are often anchored in unstructured language that must be read carefully, compared across documents, and defended later.

In short, insurance faces legal-like constraints whether it realizes it or not.

The AI factor raises the bar, not just the speed

AI is often discussed as a productivity lever, but in insurance, the real challenge is credibility. When an AI-supported decision gets scrutinized, you need to show the basis for it. If the answer is a black box, you've created a new type of risk.

That's why the industry is increasingly prioritizing accuracy, explainability, and consistency over speed alone.

It's also why "model drift" matters. If a tool's behavior changes over time, it undermines consistency and auditability in regulated workflows.

This is one place where legal has a head start. Many legal technology workflows were designed around precedent and review. The focus is less about generating text and more about interpreting documents with citations and a clear path from source to conclusion.

Insurance now needs the same.

The future of insurance work

This shift isn't theoretical. It changes how teams should define quality.

In an insurance workflow built to withstand scrutiny, a good outcome isn't only correct—it's defensible. That means:

Your conclusions should point back to the policy language. Coverage positions and claim decisions need to be anchored in the actual text, not just institutional memory. Insurance has long depended on expert judgment in how professionals read policies, interpret exclusions, and apply precedent. The next step is making that judgment visible and reproducible.

Your reasoning should be transparent enough for peer review. If a colleague can't follow how you got there, an auditor or regulator won't either. Transparent reasoning isn't a luxury in high-stakes decisions, it's a requirement.

Your process should be consistent across teams and time. Insurance is full of niches and specialized expertise, but inconsistency is costly. As experienced practitioners retire, decision quality can decline if judgment stays trapped in individuals. Understanding the policy is the hardest problem in insurance. You can't solve it with expertise that only exists in people's heads.

Your documentation should be structured for "future you." Claims files, coverage analyses, and underwriting notes should read like work that's expected to be examined later. That's the legal mindset, and it's becoming the insurance mindset.

This is also why many leaders are talking less about flashy AI and more about repeatable operating models. The most valuable AI in insurance will look like consistency, documentation, and fewer surprises.

The practical takeaway for insurance leaders

If you're leading claims, legal, compliance, or brokerage operations, the question isn't whether your teams will use AI and automation. They already are. 

The real question is whether you're building systems that will hold up when scrutiny arrives. That means setting standards for citation, traceability, and reviewability in the work itself, not as an extra step at the end.

It also means resisting tools and processes that optimize for speed while sacrificing transparency. In insurance, "close enough" is often where the risk begins. 

Insurance is learning a lesson the legal industry learned long ago: when the stakes are high, the work must show its sources.

The AI market in the insurance industry is set to hit $80 billion by 2032, yet nearly two-thirds of carriers have a gap between their AI vision and reality. In essence, carriers understand what they want from AI and see the significant value it can drive but are struggling to actually put this into practice, especially in the property sector.

Marrying vision and reality is critical, however, as carriers look to scale automation, re-orchestrate and transform workflows, and fully realize the potential of AI across the lifecycle of a claim. There is a substantial downstream impact on claimants and clients, as well. 

There are a number of steps a carrier needs to take to find success with AI in 2026 and ensure they are future-proofed and ready for the next era of property claims.

The AI Journey for Property Carriers

First and foremost, carriers need to understand where they are on their AI journey. This is important in identifying what the next steps are, what's feasible in the short term, and eventually in the long term, and aligning company communications, operations, workflow, training, and more.

At the moment, anywhere from 58-82% of carriers are leveraging AI tools in their operations, but only 12% claim to have fully mature capabilities, and only 7% have achieved scalable AI success. This means that 93% of carriers are still in the part of their AI journey in which they are identifying how to scale AI to a point at which it is driving real, measurable outcomes. What we've seen so far is that adoption of AI has been popular in areas such as intake, triage, and documentation, but fully integrated technology and end-to-end AI workflows are still far away for most carriers. This, in turn, results in a fragmented technology experience, rife with different tools, vendors, and solutions. This limits AI's impact. It keeps value confined to each step in the lifecycle of a claim, can lead to inconsistent data or silos across systems, and weakens output.

Reliance on pilot programs or point solutions is the first step in an AI journey, but it certainly can't be the last. Most importantly, this technology is rapidly advancing, and the longer it takes carriers to find value and scalability, the further they'll fall behind the competition.

The Challenges Facing Carriers

There are three main challenges facing carriers. First is integration of AI into legacy technology. The majority of claims systems weren't built with API connectivity in mind, which introduces difficulties immediately into scaling this technology across workflows. Before integration even begins, carriers need to ensure that their claims systems can support orchestration.

Second is training a disconnected property workforce. An often-overlooked aspect of AI in the property space is preparing for the challenges that can arise when adjusters are managing heavy caseloads and working in the field. Support systems are critical to success in this area, and AI and any other new tools cannot feel like a burden to them. Training and communication in best-use cases are important in presenting these tools as benefits. This can be streamlined through rollout plans that align with day-to-day workflows, prioritize flexibility, and implement continuing training opportunities.

