Paul Carroll

Change management has always been important, but we're now in a period of great change—that is often considered both complex and significant change. How has the importance of change management evolved, particularly in insurance?

Samit Bhandari Principal, PwC US

We've been driving, leading, and participating in large-scale transformations in the P&C insurance industry for 20 years. When you think about technology, we have become more experienced at leading these transformations. There's better tooling, more practitioners, certifications, accelerators, AI—the list goes on. That has helped to reduce several risks in the technology aspect. 

But it's still the same level of challenge in terms of the human aspect of what these programs are delivering, how they're changing the day-to-day work and operations, and what that means to a Customer Service Representative (CSR),[C(2]  to an agent, to an associate internally.

Paul Carroll

There's often a disconnect in technology adoption that goes unnoticed in business. A leader might mandate the use of AI, and employees will say they're using it—but are they really changing the way they do things, or are they just going through the motions to show compliance? How do you get people on the same page in a big technology transformation?

Samit Bhandari

First, just in terms of getting everyone on the same page, it's the fundamentals of what we've always said about getting stakeholders engaged—executive, top-down engagement early, right? The messaging from that level through various levels of the organization. And then just the strategic, tactical things—the change champion network and the communication early and often. When you bring in AI—we should be more nimble.

I do find in the tier two space, the carriers that we work with that are a little bit smaller are nimbler. They're better at embracing the transformation from the top levels of the organization and getting everyone on the same page to highlight that there is a significant effort being made by the organization. 

The tier ones can be a little bit less cohesive, just because they have multiple efforts going on. There may be a lack of alignment within one group to another group, or the change isn't happening to everyone at the same time. 

The second part of this is getting working product in the hands of the end users faster, such that they can get more aligned in the requirements they're specifying, what they're seeing, and then the changes they want to make. 

With the builder concept, we shouldn't have to take the waterfall approach—holding workshops, gathering requirements, then shutting that off and moving sequentially… so on and so forth. We’re able to showcase how we are doing things now, obtaining that builder mentality and increasing the number of engineers so we can take those requirements in real time and turn them into results.  

Are we there yet? No. But that path is becoming much more realistic. 

Paul Carroll

What do you see companies doing wrong about change management?

Samit Bhandari

It can’t just be about the rah-rah, the change champion network, the T-shirts, the slogans, and getting the sentiments higher. That's an aspect of change, and we need to do the communications. But companies don’t always start planning soon enough, and the budget may change throughout the transformation journey in a program. Some initiatives may not be considered a priority up front nor do they have enough budget, time and commitment throughout the program.

Companies may also underestimate what they should do on Day Two, after they’ve had the Big Bang. I was with a few executives for a tier-two carrier in the Midwest recently[C(5] , and we talked about how getting to the pilot was relatively straightforward. But the subsequent rollouts and expansions were a little bit more difficult than they had expected. And then the amount of feedback that came in immediately after they went live was even beyond their expectations.

What they often get wrong is not knowing that this is going to be a continuous and iterative process. The backlog of things to change, enhance and tweak is going to grow the minute you put something in the hands of your associates and your agents. You should have processes up front—both from a change standpoint and a delivery standpoint. From a change standpoint, to expect the amount of feedback that's going to come. But then from a delivery standpoint, to be comfortable in saying, "Hey, we're not going to knock all that off and react within 30 days, but we're going to systematically respond and talk about how we can improve continuously."
 

Paul Carroll

What are the biggest changes you’ve seen over the years?

Samit Bhandari

Early in my career, we spent so much time working with the C-suite. We built strong business cases that quantified why we were doing what we were doing, what the impetus for change was, and what the benefits would be. We went as far as having the different P&L leaders, whether it's claims or policy or billing or actuarial, sign up for the benefits the new platform would have for their constituents.

We've really moved away from that. The attitude became, “Well, we have to do something to keep up with the “Joneses”.” Now, the idea is, “Who needs a big core system from Guidewire  (We no longer have a JBR with Duck Creek so we can’t mention them. We could either reword the sentence to only include Guidewire OR just leave it at “Who needs a big core system.”? We can just take an Insurtech SaaS play and get the work done in months.”

We've unfortunately moved away from taking the time up front to think about the impact a new platform can have on end users or the top line of the organization. I think we should come back to that planning a little bit because it anchors the program in the proper tenets, and it can help assess whether you can just do this in months, given the regulatory complexities.

Once you hear from end users, you realize it's not that simple to just pick up a black box, tinker, and come back with a platform that can addresses various state rules and regulations, coverage needs and DOI expectations. 

Paul Carroll

What are some red flags that companies should watch for, to see they’re doing change management wrong?

Samit Bhandari

Watch out for comments like, "I miss the old system," or "The old platform didn't do this." You may get some of that, but you've really got to have your ear to the ground. Who is saying it? How much is that sentiment catching fire?

