The most startling article I've read in a very long time ran in The Atlantic last week, about euthanasia in Canada. The headline reads, "Canada Is Killing Itself." The deck headline says, "The country gave its citizens the right to die. Doctors are struggling to keep up with demand."

Canada's parliament legalized euthanasia in 2016, and the article says it now "accounts for about one in 20 deaths in Canada—more than Alzheimer’s and diabetes combined."

Perhaps you knew that fact, but I certainly did not, and I suspect a high percentage of you who don't live in Canada were also unaware of that trend.

In any case, it raises all sorts of moral and philosophical questions, probably moreso for those of us who've watched elderly parents gradually die. The trend will also have implications for insurers, notably life and health insurers. Those implications could grow, too, if Canada turns out to be a bellwether, and the trend spreads to other, even more populous countries.

I thought I should share. 

The Atlantic article says:

"It is too soon to call euthanasia a lifestyle option in Canada, but from the outset it has proved a case study in momentum. MAID [short for Medical Assistance in Dying] began as a practice limited to gravely ill patients who were already at the end of life. The law was then expanded to include people who were suffering from serious medical conditions but not facing imminent death. In two years, MAID will be made available to those suffering only from mental illness. Parliament has also recommended granting access to minors."

What's happening in Canada is not just an increase in number but in kind, because the country is allowing what's known as active euthanasia, not just passive euthanasia or assisted suicide. In active euthanasia, a doctor administers drugs that will end someone's life. In passive euthanasia, a doctor ends attempts to keep a terminally ill person alive, while assisted suicide is helping someone end their own life.

A few smaller countries—the Netherlands, Belgium and Luxembourg—have allowed active euthanasia since the 2000s, and a few more populous countries (Colombia and Spain, as well as Canada) have allowed the practice during the past decade. More countries and a dozen U.S. states, including California and Texas, allow some form of passive euthanasia or assisted suicide.

The potential implications for insurers seem pretty clear, especially if access to active euthanasia spreads. Life insurers will pay out on policies earlier than they would have. They will also pay out on more term life policies, the vast majority of which now lapse without a claim. By contrast, health insurers, whether private or governmental, will see claims decline. Studies vary but tend to find that 10% to 12% of healthcare costs come in the last year of life, and decisions by people to end their lives will presumably obviate many of those expenditures.

Euthanasia is a thorny subject that gets into all sorts of questions about the sanctity of life. The final years of an aging relative can also stir up bitter family arguments, some based on hard feelings that had lain dormant for decades. While my seven siblings, our parents and I were all aligned about care decisions leading up to my father's death at 82 and my mother's at 93, I've seen ferocious arguments erupt in some in-laws' and friends' families in the final weeks and months of a dear one's life.

While the insurance issues seem clear and quantifiable, if you've wrestled with the broader questions, as I have, I very much encourage you to read the piece in The Atlantic. It digs deeply into the tough stuff.

I'll just add one observation from someone who led a large, national organization, speaking to a group of healthcare CEOs. He said he'd found that a lot of the fighting over the care of dying parents occurred because of a misunderstanding. The parents felt they needed to hang on for the sake of their children, while the children felt they owed it to their parents to prolong their lives as long as possible.

He added, with chagrin, that he and his siblings had just made that sort of mistake as their father died. They had authorized something like $250,000 of care, which had lengthened their father's life by a week. During that week, their father was rarely conscious and, when he was, was incoherent and seemingly in pain.

Perhaps Canada, if it does nothing else, will bring issues about end-of-life care to the surface in ways that will allow for better conversations, in time to make a difference.  

Cheers,

Paul

Insurance fraud is caught in an endless game of cat and mouse. When fraudsters up the game, insurers get smarter about detection. But now, fraud is outpacing detection capabilities at an alarming rate. While carriers cling to conventional detection methods, fraudsters are already exploiting AI and advanced analytics to orchestrate undetectable schemes. The window for insurers to get ahead of the problem is rapidly closing.

To help carriers better understand the current impact of fraud and how it might evolve in the coming years, my team compiled the Online Fraud Insights report. By combining both quantitative data analysis and qualitative insights from our subject matter experts, we identified trends, patterns, and anomalies in fraudulent behavior related to injury claims, helping insurers evolve alongside tech.

Drivers of Insurance Fraud

When examining regional hotspots for fraud, we identified multiple demographic trends in the data:

● Higher Population, More Fraud: Cities and states with a high population density were more likely to fall victim to fraud. States like New York and Massachusetts ranked high for insurance fraud per capita, alongside cities such as Los Angeles and Houston. An increase in population comes with a higher likelihood of urban fraud rings that orchestrate staged accidents and fraudulent injury claims.

● Tourism Creates Fraud Hotspots: We identified high levels of fraud in tourism hubs like Nevada, Florida, and Louisiana. These vacation spots experience a constant influx of visitors, gig workers, and seasonal residents, leading to an increase in fraudulent injury claims related to slip-and-fall accidents, auto accidents, and workers' compensation claims.

Along with demographic indicators of insurance fraud, there were also legal environments that contributed to certain states having higher rates of insurance fraud.

● Legal Loopholes for Fraud: Certain states, such as Florida, Georgia, and South Carolina, have more claimant-friendly legal frameworks, making it easier for claimants to defraud their insurance companies by navigating loopholes in the laws. Additionally, states like New York and Florida have no-fault auto insurance laws, which require insurance companies to pay medical claims regardless of who caused the accident. This system is frequently exploited, creating an environment for fraudulent injury schemes, where doctors and attorneys collaborate to inflate medical expenses, over-treat injuries, or create fraudulent medical documentation.