Finally, expecting AI tools to drive perfection is a key challenge. This technology won't deliver perfect outcomes from day one, but through gradual improvements can drive real change in processes. If too much focus is placed on perfection, then widespread implementation can be delayed. Instead, carriers should prioritize progress first and perfection second when measuring AI against real-world baselines, with the goal of refining capabilities over time.

What Real Impact Can Look Like

When challenges are addressed and overcome, and carriers understand how to progress from point A to point B in their AI journey, real impact can be achieved. This will be seen in a number of ways across operations.

Main points of impact will be evident in faster claims reviews in which AI is helping adjusters summarize claims, extract data, and capture notes more efficiently, as well as in improved program outcomes, smoother workflows across internal and external systems, and smarter claim routing. AI tools can evaluate loss severity, complexity, and fraud risk at intake, assisting in routing claims to the right recovery resource sooner.

In addition, adjusters will see stronger field operations through enhanced drones, sensors, and tablets, which enable faster mitigation, better assessments, and quicker resource mobilization.

The data backs up this impact. Intake automation has reduced average claim processing from 10 days to 36 hours, AI photo analysis boosts claim handling efficiency by 54%, and much more.

Prioritizing Human Expertise

A key consideration in the integration of AI is the increasingly important role that humans play, and will continue to play, in this process. AI should be seen as a tool to augment and support workforce expertise, rather than replace it.

This technology is powerful, but cannot be used to replace human judgment, empathy, or real-world experience in the claims process. Losses can be ambiguous, emotionally sensitive, or require nuanced, complex coverage decisions that AI cannot handle, and require human professionals to consider context, communicate clearly, and advocate for policyholders throughout each stage.

In essence, AI is a catalyst, not a cure-all, and carriers must aim to apply AI selectively while keeping people at the center of their claims decisions. Striking this balance will be the difference in staying ahead and remaining competitive in a rapidly changing technological and regulatory landscape.

For Sedgwick's full report on "Future-Ready Property Claims," click here.

The catastrophe insurance crisis facing American municipalities isn't a crisis of risk — it's a crisis of imagination.

When the insurance industry tells municipal leaders to "invest in resilience" without developing insurance products that reward those investments with affordable coverage, they're asking communities to subsidize the industry's own innovation deficit. When carriers collect premiums from communities while simultaneously withdrawing coverage and providing no recourse to close protection gaps, they leave municipalities to bear the burden of adaptation alone. And when carriers pull away from vulnerable communities while posting record profits, their risk decisions become a way of abdicating their core function of pooling and transferring risk.

This mismatch is unsustainable and dangerous, and the consequences cascade far beyond equity concerns. Low-income residents can't rebuild after disaster. Without business interruption coverage, small businesses close, devastating local employment and sales tax revenue. Affordable housing providers, facing uninsurable properties, stop developing in climate-vulnerable areas, exacerbating housing shortages and displacement. These "equity issues" then become systemic economic failures that threaten the tax base carriers rely upon for commercial lines business.

If carriers don't evolve, they won't just be disrupted by climate change; they'll be made irrelevant by it. To meet the needs of climate-challenged jurisdictions and address these cascading failures, carriers must dramatically increase their R&D investment across product, distribution, and capital structure.

Product Innovation

The catastrophe insurance products available to municipalities today are fundamentally inadequate for current climate realities, let alone future projections. Carriers must invest in genuine product R&D across several critical areas.

Microinsurance for Vulnerable Populations: Carriers should develop lower-limit, lower-premium products specifically designed for low-income households and small businesses. Communities with functioning insurance markets maintain economic activity after disasters while communities without insurance face cascading failures that destroy the commercial premium base.

Advanced Modeling That Provides Incentives for Novel Adaptation: Current catastrophe models treat risk as static or worsening, with no meaningful premium reduction for innovative resilience projects. Carriers must bolster their modeling capabilities, so they can actually quantify the risk reduction from novel adaptation investments. This requires fundamental R&D investment in coupled physical-financial modeling, not incremental improvements to existing cat models.

Parametric Insurance for Rapid Response: Traditional indemnity insurance is ill-suited for disaster response, but parametric products that trigger immediate payouts based on objective measurements (flood depth, wind speed, earthquake magnitude) enable rapid recovery while reducing time and costs. The question is whether carriers will invest in developing parametric products tailored to municipal needs or whether municipalities will turn to specialized parametric providers who will.

Distribution Innovation

Even when adequate insurance products exist, traditional distribution channels systemically fail to reach those who need coverage most. Insurance carriers must invest in new distribution models, particularly embedded insurance mechanisms that integrate coverage directly into existing community touchpoints.

Affordable Housing: Catastrophe coverage can be embedded directly into affordable housing and stay with the property, not the individual tenant or owner. This closes protection gaps for vulnerable residents while creating stable, pooled risk for carriers.

Utility Bill: Municipalities can embed catastrophe coverage directly into customers' utility bills, creating universal protection while drastically reducing acquisition costs and adverse selection.

Employee Benefit: Municipalities and anchor institutions (hospitals, universities) employ thousands of workers, many of whom lack adequate catastrophe coverage. Embedding basic catastrophe protection in employee benefit packages closes protection gaps while creating group purchasing efficiency.