Another red flag is when business engagement and user acceptance are lacking. Users should be asking to see the new system early so they can see the results. They should want to get their hands on it and feel excited about what's going to come, because they're in it day to day.
A third warning sign is when you don't have alignment among the leadership. Somebody should be empowered to observe and then make tough decisions, to maybe say, "Hey, maybe this person shouldn't be an executive sponsor," or "We need them to shift how they're operating." Even if lack of alignment doesn't cause the program to fail during the implementation journey—after the fact, it could be just as damaging if people cast shade on what was done. "Oh, we spent too much money on that," or "We overengineered this process," or "It took too long." Leadership should wear the badge throughout the journey and even beyond.

Paul Carroll

What are three or four keys to having a successful change management program?

Samit Bhandari

I can easily flip some of those failures around into successes. Certainly alignment from the top down. The change champion network is important. Individuals should feel excited about what they're seeing. The feedback network should be working well, so you can identify pain points and then dig deeper.

Having a balance of stakeholder engagement monitoring and progress monitoring can help you identify the proper level of adoption metrics early on. 

The last thing is that awareness that it's going to be a continuous evolution and change is continuous. Even after we go live with the first pilot or the next set of states or the portal, there's going to be feedback. How we respond to it and react to it is more important than whether we got it right out the gate.

That last one is often one of the more important things to remember because a balanced mindset of "if we don't get it right, the sky isn't falling." Let's get back together as a team, a cohesive team, and respond. I was in an executive sponsor meeting this morning, and we talked about being 5% over budget. We talked about having 30 open risks[C(6] . We talked about seeing more inflow of defects at this juncture than we wanted. We had the whole C-suite on the line, and everyone was just processing the information, making sure we have the proper rigor around it, and saying, "Alright, here's how we can tackle the issues with the necessary mitigation plans. [7]

The beauty is that when the approach works well, you roll out, go live, and start changing the day-to-day operations of everyone across the different stakeholder communities, internally and externally.

And we're seeing more programs making progress. You've heard that 80% of projects fail[C(8]  (my risk team is questioning this statistic and is asking if we can confirm/source it somehow?) and all that, but that's not the case anymore in this space. The speed to market is often top of mind along with the ability to make changes, identify top-line growth by introducing new products, helping deliver significant changes faster. [9] 

Paul Carroll

Thanks, Samit.

 [C(1]Principal, PwC US should be added

 [C(2]Please spell this out in the first instance unless it is an industry known term by the audience

 [3]Please validate the '20 years' experience claim before publication and confirm the intended attribution.

 [C(4]Not necessary in this instance because the partner is speaking to his personal experience.

 [C(5]The GEP/GRP/LCP/IRP and EL, when necessary, must confirm whether the engagement letter includes restrictions regarding sharing information about the project in public communications, even if the client is not mentioned by name. They must also confirm we are not violating client confidentiality by using specific client or engagement information in the document (even when the client’s name is masked).  

 [C(6]Please confirm these statistics have been vetted and confirmed by practice leadership.

 [7]Please validate the 5% over budget and 30 open risks figures against the underlying engagement records before publication.

 [C(8]Confirm all cited information is properly sourced, independently vetted, and drawn from authoritative, authenticated sources. Check any AI-assisted or AI-generated content against factual evidence to confirm it is accurate, supported, and free from hallucinations or unsupported claims. Also confirm that use of cited materials complies with applicable Terms of Use or licensing restrictions, including whether written permission is required, and present citations on screen or in footnotes using a standard format such as APA or MLA. PwC communications, including blogs, thought leadership, and similar materials, are generally considered “commercial use.” Not all internet content is public domain or freely available for use. Refer to the AI business rules for additional information.  

 [9]Please substantiate the '80% of projects fail' statistic with a current, credible source or remove the percentage.

While we've suddenly landed in the middle of a debate about whether AI may be about to obliterate the human race, I hark back to a profile I did for the Wall Street Journal about a brilliant AI and robotics researcher from Carnegie Mellon named Hans Moravec.

The focus was his provocative idea that humans would be able to download their brains — their entire consciousness, their full personality, an exact replica of them — into computers, which could then teleport to any spot in the universe or spawn an infinite number of what Moravec called "mind children." 

The memorable headline was:

Good News: You

Can Live Forever;

Bad News: No Sex

I asked Moravec how long it would take for his vision to be realized. "Oh, a long time," he said. "Maybe 25 years."

That was 35 years ago.

So I'm not going to worry much for years about all the talk of impending doom. Timelines on sci-fi-like change tend to be way, way off. But, under the radar, there are plenty of AI issues that should be major concerns right now, including for insurers.

Let's have a look.