● Inflated Claims from Attorney-Heavy Markets: More generally, the presence of strong personal injury attorney markets leads to higher levels of fraud. Active personal injury attorney markets, where individuals are encouraged to file claims through aggressive advertising campaigns, can be found in states like Connecticut, Florida, Georgia, and South Carolina. Some law firms even go so far as to coach claimants on how to maximize settlements. This can lead to exaggeration or outright fraud, often without claimants realizing the consequences. Unfortunately, this benefits attorneys who get to cash in on the claim and leave their clients to face the legal consequences.

The Culprits of Fraud

No need to play the generational blame game -- our data proved that fraud was committed by adults of all ages. Still, each cohort had unique attributes that carriers should be aware of as they analyze insurance fraud.

As the youngest generation in the data, Gen Z was far more likely to trigger physical activity and unlawful activity-related flags than people over 35. Younger adults are more prone to riskier behaviors and are sucked in by cultural pressure to overshare on social media, exposing their fraud online. For this reason, it's no surprise that claimants ages 18-25 ranked No. 1 for fraud found on Instagram and TikTok.

In the thick of their career life, Millennials were most likely to be identified for workplace-related fraud. Claimants aged 25-44 were most likely to be flagged due to their association with a business and had the highest percentage of fraud found because of activity on LinkedIn. Gen X fraudsters demonstrated similar behavior to their millennial peers, but with a higher income and more flexibility, this group was slightly more likely to be flagged for travel-related fraudulent activity. In general, those over 35 are more likely to publicly discuss their illness, pain, or injuries. Sometimes, this reinforces their claims, and other times, directly contradicts them.

Unsurprisingly, Facebook was the outlet of choice for Baby Boomers, who had the highest percentage (80%) of fraud found through the platform. Claimants over 65 also had the highest rates of fatality flags, possibly pointing to intentional or unintentional casualty fraud.

As we examined the lines of business most likely to be flagged as fraudulent, auto and workers' compensation claims accounted for the majority of flagged alerts in terms of raw volume. Even so, the highest referral rate to SIU teams came from disability claims, with 8.9% of monitored cases surfacing strong enough contradictions to warrant further investigation. Fraud alerts in disability lines are found mostly through a claimant's association with another business, suggesting claimants are reporting injuries from side gigs or personal businesses through their employer.

Insurance Fraud Predictions

Over the next five years, we predict fraudsters seeking larger payouts will become more creative. Insurers can expect an increase in sophisticated deception tactics, designed and optimized using artificial intelligence.

In the coming years, we anticipate a renewed focus on the regulatory environment in the U.S. This will shift policy toward insurer collaboration, an essential tool to increase fraud prevention rates. We also predict that insurers will increasingly use AI-driven risk assessments to identify high-risk claims, fighting fire with fire with more scalable, efficient operations. By changing the regulatory environment and adopting advanced detection strategies, carriers can effectively stay ahead of emerging fraud.

Recently, after 42 years on the job, Laura Whitman closed her office door at a regional insurance provider in Ohio for the last time. She was the go-to expert on complex risk underwriting, knowledge no internal system could fully capture. Her company hosted a retirement party. But they still haven't filled her role.

The insurance industry has known this was coming. Its workforce has been aging for decades, and now, the wave of retirements is here. Nearly 400,000 insurance professionals like Laura are expected to leave the field by 2026. That's a staggering loss of institutional knowledge, capacity, and experience.

But younger workers aren't exactly lining up to take their place. According to a recent report, 67% of Gen Z consider insurance boring, and fewer than one in three find it appealing—ranking it last among 12 industries. That perception gap matters. The field is shifting fast toward roles in data analytics, cybersecurity, and AI, and new skills and fresh talent are critical to keep up.

Fixing this isn't just about posting more jobs or offering bigger signing bonuses. It requires a fundamental shift in how insurers attract, develop, and grow the next generation of talent.

Why early career talent isn't seeing insurance (yet)

Traditional recruiting in insurance has long relied on job boards touting stability, campus visits to business schools, and internal referrals. That approach worked when talent flowed in passively. Today, it's invisible to the people insurers most need to reach.

Gen Z, on track to become the largest generation in the workforce by 2035, wants careers that offer purpose, growth, and social connection. Few associate those traits with insurance. At the same time, insurers need talent skilled in AI, cybersecurity, and analytics, roles that young professionals may not even realize exist in the sector.

It's more than a branding problem. Many organizations still center recruiting on external hiring for narrowly defined roles, without showing candidates where they can grow. That limits who applies and what they imagine possible once they're in the door.

From filling jobs to building careers

To compete for talent, insurers need to stop filling vacancies and start building visible, viable careers. That begins by showing candidates where the industry is going, not just where it's been, and by backing that up with real investments in internal mobility, skill-building, and development support.

If traditional recruiting isn't delivering the workforce insurers need, then the next step is clear: attract new talent by showing what's possible, then grow it from within. Many frontline and mid-career employees already understand the business and the customer. What they need is a clear, supported path forward.

Ultimately, the wave of retirements goes beyond a hiring challenge. It presents an opportunity for business continuity and transformation. As experienced professionals exit, insurers have the chance to rethink how critical knowledge is captured, how modernization efforts are staffed, and how emerging talent is developed. With digital-native competitors gaining ground, those who invest now in building a resilient internal talent pipeline will lead the industry forward. Workforce development isn't just an HR initiative, it's a lever for long-term competitive advantage.

Making career paths real, and accessible

Tailored career pathways can help insurers both attract and grow talent, especially in fast-changing areas like digital claims, analytics, and cyber risk. These paths don't always require degrees to begin. Many start with certifications, licensing support, or short-form programs that build job-ready skills.