Small Business District: Local business districts and chambers of commerce can serve as aggregators for embedded small business catastrophe coverage. Small businesses often lack business interruption and property coverage. District-level embedded insurance solves the distribution problem while enabling risk pooling across merchants.

Capital Structure

Perhaps the most fundamental innovation required is at the capital structure level. Traditional reinsurance capital is designed to maximize returns for institutional investors and reduce exposure to complex, small-scale risks that don't fit standardized cat bond structures.

Community Development Reinsurance Institutions (CDRIs) — mission-driven nonprofit reinsurers structured similarly to community development banks — offer an alternative capital model specifically designed to support municipal resilience and insurance market function.

Traditional carriers should see CDRIs not as competitors but as catalysts for market development. By providing reinsurance for products serving underserved markets, CDRIs enable primary carriers to write business they otherwise couldn't while maintaining risk tolerance within board-approved limits. This expands the overall insurance market rather than displacing existing business.

Yet most carriers remain unaware of or unengaged with the emerging CDRI sector. If carriers don't invest in understanding, partnering with, and leveraging mission-driven reinsurance capacity, they'll soon discover that CDRIs have enabled an entirely new ecosystem of insurance providers serving municipalities—providers who didn't need traditional carrier participation to succeed.

The Choice Before Carriers: Lead, Follow, or Become Obsolete

The insurance industry faces a stark choice: invest in the R&D necessary to develop products, distribution channels, and capital structures adequate to municipal climate realities or continue business as usual and risk becoming obsolete.

The communities that carriers are abandoning won't simply accept uninsurability. They'll build alternatives — self-insurance pools, parametric coverage through specialized providers, embedded insurance through MGAs, risk financing through CDRIs and innovative bonds that will chip away at carriers' market relevance. Eventually, the "alternative" insurance ecosystem serving municipalities will become the mainstream, and traditional carriers will lose out on large parts of the market.

This isn't speculation; it's already happening. The only question is whether the insurance industry will rise to meet the demands and challenges in the municipal markets.

While the excess and surplus lines market was once an option of last resort, today it is all too frequently a first step in the process of insuring risk.

A coastal property facing new catastrophe models. A business navigating cyber exposure. A specialized liability account that has outgrown admitted guidelines. For agents and brokers, E&S is now part of everyday operations.

This shift has forced the distribution side of the industry to move faster, communicate more clearly and operate with greater precision. Technology is becoming a bridge to help insurance agencies keep pace while also strengthening relationships with their carrier partners.

Speed matters, but clarity matters more

Unlike admitted markets, where rates and underwriting changes can take time to filter through regulatory processes, non-admitted appetites can shift quickly based on loss trends, capacity and real-time market conditions. A carrier writing a class of business today may pull back tomorrow or adjust pricing as results demand it.

For the retail agent, this reality creates a constant challenge: Where does this risk belong currently? In the past, the answer has required multiple submissions, follow-up emails and trial-and-error market shopping. That cycle slows service, strains staff resources and frustrates underwriters who receive incomplete or misrouted submissions.

Technology can help avoid this cycle of frustration and wasted time, not by replacing relationships but by reducing the friction that can damage them.

Streamlined placements support better partnerships

Modern E&S placement platforms are designed to make submissions cleaner, faster and more consistent. The best tools help agents submit once, validate completeness and route risks to the right markets based on current appetite.

This kind of upfront triage benefits all involved. Agents spend less time chasing dead ends. Underwriters spend less time sorting through half-built submissions. Carriers receive applications closer to their appetites, with clearer exposure data and fewer missing pieces.

The result is a more efficient exchange that respects the time and expertise on both sides of the relationship. We’re seeing that with the deployment of Xchange - Powered by SIAA, which provides our members a faster, cleaner and easier way to access and place E&S business. 

Reducing errors and improving underwriting confidence

One of the most persistent challenges in E&S is submission accuracy. When clients want fast answers, agency teams sometimes make assumptions to move the process along. These seemingly educated guesses can create big delays later when an underwriter must circle back for corrections.

Technology that enriches submissions with third-party data sources can reduce the burden. Property records, hazard data and other verification tools can help confirm details before the submission ever reaches the carrier.

Doing this leads to fewer surprises, fewer resubmissions and a smoother path to a quote. More importantly, it helps carriers trust what they are seeing, which ultimately contributes to stronger carrier-agent relationships.

AI should be an optimizer, not a replacer

Artificial intelligence is playing a growing role in the E&S workflow, but the industry must be clear-eyed about its realities.

AI can help organize information, identify inconsistencies and accelerate routing. It can reduce manual data entry and make it easier for agents to package risks in an underwriter-ready format.

What it cannot do is replace underwriting judgment.

Complex accounts still require human experience, context and expertise. Technology works best when it clears away administrative clutter so underwriters and agents can focus on conversations that matter: coverage structure, risk controls, exclusions and long-term strategy. When positioned correctly, AI supports relationships rather than threatening them.