I'll start with Bill Gates's recent manifesto, which, among other potential dangers from AI, called out the prospect that AI will supercharge the work of malign actors, perhaps leading to bioterrorism, massive cyberattacks, and more. While we can discuss the potential long-term threats to humanity from AI, these are the kinds of threats I think we need to focus on today. These threats are already being pursued, whether by individuals looking to extort massive amounts of money or by nations looking for weapons in an increasingly belligerent world, and AI clearly provides exponentially more computing capability.

The MIT Technology Review goes into detail about how AI might produce a devastating bioterrorism attack: "A bioweapon might be a highly lethal virus that targets people according to their genes. It could be a fungus that wipes out a crop and causes food insecurity. Perhaps it would be a tasteless, odorless toxin that could be slipped into a region’s water supply, undetected....

"Today, AI bots can answer questions on topics spanning all realms of science. Anyone can use large language models trained on the knowledge and experience of 'almost every scientist who ever lived on this planet,' says Dunja Sabra, a biosecurity researcher at the University of Hamburg in Germany. Those models can provide instructions and video training on how to conduct experiments.

"Combine that with advances in biotech that have made gene editing and synthetic biology tools much more accessible (the “DIY biology” movement has already enabled many people to set up labs at home), and you’ve got a potentially very dangerous situation."

Wired, meanwhile, warns about all the vulnerabilities in software that AI bots are finding, and it's not hard to imagine how those weaknesses could be exploited. In July, hackers, thought to be based in Iran, disrupted 30 municipal water systems in Minnesota, and you can be sure Iran will ramp up attacks as fast as it can. North Korea, China, Russia, and other countries could stage similar, small attacks or could even try to shut down electric grids and stall commerce by using weapons of not-quite war.

Cyber attacks could easily lead to massive business interruptions of the sort insurers routinely cover and could increase geopolitical risks of every flavor.

Businesses and governments understand that the bots are making them vulnerable and are working as fast as they can to plug the holes, but they won't find all the holes, at least not right away. And you be sure that some hacker cartel or foreign government is storing up what are known as "zero day" vulnerabilities that can be unleashed on unsuspecting businesses and societies. 

There is also massive potential for operator error as AI is deployed more broadly. For instance, the plan to use AI in air traffic control, just now going live for DC-area airports, strikes me as an accident waiting to happen. I hope everything goes smoothly, but the potential for trouble is so great that insurers and everyone else should be wary. Air traffic control hasn't exactly acquitted itself well lately, and AI tends to amplify flaws by making everything happen faster. 

That's where I think the focus should be over at least the next couple of years, both for insurers and for society writ large: on the potential bio, crypto and other deliberate, organized attacks that AI makes possible from bad actors who would profit from those attacks, as well as on the potential for catastrophe as AI gets more involved in mission-critical efforts. 

Yes, there is always the possibility that the search for superintelligence could create an AI that will go rogue and do indescribable damage to the whole human race, for no apparent reason, so government officials should be erecting guardrails.

But business is already circumscribing what AI can do. Insurers certainly are. They've realized that justifying a decision with "the AI says so" won't fly, so they're requiring that every decision be explainable and are greatly limiting what actions an AI can take without explicit human permission.

The whole superintelligence debate has so many dimensions even beyond the technical ones — there are political elements, issues related to business models, massive public relations concerns, etc. Here, for instance, is a column in the WSJ that argues the whole apocalypse debate is an attempt by AI's leading developers to duck responsibility. 

The issue reminds me of Winston Churchill's description of the Soviet Union after it allied with Nazi Germany in 1939: "a riddle wrapped in a mystery inside an enigma." And I'm supposed to understand the flow of technology revolutions, having followed them for decades.

Fortunately, I think we can wait to puzzle out all the implications of this AI revolution — as long as we don't take our eye off the ball on the imminent threats it creates.

Cheers,

Paul

The Problem

Non-flood damage from water costs homeowners billions a year, and insurers pay $15 billion in claims, just in the U.S. One in 60 U.S. homes suffers water damage each year, and the average claim is nearly $14,000—which doesn’t even include the deductible the policyholder pays, or the huge hassle they go through. Typical solutions involve expensive shutoff valves or a host of small sensors placed throughout a home, but providers have struggled to make a compelling case that their programs deliver an ROI.

The Solution

LeakBot takes a different approach, one that begins before a leak occurs and stretches to a solution. The approach starts with a LeakBot sensor that policyholders clip to the main water pipe in the house. The sensor detects any leaks anywhere in the house that exceed a teaspoon per minute. LeakBot’s app walks a policyholder through steps to find leaks, then, if necessary, sends one of its plumbers to visit the house to find any remaining ones, at no charge. The LeakBot plumber documents the repair, giving carriers visual validation of every claim saved.

The approach finds even tiny, hidden leaks weeks or months before they can cause significant damage. In 2025, LeakBot completed 7,000 leak repairs, 1,470 of which revealed water damage that had already begun.