For some, they're a launchpad into a degree. For others, they offer a direct path to promotion. Think:

• Data analytics certificates for claims specialists
• Compliance credentials for service reps
• Digital tools training for marketing assistants moving into product roles

When programs are stackable, visible, and linked to advancement, they help employees see a future inside the organization and give insurers a scalable way to meet future workforce needs.

The good news? Many insurers already have structured role levels and internal promotion cultures. What's often missing is the connective tissue: visibility, support, and educational investment.

Building workforce readiness from the inside out

Workforce development only works when it's grounded in what employees actually need to grow. That means moving past one-size-fits-all training toward people-centered development aligned with business priorities.

Start with your people. Some employees want to move up. Others want to reskill into a different role. Others want to deepen where they are. The right education strategy gives them all a path.

Second, make sure these opportunities are accessible. Many of the workers who could benefit most from career development, including those in high-turnover or underserved segments, are also the least likely to have time, money, or flexibility to pursue traditional degrees. That's why programs need to be designed with real-world constraints in mind: shorter formats, stackable credentials, online options, and upfront tuition support.

Lastly, track what matters. Course completions are useful but not the whole picture. Focus on outcomes like retention, internal promotion, and movement into future-critical roles. That's where you'll see the impact.

Taken together, this kind of approach delivers on both sides. Employees get a clear, supported path to growth. Employers get a stronger, more resilient talent pipeline. And the industry starts to look less like it's falling behind and more like it's building for what's next.

Now is insurance's moment to step in

In 2025, white-collar job cuts are reshaping early-career opportunities. Industries that Gen Z rank as most appealing—tech, media, entertainment—are not offering the same volume of entry-level roles they once did.

That's where insurance can step in, as a stable, future-facing industry with the potential to offer real career paths, not just jobs.

The cost of doing nothing is steep. If roles like Laura's go unfilled, or are filled without the right capabilities, insurers risk more than backlogged claims. They risk customer churn, compliance failures, and stalled digital transformation. According to McKinsey, organizations that invest in skills transformation are up to 2.5 time more likely to succeed in transformation initiatives. Conversely, the loss of institutional knowledge can cost companies millions in operational inefficiencies, rework, and delayed innovation.

The coming retirement wave is more than a moment of transition. It offers a chance to reshape what a career in insurance looks like. By focusing on upward mobility and building accessible, future-aligned learning pathways, insurers can attract the talent they need and finally tell a story that resonates with the next generation.

Geopolitical unrest, inflationary pressure, supply chain disruption, and natural disasters. What does this mean? It means the insurance industry is in polycrisis. However, none of this is new to the industry. What is new is the sheer volume of information and noise the industry is bombarded with daily. This distorted signal-to-noise ratio is making risks feel bigger, newer, and more threatening than they may be.

During insurtech Send's INFUSE webinar, I had the opportunity to share my thoughts about the current polycrisis in the insurance industry. While my fellow panelists called for a unique approach and deeper partnerships with brokers to handle connected risks, I challenged conventional thinking about risk in 2025.

Is there a polycrisis, or is it just a perception?

There is no denying that we are seeing a severe increase in connected risks. However, the term "polycrisis" doesn't need to be intimidating to a world that feels increasingly uncertain. We've been here before, just not with the same amplification. Climate, inflation, and supply chain risks have always been part of our industry's history. What has changed is the velocity of news cycles and the way anxiety spreads faster than fact. During the webinar, I presented data on hurricane frequency. The diagram below shows no significant rise in hurricanes over time, even as public perception suggests otherwise. This is why, for insurers, panic doesn't help; perspective does.

Innovation is critical; so is discipline

Still, we cannot ignore today's connected risks. They are complex and always evolving, and industry should, too, in response. Innovation is critical. My fellow panelist Neal Croft suggested that insurers must embrace continuous, real-time underwriting powered by high-quality data, scenario modeling, and adaptive decision-making tools. While I agree, I also believe we should be cautious not to discard the fundamentals of underwriting for the sake of transformation. Before we declare traditional models obsolete, we must ask ourselves: Are they truly broken, or are we not using them properly in a modern context?

Crisis in insurance means new opportunities

At Cross Cover, we've seen firsthand that elevated risk creates need, need creates innovation, and innovation creates business. Many of the most resilient and profitable firms I've encountered were born or transformed during turbulent periods.

Today's polycrisis is no different. Whether it's using parametric products to close protection gaps, deploying smarter data to assess climate risk, or simply partnering more deeply with brokers and clients, there is no shortage of ways to turn complexity into commercial advantage.

Filtering Out the Noise

As we discussed during the INFUSE webinar, the biggest risk isn't complexity; it's losing our grip on what's noise versus what's a true signal. We must:

• Challenge narratives that push fear instead of facts.
• Check our biases when reviewing risk data.
• Tune out the noise and focus on actionable signals.

Markets are resilient. So are insurers. Our job isn't to predict the next crisis perfectly; it's to respond with clarity, courage, and composure when it arrives. There's no doubt the insurance industry is standing at an inflection point. But we don't need to overreact. We need to recalibrate. Yes, embrace innovation. Yes, evolve underwriting. But do it with a clear head. We're not new to crises; we're built for them. Let's never forget that.

Vertical AI agents, tailored for industries like insurance, healthcare, etc., are redefining how organizations handle complex, open-ended, and multi-step workflows. 

By combining domain-specific knowledge graphs, advanced retrieval methods such as GraphRAG, and orchestrator-worker agentic patterns, these agents can reason, decide, and act with transparency. When paired with hardware such as AR/VR devices, SmartGlass, telematics, and wearables, they enable contextual understanding of the environment and real-time decision support. 