Strengthening carrier relationships through better submissions

Carrier relationships are built on trust, consistency and professionalism. In the E&S space, where underwriters face heavy submission volume, standout agencies are those that deliver clear narratives and decision-ready accounts. Technology helps agencies meet that standard at scale.

By standardizing intake, improving exposure clarity and managing workflow discipline, agents become better partners to their markets. Carriers benefit from lower acquisition expense per policy, improved risk selection and fewer wasted cycles.

Over time, these operational advantages translate into stronger long-term collaboration.

Carriers tend to prefer distribution partners who can deliver reliable data quality and efficient servicing without requiring carriers to expand headcount at the same rate as submissions.

Agencies that adapt will protect their growth

For agents and brokers, the risk of ignoring technology is not about missing a trend. It is about falling behind both the market and the competition.

As risks become more complex, turnaround time is becoming a competitive differentiator. Agencies relying solely on inbox-driven workflows will find it harder to shift books of business, maintain service levels and compete for talent.

The goal is not to adopt technology for the sake of shiny tech solutions. Rather, the goal is to protect the value of the agency by making the placement process faster, cleaner and easier to hand off to the next generation.

Relationships remain at the center

E&S will always involve more complexity than standard business. But complexity does not have to mean inefficiency. With the right technology, agents and brokers can keep pace with a rapidly evolving market while building better carrier relationships through stronger submissions, smarter routing and clearer communication.

The future of E&S distribution will not be defined by replacing people. It will be defined by empowering them. When technology reduces friction, relationships have room to grow.

We've all had a chuckle about the occasional hallucination by generative AI: the time when it recommended using glue to keep cheese from sliding off a piece of pizza; when an Air Canada chatbot promised a passenger a bereavement fare despite a policy to the contrary, and the airline had to live up to that promise; when a lawyer unknowingly submitted a brief to a judge that was based on citations from court cases that never happened; and so on. 

But a couple of recent stories go well beyond the chuckle level. While generative AI continues to show all the promise in the world, these stories demonstrate consistent problems that, unchecked, would lead to severe consequences. 

Let's start with the one about war games in which large language models almost always recommended escalating to nuclear weapons.

As Axios reports, a researcher at Kings College London pitted three popular LLMs — GPT-5.2, Claude Sonnet 4 and Gemini 3 Flash — against each other in 21 war games in which the AIs acted as the leaders of major nations. The scenarios included threats to survival, but also included lower-stakes conflicts, such as border skirmishes and resource competition. Yet 95% of the time, at least one of the LLMs "used" nuclear weapons, and escalation typically ensued. 

For anyone without a strong Dr. Strangelove streak, those results reflect a scary misjudgment. While the U.S. and the Soviet Union considered tactical nuclear weapons to be legitimate parts of their arsenals in the early years of the nuclear age, those were also the times when the countries casually considered using nuclear weapons for industrial uses such as mining and natural gas extraction. It's been clear for decades that nuclear weapons are simply too powerful for their effects to be limited to legitimate military or industrial targets. 

Even at one kiloton, the smallest payload for what's considered a tactical nuclear weapon, the explosion would be 100 times as powerful as the biggest conventional bomb in the U.S. arsenal. At the top end of the range for a tactical nuclear weapon (generally considered to be 100 kilotons), the explosion would be some seven times as powerful as the bomb dropped on Hiroshima, which destroyed a military target but also killed an estimated 140,000 people, the vast majority of them civilians. The radiation released can also reach far beyond the targeted area. 

While the Kings College researcher noted that no one is handing AIs the keys to nuclear weapons systems, he said, "Militaries are already using AI for decision support — and research suggests those systems may lean into rapid escalation under pressure."

The other article that caught my eye relates to ChatGPT Health. The app, launched in January, is consulted by some 40 million people every day — and a study found the potential for major problems with the app's diagnoses. For more than half of the study's hypothetical patients who should have sought immediate medical care, ChatGPT Health told them they should stay home or wait to schedule a regular appointment with a doctor. 

The article, in the Guardian, said: "In one of the simulations, eight times out of 10 (84%), the platform sent a suffocating woman to a future appointment she would not live to see.... Meanwhile, 64.8% of completely safe individuals were told to seek immediate medical care."

For the study, published in the journal Nature Medicine, researchers created 60 realistic patient scenarios covering health conditions from mild illnesses to emergencies, then presented those scenarios to ChatGPT Health in various ways: changing the gender of the patient, sometimes providing test results, sometimes adding comments about what "friends" advised, etc. Three independent doctors reviewed each scenario and agreed on the level of care needed, based on clinical guidelines.

The study found that ChatGPT Health did well on textbook emergencies such as stroke and severe allergic reactions. But "'what worries me most,'" a doctor is quoted as saying in the article, "'is the false sense of security these systems create. If someone is told to wait 48 hours during an asthma attack or diabetic crisis, that reassurance could cost them their life.'”

Any number of health experts have extolled the potential for AI-based health advice, coupled with wearables and telemedicine, to revolutionize healthcare — providing care to the elderly and to people in rural areas, who would otherwise have difficulty getting access, while slowing the inexorable rise in healthcare costs. And I've bought in: Chunka Mui, Tim Andrews and I included a lengthy scenario about the potential for AI-based healthcare in our 2021 book, "A Brief History of a Perfect Future."

I still think the potential is there, too. As OpenAI, the developer of ChatGPT, told the Guardian, the app is updated and improved all the time, and I hope they keep charging ahead. (OpenAI also said it doesn't believe the study reflects how people actually use ChatGPT Health.)

But I also hope they are constantly checking for problems such as those identified in the study, and anyone else using AI in situations with major consequences should exercise similar care. That includes insurers, and not just in healthcare. As we feel our way toward using AI agents, we need to be very careful to not only vet them before putting them into production but to then supervise them — because they absolutely will make mistakes — and to keep improving them.

Cheers,

Paul

In the timeless words of Sun Tzu in The Art of War: "If you know the enemy and know yourself, you need not fear the result of a hundred battles." Today, in the battle against fraud in business and government programs, entity resolution—combined with knowledge graphs and geospatial analytics—serves as that ultimate weapon, akin to Excalibur, the legendary magical sword that could cut through anything.

When it comes to fighting fraud, it cuts through layers of deception, revealing hidden connections between people, businesses, transactions, and locations that fraudsters purposefully endeavor to keep obscured. By mapping out entities and resolving disparate records across dispersed systems to the real individuals and organizations behind them, investigators gain the clarity to validate transactions, expose invalid transactions, and dismantle fraudulent networks.

Fraud in government programs and business operations thrives in the shadows of fragmented data: mismatched names, shell companies, fake addresses, synthetic identities, and manipulated locations. Without a unified view, billions of dollars are lost annually to schemes like improper benefit claims, procurement kickbacks, subsidy abuse, "paper mills," and phantom vendor payments.

Entity resolution bridges these gaps, linking records across databases—names and addresses, tax filings, business registries, transaction logs, social media, and public records—to create a "360-degree" profile of every entity involved.

Entity Superpower — Unmasking the True Actors

At its heart, entity resolution determines when multiple records refer to the same real-world person, business, or location, despite variations in spelling, abbreviations, typos, or deliberate obfuscation. Advanced algorithms and machine learning handle the noise: "John A. Smith LLC" might resolve to the same entity as "JAS Enterprises" owned by "Jon Smith," especially when tied to shared addresses, phone numbers, or transaction patterns.

When integrated into knowledge graphs, these resolved entities form connected networks of relationships—ownership links, family ties, shared board members, or transaction flows. Adding the basics of address geocoding and geospatial analytics overlays physical reality: mapping addresses, proximity of claimed locations, or clustering of suspicious activities in specific regions. This data fusion transforms isolated data points into a battlefield looking glass that maps where fraud patterns emerge clearly.

Consider a classic red flag in government-funded programs: more licensed or funded daycares than the number of children in an area could possibly require. Entity resolution uncovers this by resolving provider records to actual owners and cross-referencing enrollment claims against demographic data. Knowledge graphs reveal networks of colluding owners registering multiple entities at the same address or funneling funds through connected shell companies. Geospatial views highlight unnatural concentrations—clusters of daycares in low-population rural zones or urban blocks with improbable child-to-provider ratios—signaling potential ghost operations or subsidy farming.

So, as with childcare, insurance companies may apply entity resolution to chiropractors, MRI facilities, and clinics, but in addition now the named insured, agent, claimant, and adjuster meld in with medical providers, equipment, legal staff, vendors, and others in the graph across any line of business. As lines are combined and companies join forces, this process can literally map trillions of dollars of historical premiums and claims that could influence real-time payments.

The King's Sword Trumps All Use Cases

Drawing from innovative applications across business and government using knowledge graphs for fraud detection, the combination of entity resolution, knowledge graphs, and geospatial tools exposes fraud across diverse domains:

• Government Benefit and Subsidy Fraud: In childcare subsidies, housing assistance, unemployment benefits, or agricultural grants, resolved entities expose operators claiming impossibly high volumes. Geospatial analysis flags unnatural provider distributions relative to demographics, while knowledge graphs uncover collusive networks funneling funds through connected shells or using stolen identities for enrollment claims.
• Procurement and Contract Fraud: Vendors often conceal conflicts via layered ownership or bid-rigging. Entity resolution connects bidders to officials' associates or hidden entities; geospatial overlays reveal fictitious delivery sites or illogical routing; graphs detect circular payments or anomalous bidding patterns indicative of corruption.
• Fake Business and Identity Schemes: Fraud rings create phantom companies for loans, grants, tax credits, or PPP-style programs. Resolution merges digital and physical footprints—such as mismatched websites/IPs with abandoned addresses—while geospatial clustering pinpoints registration hotspots tied to broader scams.
• Money Laundering and Illicit Flows: In trade-based or benefit-related schemes, resolved entities link actors across jurisdictions. Knowledge graphs map multi-hop transaction chains; geospatial tools visualize fund movements against claimed origins, exposing laundering through high-risk locations or mismatched geographies.
• Insurance Claims Fraud: In property insurance schemes, fraudsters stage incidents like water damage during homeowners' vacations, directing repairs to complicit restoration providers. Entity resolution links claimants, properties, and service providers across cases, revealing common identities or ownership ties; knowledge graphs highlight recurring patterns in damage types, timing, and vendor involvement; geospatial analytics maps claim locations against provider clusters, unmasking organized rings exploiting insureds and property owners.