This article explores why such agents are well-suited for scenarios such as risk assessment and long-term care adjudication.

Problem Statement: What?

Insurers have often leveraged robotic process automation (RPA) for efficiency, and this approach has reached a saturation point. Moreover, risk postures and customer needs are changing dynamically, and insurers require capabilities to improve customer experience and provide differentiated services. In this context, let us consider the following two scenarios.

Long-term care (LTC) claims - With the rise in U.S. aging population and shortage of skilled workers such as registered nurses (RN), licensed practical nurses (LPN), certified nursing assistants (CNA), etc., the burden on nursing staff increases in terms of documentation of care and services with respect to activities of daily living (ADLs), evaluation and approval of care plans, health status, etc. This information is critical for adjusters to evaluate policyholders' eligibility and medical assessments, along with reports from facilities and monitoring devices, all while complying with HIPAA and insurance regulations.

Auto-insurance claims – With advancements in vehicle technology such as advanced driver assistance systems (ADAS), software-defined vehicles (SDV), telematics, etc., there is also increased need to assess behavioral patterns, dynamic risk posture, liability, and fraud risks.

These workflows demonstrate the following traits:

• Heterogeneous data - comprising structured data such as policy and activities of daily living (ADL) scores, and unstructured data such as provider notes, reports, and images/videos (facility inspections, etc.).
• Domain-specific logic – requiring specialized skill sets to reason and infer care eligibility criteria, ontologies, risk posture, policy clauses, and regulatory compliance.

Solution - Agentic AI: Why?

The dynamic and variable nature of the above workflows requires knowledge workers with the ability to understand the context to plan, act, and reflect on the chosen path. This complex task is best suited for agentic AI (orchestrator-worker agentic pattern), as it is an open-ended problem that needs specialized skills to learn and adapt to the environment. At a high level, it involves the following components:

• An orchestrator to analyze the request, plan and decompose tasks into sub-tasks/workers, orchestrate the workflow, and synthesize the results.
• Workers/specialized agents – such as wearable worker to process mobility data and generate structured events, telematics worker to normalize trip/vehicle signals into standardized event schema, knowledge graph query agent to map intents to graph queries – patients' history, claims history, prior similar incidents, etc.
• Knowledge graph and graph database – canonical domain model, enabling reasoning and inference. Schema: ontology (OWL/SHACL), nodes for entities (patient, policy, device, trip, event), edges for relations (caused_by, observed_at, claimed_in)
• RAG + LLM service – assemble knowledge graph-grounded context and retrieve documents from vector database to produce answer/plan

Long-Term Care Claims Adjudication: How?

Consider a scenario where a policyholder files a claim for in-home care services. Wearable devices track their mobility and heart rate patterns, and a care provider uploads ADL assessment forms and daily care logs.

The following is the conceptual flow:

1. Wearable device/event triggers a request, wherein wearable data worker ingests mobility and physiological data, runs on-device pre-processing to compute ADL scores, and flags anomalies (e.g., sudden decline in mobility).

2. Data is transmitted securely to the orchestrator, where it analyzes and routes to appropriate workers such as:

• Policy knowledge graph query agent to match ADL scores against benefit triggers in LTC policy
• Document worker to parse care provider notes for evidence supporting claim eligibility

3. Knowledge graph encodes the domain ontology (policies, events, claims, etc.) and GraphRAG starts with semantic retrieval, then expands context through relationships for multi-step reasoning to ensure outputs are grounded on facts.

4. Compliance worker validates whether the recommendation meets both insurer policy and local regulatory guidelines.

5. Orchestrator then synthesizes the results/decision and sends to a claims adjuster for review and approval.

6. All decisions and their sources/origins are appended to knowledge graph with audit trail for regulatory/ML training.

Potential Benefits

Agents help alleviate the burden on knowledge workers by augmenting them and orchestrating and synthesizing complex processes such as claims through specialized workers. This enables delivery of the following benefits:

• Faster, evidence-driven claims processing
• Improved quality of care by synthesizing real-time/near-real-time information from edge devices, wearables, etc.
• Reduced false positives/fraud rates through correlation/identification of fraudulent rings
• Grounded response that is explainable and traceable to improve trust

The Way forward

Graph-based retrieval, structured communication protocol, IoT/wearables/edge devices, and multi-agent orchestration are converging into a practical toolkit for industry-specific AI.

For organizations to scale responsibly, the path forward is:

• Problem awareness and choosing the right high-impact tasks/processes for Agentic AI. It is not a silver bullet for all problems. Tasks that are repeatable, generic/not specialized with fixed/pre-defined paths, etc., are not ideal candidates for Agentic AI to justify the ROI.
• Define and embrace domain ontology to capture knowledge in knowledge graphs to power the LLMs with grounded context.
• Implement GraphRAG retrieval with provenance support. This ensures transparency, accountability, and trustworthiness in decision making.
• Iterate and integrate with environmental data such as wearables, facilities, provider networks, etc.
• Iterate by adding specialized workers as the workflow expands or needs change.
• Measure the outcomes to demonstrate the value and recalibrate/adapt to changing needs.

Vertical AI agents are no longer a research-only concept. If harnessed at the appropriate value chain, the power and benefits they unleash will be a game changer for any industry.

For years, the dominant question in insurance IT has been: Which option, buy or build, will cause fewer headaches? But as software capabilities become more commoditized, and as AI and low-code tools put more power directly in the hands of teams, a simpler question is emerging:

Who can get this done?

The tools and capabilities all work. Everyone has application programming interfaces (APIs) you can connect to. However, while it's possible to stand up a customized experience for a new product in a matter of weeks, shipping something truly integrated, flexible, and usable is a different story.