In auto insurance, staged accidents generate multiple unrelated passengers all seeking medical treatment from the same provider and being represented by the same lawyer even though they themselves may live far apart and curiously are frequently unable to be located.

The schemes for various lines of casualty and property in auto, home, workers' compensation, and commercial insurance all are well mapped by the NICB (National Insurance Crime Bureau). And new schemes are emerging all the time — especially with the backing of transnational criminal organizations, but also with just everyday people getting creative with generative AI.

En Garde — the Industry Keeps Its Hand on the Hilt

As fraud schemes grow more sophisticated with digital mapping tools and global reach, entity resolution in knowledge graphs—enhanced by geospatial context—will only sharpen. Real-time monitoring, AI-driven anomaly detection, and dynamic mapping will make deception harder to sustain. The result? Interdiction of transactions. Faster and better recoveries. Frustrated, if not deterred, criminals. Lower premiums for insureds. Safeguarded public funds.

In the war on fraud, knowledge is power—but resolved, connected, and spatially aware knowledge is the key to victory. Like Excalibur drawn from the stone, we across these industries, companies, and public bodies draw data from our legacy and modern systems. This combination of data and technology empowers those who wield it to cut through illusion and restore justice.

Let's talk about something that's been hiding in plain sight in insurance and healthcare operations for the better part of three decades: You have no idea what your processes are actually doing.

I don't mean that as an insult. I mean it as a structural observation. You have dashboards—God, do you have dashboards. Gorgeous ones with KPI tiles and sparklines trending whichever way the builder needed them to trend. You have reporting teams producing decks for Monday standups—assemblies of data that's six weeks old, filtered through three layers of organizational telephone, and crafted—not maliciously, but inevitably—to support a story someone already believed.

What you mostly don't have is instrumentation. Real instrumentation. The kind that tells you, in something close to real time, what your core processes are producing, where they're breaking, and what that's costing you.

That gap is about to get much more expensive to ignore.

Process excellence folks will recognize DMAIC—Define, Measure, Analyze, Improve, Control. The problem is that in most operations, the M and the A have always been the expensive, politically fraught parts. So organizations Define—sometimes brilliantly—and then Jump. Straight to Improve. They hire consultants, run workshops, launch initiatives, celebrate launches. A year later, they do it again. That isn't improvement. It's expensive thrash—innovation theater in a process‑excellence costume.

Instrumentation was always theoretically worth it. It just never made it to the top of the list.

Enter AI, which changes this calculation in two ways—one a mandate, one a gift.

The mandate first, because it's the one that gets you fired.

You can't drop operational AI into a live process environment without knowing precisely what it's doing. AI systems in claims processing, prior authorization, utilization management—these make decisions at a speed and scale no human team can realistically audit afterward. If you don't have instrumentation showing, in near‑real‑time, what your models are producing, where they're drifting, and where edge cases are piling up into systematic errors, you'll have a very bad day. Possibly a regulatory very bad day. Possibly a front‑page very bad day.

Operational AI forces the instrumentation conversation in a way Six Sigma consultants never could.

Now the gift.

AI also makes instrumentation cheaper and easier than it's ever been. Process‑mining tools can map your actual workflows—not the idealized Visio diagram, but what's really happening—by reading keystrokes, logs and system events that already exist. Natural language processing (NLP) can monitor unstructured outputs: call transcripts, clinical notes, adjuster comments, member complaints. Modern data pipelines can connect legacy systems in a fraction of the former time and cost. All without creating risk or dependencies.

By instrumenting your operation for AI, you end up using AI to measure what you should have been measuring all along. The mandate and the gift are the same. You don't get the AI transformation without building the measurement infrastructure—and once you've built it, you finally have something most organizations have never possessed: a real‑time picture of their own operations.

The counterintuitive part nobody talks about: people assume a fully instrumented, heavily automated operation becomes robotic. Soulless.

The opposite is true.

When 80% of your operation runs smoothly—instrumented, measured, automated, in control—something remarkable happens to your meetings. The variance archaeology, the defensive explaining, the "why did this metric move?" inquisitions—all move into dashboards that don't need a room full of people to interpret. What's left in your daily standup are exceptions. Real exceptions. The claim that fell outside every parameter. The member experience that defied categorization.

Exceptions are where operations learn. They're where customer‑service stories live—the quietly devastating and the genuinely remarkable—and those stories, surfaced in a room of engaged humans, are where innovation happens. Not in workshops or hackathons, but in noticing an exception, connecting it to context, and realizing it points to something structural.

The daily meeting becomes tactical again—focused on real issues, resolved quickly, without drifting into philosophical fog. Strategy moves to the quarterly business review, where it belongs. Mixing the daily and the quarterly is how organizations end up doing neither well.