And here's the truth: Either a product fits you, or you need to fit the product.

Most packaged software products assume the latter. You're expected to adapt to their processes, compromise your priorities, and sacrifice competitive edge just to match their system. But no single technology platform fits every insurer. Forcing fit often means slowing down or giving up what makes you different.

The real challenge is less about choosing the right technology and solutions; it's how they get delivered.

And delivery only works if you have the people to do it. Insurance talent is aging out. A significant portion of the insurance workforce is nearing retirement, and with it goes deep domain knowledge of both the systems and logic behind them. As these experienced professionals exit, many insurers are left with delivery and IT gaps that they can't easily fill.

Where Execution Breaks Down

Insurers and MGAs consistently voice the same frustrations:

"We didn't realize how many things would be out of scope."

"We're paying for change requests for things we thought were standard."

"The vendor team we thought we'd be working with disappears once we sign the contract."

Too often, delivery is handed off to offshore teams who were never part of planning and won't be around for iteration. Or it's handed over to internal IT groups already stretched across 10 priorities. Or, in the worst case, it's controlled by product vendors who monetize lack of flexibility, turning every change into a new contract.

This is what happens when product and delivery get disconnected: Timelines slip, no one owns the outcome, and every small change turns into a negotiation. It's not always a technology problem. It's an execution and fit problem.

What Execution-First Looks Like

Execution-first delivery means your partner goes beyond standing up a platform; they work with your team to engineer outcomes. More than just handing over a minimum viable product (MVP) that checks the boxes, think about how to build processes where your partner can stay close to the business, adapting in real time, and delivering something that actually works in the field.

In the best cases, that looks like:

• Engineers who understand the business logic
• Iteration without change request delays
• The same team staying accountable post-launch

Execution-first partners are those who stay close to the work and ship with you, not ones who add layers between planning and delivery.

Insurers that succeed do so not just because they chose the right platform. They built the right team and partners around it. Execution is what determines whether your road map turns into reality, or a backlog of change requests and mounting frustration.

The real question is no longer what to choose. It is who can deliver.

Insurance, like many industries, is in full sprint toward artificial intelligence. Conferences are packed with AI demos. Strategy decks are flooded with automation goals. And boardrooms are asking "Why not AI?" 

Instead, many might want to ask, "Why now?" or even, "Are we ready?"

From where I sit, too many carriers are chasing AI before they've laid the groundwork. They're eager to run with advanced tools, but their operations are still learning to walk.

Let's be clear: AI and automation are powerful upgrades, like switching from hand tools to power tools when building a house. But it doesn't matter how advanced your tools are if the foundation is cracked. In the same way, if your core systems are fragmented, inefficient, or poorly integrated, AI won't fix them. It will just amplify what's broken.

The Risk of Skipping the Basics

I've seen this happen more than once. A carrier gets excited about AI-powered underwriting or virtual claims assistants. They invest in the tech, build a team and expect results. Immediately. But within months, the project stalls. Budgets balloon. Stakeholders lose faith. Or worse, the tool works, but they produce outputs that are unusable because the surrounding systems aren't connected.

AI doesn't work in isolation. It needs clean, structured, reliable data. It needs integrated workflows. It needs clear visibility into the customer journey. And it needs all of that before you turn on your first model.

Too many insurers are trying to build a smart home, installing smart bulbs, thermostats, and locks, without fixing the faulty wiring behind the walls. Layering AI on top of outdated systems, manual workarounds and siloed data means you're not innovating. You're firefighting.

What Are the Facts?

According to a 2024 Deloitte survey, between 70% and 80% of U.S. insurers have implemented generative AI in at least one business function, such as claims, customer service, or distribution. That aligns with broader findings that indicate that by the end of 2025, around 91% of insurance companies worldwide will have adopted some form of AI technology. Some AI-powered claims automation is already cutting processing time by as much as 70%.

But that adoption isn't without friction. According to another recent survey, 74% of insurers still rely on outdated legacy systems for critical operations like pricing, underwriting and rating.

That gap reveals the heart of the issue: Enthusiasm for AI is real and fast, but operational maturity often isn't keeping pace.

The Must-Haves Before You Automate

If you're an insurer considering AI, there are ways to implement it. Before doing that, I caution you to take a hard look at your operations first. Ask:

• Are our core systems integrated?
• Is our data clean, consistent and accessible in real time?
• Do we have automated workflows that allow AI to act, or do we still depend on email and spreadsheets to get things done?
• Can we trace and audit every customer touchpoint across systems?

If the answer is no to any of the above, AI won't help you, at least not yet. And that's not a critique on AI. It's a call to action for operational readiness.

Modernization First, Then Automation

The insurers seeing real results from AI are the ones who took the time to modernize their business first. They invested in workflow automation. They connected their systems. They focused on data quality and governance. They created operational environments that are scalable, transparent and efficient.

Only then did they start exploring AI, not as a gimmick but as an extension of the maturity they'd already built. The difference is obvious. Their projects hit milestones. Their tools integrate seamlessly. And their teams actually trust and use the outputs.

AI Can't Fix Ops, But Ops Can Make AI Work

AI adoption feels inevitable (maybe even urgent), with pressure coming from all sides. But urgency without readiness doesn't lead to progress. It leads to wasted time, money and trust. You wouldn't run a marathon if you hadn't tried a 5K first, right? The real opportunity in insurance isn't just about being early to AI, it's about being ready for it.

Operational maturity isn't glamorous. It's not as flashy as a chatbot demo or as headline-grabbing as an AI-powered claims system. But it's the difference between innovation that sticks and ideas that quietly fail.

AI will transform insurance; there's no question about it. But it's not a quick fix. It's a multiplier, not a miracle. And only those who've done the hard work of modernization will see it pay off.

Recently, third party administrators (TPAs) have experienced inorganic growth through acquisition. Acquiring new lines of business or further expanding existing lines of business has been at the core of business growth strategy for several TPAs. TPAs then attempt to reduce cost to improve business unit (BU) margin and overall profitability. While that is a viable growth strategy, TPAs often ignore an obvious growth lever – cross-selling and upselling existing clients.

The value of cross-selling and upselling has not been lost in the insurance space – indeed, carriers have used data- and analytics-driven models to cross-sell and upsell and improve productivity. Those carriers have seen revenue increases of up to 30% increase. But the cross-selling/upselling focus has historically been in the retail markets, driven by financial advisors and brokers. For TPAs, sales strategy is typically fragmented.

Putting the Pieces Together

A TPA sales and renewal strategy typically faces several key obstacles that inhibit cross-selling and upselling:

1. Lack of Financial Alignment Between Performance and Cross-Selling – Most TPAs report on margin at two levels – business unit (BU) margin and enterprise margin. As a result, BUs are given incentives to consider only their own profitability and growth, not the profitability of the enterprise. This results in missed opportunities – for example, a healthcare claims TPA missed opportunities to provide tort claims administration because the client was unaware the TPA provided the additional services.

2. Disjointed Relationship Management – Client relationships are often owned and managed at the BU level. This structure creates a challenge, as leaders of individual lines of business rarely identify or pursue opportunities to offer services from other units. While some TPAs use enterprise-level client relationship managers, these roles often lack the authority or visibility needed to serve as the primary point of contact for all client interactions.

3. Lack of Pricing Discipline – Pricing for both new business and renewals is usually a struggle for several reasons. First, TPAs tend to offer several different pricing models for services. Each of those comes with different cost assumptions, which may or may not accurately reflect the true cost of the service offered. Second, pricing may be done in a shared service model or it may be done at a BU level. At the BU level, cost assumptions may not be the same for each line of business. For example, are corporate functions (e.g., finance) included in costs underlying pricing assumptions? 

Cost allocation mistakes can lead to unprofitable underwriting. In one instance, a TPA made faulty assumptions on labor costs, where claims adjuster productivity was 33% worse than the numbers used in pricing assumptions. Additional claims personnel were needed to meet overall production requirements. The same TPA also failed to allocate enterprise costs when calculating pricing. What appeared to be a 20% margin on paper actually was a single-digit or even negative margin in reality.

These challenges can be overcome, but TPAs must understand what cross-selling and upselling opportunities can mean to both topline revenue and overall profitability. Although TPAs typically have nascent cross-selling capabilities, there is significant upside to TPAs that invest appropriately.

So what should TPAs do if they want to expand their cross-selling and upselling capabilities?

Unlocking Value Through Cross-Selling

For TPAs to unlock value through cross-selling, TPAs need to ensure they have met all the activation requirements:

1. The functional capability to cross-sell and upsell
2. Strong pricing and cost discipline
3. Organizational enablement

Functional Capability to Cross-Sell and Upsell

Cross-selling and upselling for TPAs begins with good-quality data. TPAs need to use their data to understand 1) the factors that drive a customer to purchase additional services and 2) the best timing for making the purchase. This will prove to be a significant pain point for TPAs to overcome, but this will be a critical market differentiator. TPAs often segment customers based on line of business and size of account. Understanding how these interactions and key cross-selling factors will differ based on the type of client is a major opportunity.

In addition to data components, TPAs also need to ensure that there is the necessary technology. Specifically, the use of common CRMs, customer segmentation tools, centralized reporting tools, and leveraging AI for automation purposes are all critical tools that help enable cross-selling and upselling.

Strong Pricing and Cost Discipline

Cross-selling can drive revenue, but the true value is in unlocking additional profitability, both at the BU and enterprise level. To do so, TPAs need to understand a service's cost drivers and more importantly, how cross-selling and upselling can reduce traditional costs. TPA services typically have three cost types – BU labor, BU non-labor, and enterprise/corporate costs. The graphical representation below highlights how disciplined pricing teams will incorporate these three costs (shades of blue), and then adjust pricing to the target profitability (in green).

When TPAs attempt to cross-sell or upsell, they need to ensure their pricing team has accurately included all cost variables in their quote. 

Here's an example:

If a TPA is currently providing P&C claims support to a client, and now can provide cyber claims support, the TPA has to have a strong understanding of cyber claims support pricing. If the TPA is leveraging a claims-driven pricing model (i.e., price to process X amount of claims over Y time leveraging Z individuals with certain processing assumptions), then profitability against the BU labor costs depend on how accurate the processing assumptions are for the TPA. Accuracy promotes stable margins, while inaccuracy invites cost creep. But cross-selling should inspire a reduction in BU non-labor costs. For example, marketing expenses that are typical for the cyber business should not exist in a cross-sale scenario, thus improving profitability both for the BU and the enterprise. But the opportunity does not stop there. Allocating corporate costs ensures profitability against shared resources, but it also helps to benchmark spend against revenue.

In the case of cross-selling and upselling, TPAs have the opportunity to improve the spending ratio between corporate support and the revenue generated, which should lead to improved enterprise margin, assuming scalability.

Organizational Enablement

Last, but certainly not least, TPAs need to integrate the cross-selling and upselling process into their sales dynamic. It is not uncommon to see lines of business, enterprise customer relations teams, and other teams involved in the sales and renewal processes, with significant friction as the usual result.