The even better news is that a genuinely well‑run operation—one that knows what it's doing, measures what matters, and improves based on evidence—can deliver on a real mission. Instrumentation isn't separate from culture; it's the infrastructure culture runs on.

The more automated your operation becomes, the more human it can afford to be.

The instrumentation imperative is real, and AI is making it urgent. The organizations that win will be the ones that treat it not as compliance, but as what they should have built 20 years ago—finally achievable, finally affordable, and harder to ignore every quarter they wait.

Paul Carroll

AI is everywhere in insurance right now. Where do you see it being used especially well?

Dr. Michael Bewley

One mature application is the use of something called supervised machine learning, for aerial imagery. The application provides a way of getting reliable recognition of objects and images, which can be really informative about a property. Then you can use what you see in trusted frameworks. You know, given the roof had large patches of rusting or missing shingles or a hole in it before the event, what's the likelihood of damage in the event? That can be modeled in a pretty clean way.

But there's a whole spectrum of AI from really quantifiable, reliable, and well-understood systems all the way through to things where it's all about creativity. You throw in an idea, and it comes up with some more ideas. 

Even traditional risk modeling can be seen as AI. You're trying to predict the likelihood of claims.

Paul Carroll

What are the risks associated with using AI?

Dr. Michael Bewley

You've got to get the newly available data and realize it's amazing but then apply it carefully, because all data comes with uncertainty. Even if we're really confident it's a solar panel on that roof, we'll tell customers we're 98% confident. There's a 2% chance we're wrong, and saying so allows insurers to treat the data in a more nuanced way.

Paul Carroll

There's growing pushback against AI-based property assessments. People are told they have a roof problem through AI and aerial surveillance, and while they may acknowledge the issue is real, they resist being charged based on that information. How do you address this customer trust challenge?

Dr. Michael Bewley

That we can determine a roof's condition remotely is really valuable—not just to the insurer, but to the insured. Not many people climb on their roof on a regular basis. The fact that we can not only say there's an issue with that roof, but we can show the image it comes from is really important.

If we can tell someone their roof is damaged, they can fix it. They can reduce their risk, and that's in everyone's interest.

Paul Carroll

Organizations are dealing with information in countless different forms—one insurance system had 37 ways that San Francisco was described, from San Francisco to San Fran, SF, Frisco and so on. Is this data uncertainty part of the reason property-related decisions are still so difficult to make?

Dr. Michael Bewley

Just having so much more data today doesn't necessarily make for good decisions in and of itself. There are so many questionable sources of information out there, and there are so many sources where it's unclear how accurate they are because you can't actually see the provenance. It's very difficult to ascribe a level of trust.

This is why we've hinged our whole strategy on aerial imagery. We bring in third-party data and other information, but the core is what your eyes can see.

Insurers are being bombarded by a huge range of information from different vendors and open information out there on the Web. So we're very particular about how we form our information, and we make that transparent to the user. Every bit of data that we serve up in our APIs comes with a link so you can go and look at the photo. 

It's well-articulated information that matters. Volume can actually be a detractor because you get lost in the noise.

Paul Carroll

The insurance industry has historically moved slowly, but in catastrophe response, speed is critical. Where do we speed things up?

Dr. Michael Bewley

The challenge is that a catastrophe is a continually unraveling scenario. It's not just that the cat event occurs, then we're done, and we all move on. The hurricane makes landfall, properties get damaged, the storm keeps moving, further events occur, there are recovery efforts, and so on. So while speed is good, clarity is important, as well. 

If there's an event that we're going to capture with our cameras, we'll get a plane up in the air as soon as it's safe. As soon as we capture some valid imagery, we turn it around as fast as we can, using AI. In Hurricane Milton, I think we flew over 100 flights because there were so many things going on—the weather changes, what's going on on the ground changes.

Paul Carroll

Would you talk a bit more about how insurance can move from the traditional repair-and-replace model to a Predict & Prevent approach?

Dr. Michael Bewley

That's a great question. If we step back from the catastrophe-specific discussion, our regular capture program covers most well-populated areas multiple times a year. We’ve done this for a decade now in the U.S. and 18 years in Australia.

The regular uptake of imagery, year in, year out, shows you where things are today and where they've been historically, and then captures an event in that context. A really good example is our new roof edge product. We've run AI on stupendous quantities of imagery. We've looked at our full imagery archive in the U.S. and run every single house on every single historical date to work out when a new roof got put in. If an event is coming up, you can start to feed that into an understanding of whether the roof is getting to end of life anyway, so maybe it's time to replace it. Maybe that reduces the risk. You can have a mature discussion between the insured and the insurer about that. 

The exact same imagery is being used by insurers, by local governments, by construction, by town planning, by environmental groups, by so many different sorts of people. So they can have discussions about how to remediate the risks on a property before an event happens. We can talk about how we plan towns better. It's wonderful if we can all look at that same source of truth.

Paul Carroll 

What is one challenge you'd like to offer to insurers about their assumptions on property risk? What are they missing that they should understand?

Dr. Michael Bewley

I think the challenge is really for them to understand that there are new. high-quality sources of information available. They may be used to doing things a certain way with limited information, so they have to understand the incoming information and make good use of it.