While the exact method by which a TPA manages these various teams to achieve cross-sale and upselling goals will vary, there are general concepts that can push the organization in the right direction.

1. Tie Financial Incentives to Cross-Sale/Upselling Metrics – TPAs should consider setting enterprise goals that highlight the importance of cross-selling/upselling. Rather than tying any particular individuals' compensation to a specific target, the enterprise could set a target that X amount of revenue is derived from selling new services to existing clients for the enterprise to be eligible for specific incentives (e.g., a minimum amount of cross-sales yields a 5% bonus toward every employee in the company).

2. Automate Where Possible – Too often, insurance carriers and distributors try to consciously engage in cross-selling, rather than trying to embed it into their sales process. For TPAs, the opportunity exists to automate the upselling and cross-selling processes, particularly for small or medium-sized employers or customers. In these situations, where renewals are likely already automated, TPAs should consider redesigning processes and responsibilities to suggest services leveraged by customers of similar size and industry.

3. Emphasize Customer-Centric Relationship Models – For larger clients, TPAs can leverage client-centric customer relationship teams to serve as single points of contact. This approach can reduce friction and encourage a "single TPA" interaction with the customer. The customer relationship management team then triages to the relevant internal teams. In these scenarios, the TPA can much more easily showcase its product shelf to customers and encourage cross-selling/upselling opportunities.

What's Next?

For TPAs, an investment into cross-selling and upselling provides an opportunity for organic growth. While the M&A market remains hot for TPAs, TPAs will eventually reach a point where inorganic growth is no longer a viable long-term strategy. Instead, TPAs will need to modify their growth goals to include customer penetration goals.

To do so, TPAs must act now to develop the organizational muscle necessary to grow this capability.

TPAs that are interested in pursuing cross-selling and upselling initiatives should:

1. Assess their readiness against operational, technical, growth strategy, and organizational dimensions.
2. Identify the factors that lead to opportunities
3. Develop data models and pilots to test these factors
4. Refine these models and roll out for broader application

Doing so will position TPAs to develop deeper relationships with their customers and with that, secure profitable clients for the future.

With about three million "18-wheeler" trucks operating in the U.S. and the demand for timely delivery of cargo, the transportation sector is often defined by its complexity and the speed at which decisions must be made. 

For freight brokers and motor carriers, arranging cargo insurance has, until recently, been a manual and time-consuming process. The value of many shipments is higher than the $100,000 basic cargo insurance coverage carried by most trucking companies. It could take a day or more to secure supplemental coverage for high-value or specialized loads. In an industry where hours can mean the difference between winning and losing a shipment, this lag is more than an inconvenience; it is a money-loser.

Recent advances in embedded insurance technology are beginning to emerge as a game-changer. In consumer transactions, embedded insurance has been commonplace for years. Examples of this are the travel insurance offered during flight bookings or ticket insurance at checkout for major sports events and concerts. But in business-to-business (B2B) transactions, embedded insurance has been limited. Now, that is shifting, and the implications for freight risk management are significant.

Moving Beyond the Consumer Model

Embedded insurance, when you break it down to basics, refers to the integration of insurance products into a broader transaction or workflow. This allows users such as freight brokers, shippers and motor truck carriers to access and bind coverage without leaving their primary platform. In the consumer world, embedded insurance is often a simple checkbox at the point of sale. In the B2B context, the complexity is greater, and the requirements for expertise are much higher.

The challenge in commercial freight has been twofold. First, there is the need for custom coverage based on shipment specifics such as the goods being transported, the value of the load and the points of origin and destination. Second, there has been a lack of digital infrastructure to support instant quoting and policy issuance at the moment a load is arranged. As a result, many freight brokers and motor carriers have relied on time-consuming manual processes such as emails and phone calls. This causes friction, delays and sometimes lost business opportunities.

A Case Study in B2B Embedded Insurance

The sort of system developed at Logistiq Insurance Solutions, called Freight Insurance Fast, addresses these challenges. The program runs on a software application with an application programming interface (API) designed to integrate directly into transportation management systems (TMS), load boards, and other logistics platforms. This enables freight brokers and motor carriers to access high-quality shipper's interest policies instantly, precisely at the point in the workflow when insurance is most often required, which is when the details of a load are being finalized.

Early adopters have seen immediate benefits. Users can secure supplemental cargo coverage in seconds rather than hours or days. This reduces manual administrative steps and keeps shipments moving. This timeliness is especially valuable because rising insurance premiums have led many motor carriers to lower their standard ($100,000 per load) coverage limits, making spot insurance a necessity for higher-value loads. Verisk CargoNet reports 3,625 cargo theft incidents in 2024, a 27% increase from 2023, with the average loss climbing to approximately $202,000 per theft, underscoring the need for adequate insurance on high-value loads.

Efficiency, Transparency, and the Role of Expertise

The integration of embedded insurance into freight workflows does more than save time. It also creates a more transparent and informed decision-making process. By using shipment data already entered into a TMS, the system can accurately price risk and present coverage options for the specific load. This not only reduces errors and gaps in coverage but also helps brokers and carriers make better risk management decisions without leaving their "go-to" workflows.

It's also important to note that embedded insurance in the freight sector is not only about automating transactions. The most effective embedded solutions combine digital convenience with deep insurance expertise. This ensures that users are not simply rushed to a purchase but are guided toward the right coverage for their needs.

The Road Ahead for B2B Embedded Insurance

The adoption of API-based, digitally delivered embedded insurance in B2B logistics is still in its early days, but the momentum is clear. As more TMS providers, load boards, insurance agencies, and vetting companies recognize the value of integrated insurance solutions, we can expect to see broader uptake and continued innovation in this space. For freight brokers and motor carriers, the result is a more agile, efficient, and resilient supply chain, one where risk management keeps pace with the speed of business.