In the AI space, the challenge is sifting the signal from the noise. There is genuinely a bunch of AI stuff, particularly the stuff that's in the media a lot, that one needs to treat very carefully. All the large language models and Gen AI imagery stuff—there is a place for that in insurance, but it's different from the more tried-and-tested machine learning approaches, and we have to weave that in carefully. It's very important to understand the full tapestry of AI solutions that there are and not to get them muddled up. 

Gen AI opened up a new world. It is absolutely revolutionary. I think it's on the level of the internet being invented or the personal computer. So you definitely don't want to sit by and say, "Well, I'll wait and see what happens," or "This one's not for me." You've got to get involved.

But as with the personal computer and the internet coming online, there's uncertainty about how to use it. There's uncertainty about what the impact will be. You just have to get in there and get involved. But you have to do it with wisdom and care.

Paul Carroll

Yeah, I think we've just scratched the surface. This is quite a ride we're on.

About Dr. Michael Bewley

Dr. Michael Bewley’s passion for AI began in 2007. Graduating with degrees in electrical engineering and physics (University of Sydney), he received the University Medal for using machine learning (ML) on brain scans to detect Alzheimer’s disease. He joined Cochlear to work on implantable hearing solutions, also implementing its first customer-use product analytics.  A sea-change led to a PhD program at the Australian Centre for Field Robotics, using ML to interpret sea-floor imagery from autonomous submersible surveys. He also established a data science team as Lead Data Scientist at the Commonwealth Bank.  Mike joined Nearmap in 2017 and is now VP of AI & Computer Vision, leading the development of AI technology, applying petabyte-scale deep learning on geospatial imagery and AI data sets.

Subpoenas are a routine part of claims investigations, coverage disputes and regulatory inquiries. What isn't routine is the pace at which they're arriving.

A new analysis from Wolters Kluwer CT found that U.S. subpoena volumes reached 498,000 in 2025, with growth accelerating year-over-year. After a brief 3% dip in 2020, volumes have climbed every year since, growing 13% in 2023, 10% in 2024 and 8% in 2025. Insurance is absorbing more of that increase than any other sector.

Insurance-related subpoenas grew 65% between 2019 and 2025, making them the fastest-growing category in the data. Roughly 80% of that activity is concentrated in California, Florida, Georgia and Texas. Each of these jurisdictions has its own combination of regulatory activity, litigation trends and natural disaster exposure fueling the increase.

Florida saw the sharpest increase of any state, with volumes up 86% since 2019. Hurricane claim investigations, growing demand for insurance-related records and a wave of litigation filed ahead of major tort reform measures all contributed. Much of this reflects heightened scrutiny from state regulators examining how insurers handle claims in disaster-affected areas, which generates a downstream surge in records requests and legal process activity. California volumes rose 54%, driven by insurance coverage disputes, surplus line insurer activity and new privacy compliance requirements. California's evolving regulatory landscape, particularly around data access and consumer protection, has expanded the scope of what gets subpoenaed and how quickly insurers are expected to respond.

The implications extend beyond claims departments into insurance legal operations and compliance.

How intake processes need to change

For most insurers, subpoena intake and response processes were built for a different volume environment. Many still rely on manual workflows to receive, triage and route incoming legal documents across departments and jurisdictions. Gradual increases are manageable. A 65% jump in six years exposes the limits of processes that were never designed for this pace.

Missed response deadlines create legal exposure. Misdirected documents delay claims resolution. Inconsistent handling across state lines introduces compliance risk, particularly for multi-state insurers navigating different procedural requirements in each jurisdiction. The operational cost of getting it wrong is compounding as volumes climb.

Jurisdictional complexity adds to the burden

The geographic concentration of subpoena growth creates a particular challenge for insurers that operate across states. Multi-state insurers are managing higher volumes under different rules, different timelines and different regulatory expectations in each jurisdiction.

With roughly 80% of insurance-related subpoena activity concentrated in four states, organizations with significant exposure in Florida, California, Georgia and Texas face a disproportionate operational burden. The resource allocation models and response frameworks that worked five years ago may no longer be adequate for today's volume and complexity.

What insurers should do now

The subpoena data points to a broader reality about litigation complexity that extends beyond any single sector. Regulatory scrutiny is increasing, data access expectations are broadening and legal activity in key sectors is accelerating. These are structural trends, not temporary spikes.

Insurers managing legal process intake through fragmented, manual systems are absorbing unnecessary risk. The organizations best positioned to handle this environment are the ones treating legal process management as an operational discipline rather than an administrative afterthought.

That means evaluating how subpoenas and other legal documents are received, tracked and routed across the organization. It means understanding jurisdictional requirements at a granular level and building response protocols that account for the specific procedural obligations in high-volume states. Subpoena volume trends also signal where litigation and regulatory activity are heading, which should inform how insurers staff and structure their legal process operations.

If these trends hold, the gap between current legal process volumes and most insurers' capacity to manage them will only widen. The question for insurers is whether their legal operations are built for the volume they're handling today or the volume they were handling five years ago.