As the insurance industry continues to explore the possibilities of embedded products, the freight sector offers a compelling case study in how the well-thought-out integration of insurance can help to improve operational efficiency and provide better risk outcomes. The lessons here are broadly applicable; when insurance is delivered at the right moment, in the right context, and with the right expertise, everyone from the underwriter to the end customer enjoys the benefits.

As natural perils intensify and become more unpredictable, the imperative to accurately assess, price, and mitigate risk has never been more critical. While catastrophe modeling has long been our industry's bedrock, the sheer volume of unstructured, real-time data presents both a formidable challenge and an immense opportunity. Large language models (LLMs) are poised to unlock this potential, fundamentally reshaping our approach to risk management.

The Unified Risk Picture

The true power of the LLM revolution lies not in isolated capabilities but in their synergy. Consider a widespread, sudden event, such as the 2013 Eastern Canada ice storm or a large-scale power outage in a major European city. Traditionally, assessing the full impact involves sifting through disparate sources post-event. With LLMs, initial signals—from real-time social media chatter, emergency service dispatches, and local news reports—can trigger immediate, cascading actions. The system can instantly identify affected properties, cross-reference policies for specific peril clauses (e.g., wind or ice damage), and flag potential exposures. This shifts us from a reactive, post-event data collation process to a dynamic, pre-emptive risk mapping system.

This unified view offers profound value to underwriters globally. Imagine an alert during the 2021 British Columbia atmospheric river event, or a significant flood in Central Europe. An LLM-powered system could not only identify affected properties but also pinpoint those within a portfolio with a high probability of claims based on a newly identified overland water coverage trigger. It could even enrich property data with nuanced details like "a heritage brick house, built in the 1920s, with a newly reinforced basement, surrounded by mature trees in a high-wind zone"—all extracted from unstructured notes, historical records, or public descriptions. This unified risk picture transcends traditional modeling; it becomes a living, breathing digital twin of our entire exposure, continuously updating as events unfold and new data emerges.

Going Beyond Assessment

This continuous stream of hyper-accurate data ushers in a new era of risk management. Instead of merely assessing losses after they happen, insurers can begin to mitigate them before they become catastrophic. The same LLM-driven system calculating real-time exposure could send targeted alerts to policyholders in the path of a disaster, such as the 2016 Fort McMurray wildfire or large-scale blazes in Australia or California. These alerts could offer tailored advice on securing properties before evacuation, moving beyond generic warnings to actionable, personalized guidance.

This fundamentally changes the insurer's role from a financial backstop to a genuine risk partner. It moves beyond simple risk pricing to genuine risk prevention. For agricultural clients, the system might not just assess hail risk but also provide hyper-local weather alerts combined with tailored advice on securing specific crops or equipment. This shift to pre-loss mitigation is not just about reducing claims; it's about building a more resilient society, one informed decision at a time. The strategic value of unstructured data in enhancing resilience, as highlighted by extensive research in risk management and supply chain logistics, underscores this imperative.

Challenges and the Human Imperative

Of course, this transformative vision is not without its challenges. The most significant is trust. An LLM's ability to extract nuanced insights from complex data is only as good as its underlying training and validation. In a highly regulated industry, we cannot afford for models to "hallucinate" or misinterpret critical policy clauses. The ethical implications are equally immense. The use of public and social media data, for instance, must be handled with extreme care to protect privacy and security, adhering to evolving global data protection regulations like GDPR and Canada's privacy laws.

Academic and industry literature consistently emphasize the necessity of a "human-in-the-loop" model. The future of this technology isn't about replacing human experts but augmenting them. Actuaries, underwriters, and claims adjusters will remain essential, but their roles will evolve. Instead of manually sifting through vast datasets, they will become critical thinkers and validators, leveraging LLM-generated insights to make faster, more precise, and more strategic decisions. This human oversight is the crucial final check, ensuring that while machines process the deluge of data, human expertise guides the way forward, maintaining fairness, explainability, and equitable outcomes.

A Phased Approach to Adoption

Given the transformative potential and the identified challenges, our recommendation for executive leadership is to pursue a phased, strategic adoption of LLM technology. This is not a "big bang" project but a continuous evolution, built on measured steps and clear governance.

1. Phase 1: Pilot and Validation. Initiate small, focused pilot programs to test LLMs on specific, well-defined problems. A strong starting point could be using LLMs to parse and extract key data points from a limited set of complex policy documents or to analyze drone footage from a past significant event (e.g., a regional flood or wildfire) to improve preliminary damage assessment protocols. This phase is critical for proving the concept, building internal trust, and demonstrating tangible ROI.

2. Phase 2: Integration. Once pilot successes are validated, focus on integrating LLM capabilities into existing systems. This involves building robust bridges to connect unstructured data insights from LLMs with the structured data in our core catastrophe models, policy administration systems, and claims platforms. The goal is seamless data flow and enhanced decision support.

3. Phase 3: Scale and Governance. As the technology scales across the enterprise, establish comprehensive governance frameworks. This is a non-negotiable step to ensure data integrity, address potential biases, and maintain strict compliance with all relevant regulatory standards globally. A dedicated, cross-functional team—comprising legal, compliance, IT, and business leaders—will be essential to guide this process, ensuring responsible and ethical deployment.

By taking this measured and strategic approach, we can harness the immense power of LLMs to move from a reactive to a truly proactive model of risk management. This will not only strengthen our organizational foundations but also enable us to provide greater stability, security, and peace of mind to our policyholders worldwide, solidifying our role as essential pillars of societal resilience in an increasingly uncertain future.