More than two dozen carriers provide the LeakBot solution for free, paying LeakBot $5 a month per house, all in. LeakBot reduces the number of water loss claims by 60%, generating a significant ROI for its carrier partners.

LeakBot covers the full loop, detection through documented repair, which makes it the only true end-to-end platform in this space.

The Documentation

LeakBot has produced a white paper that clears the actuarial bar for detail and reliability. The paper draws on hundreds of thousands of device-exposure years. It uses cohort-level (not anecdotal) evidence. It relies on auditable repair records — rather than programs still running on pilot-stage promises.

Sponsored by Leakbot

There is always going to be another new platform, another integration, and, increasingly, another conversation about AI. But the value of technology isn't in simply having more of it. It's how well it connects and whether insurers get better information when they need it.

At a time when AI is redefining how the insurance industry will function for the next 30 years, a converged environment unifies underwriting, policy administration, billing, claims, and customer engagement on one operating platform, rather than a chain of systems handing data to each other. A claims signal reaches pricing while the claim is still open, informing the next risk assessment in real time. An underwriting decision immediately shapes how service and retention teams treat that customer afterward. Information moves in near real time, following what the business needs rather than the limits of the original system architecture.

The Case for Converged Platforms

Across this platform, AI becomes part of how the workflow runs. An agent reviewing a claim has policy history, the underwriting file, and prior customer interactions available at the same moment, because they live in the same context rather than three different logins. That context lets AI move from flagging an exception to resolving it, and from recommending a next step to taking it, with a human reviewing the outcome afterward. The reasoning behind each action stays visible and auditable, so the people accountable for the outcome can see exactly how the system reached it.

The practical benefits show up quickly. Teams configure new products within a shared environment instead of rebuilding them system by system, so what used to take months of coordinated releases now takes days. Renewal pricing, fraud screening, and service routing draw on the same live data rather than nightly batch exports, so all cross-referenced with each other and reflect the real-time state of the business.

The autonomous AI platform we are describing doesn't mean anyone should abandon the judgment that experienced underwriters and claims handlers bring. A converged environment pairs automation with quality gates. An organization can route a routine renewal straight through and hold a complex commercial risk for human review, using the same data for both. The goal is to make sure the people making the decision are working from the full picture, instead of whatever fragment their siloed, legacy systems offer.

Rethinking Process Design in an AI-First World

It is important that companies should think carefully about the shape of their AI models. There is a natural tendency for people to focus on human-driven, effort-based, bottom-up processes and then apply AI efficiency factors to achieve better productivity results. I believe that approach is fundamentally flawed as it quietly assumes the old process as the correct baseline, when the old process is the very thing that should be questioned.

Companies should work back from the assumption that every function is AI-automated and design that as the starting point, not the improvement point. Then justify, line by line, the human effort that is genuinely needed to be added back: the judgment calls, the client-facing decisions, the regulatory sign-offs, the things an autonomous AI agent demonstrably should not be permitted to do.

The best approach is to question the requirement itself and what value it delivers to the business; wherever possible, delete the step entirely rather than automate it, simplify the real requirements that survive, and only then automate the complete end-to-end process with the minimum human intervention possible. There is no business value in automating a step that should have been removed completely, as this is the most expensive mistake that organizations make.

The clearest sign that this transformation is working comes down to a simple test. Can a service representative see a customer's full history the moment the call connects, without transferring them or putting them on hold while three systems are checked in sequence? Can an underwriter check how similar risks have actually performed against what the rate manual predicts? Can a claims handler settle a straightforward loss the same day it's reported, because the verification, the policy check, and the payment authorization all draw from the same source of truth?

That is the standard worth building toward. For companies where coherence is increasingly a transformational priority, my advice is to build an environment where every part of the business can see what every other part already knows, and where AI has enough context to act on that knowledge instead of merely describing it.

The NAIC's AI Risk Evaluation Supplement is the questionnaire regulators will draw on to examine how insurers govern their AI. It asks whether a model's outputs were tested for accuracy, but only for the models a carrier itself rated high risk. When a customer-facing model gets a coverage question wrong, the policyholder acts on the wrong answer. 

That is the exposure a high inherent risk rating exists to capture. Rate one low instead, and the inquiry ends there. The model keeps answering coverage questions, the answers keep reaching policyholders, and nothing requires the carrier to check whether they were right. Few carriers check on their own.

What a low rating leaves unchecked surfaces later, in a claim file rather than an examination. Misrepresenting policy provisions is a listed unfair trade practice in nearly every state, and a dated coverage statement the policy contradicts is what a misrepresentation claim is built from. The customer who heard "you're covered," paid for the repairs, and then had the claim denied is the one who brings the complaint.

When that complaint arrives, a court will ask how often the model's answers were checked and what the checks found. Errors and omissions underwriters, who now ask about AI at renewal, want the same answers. A carrier that rated its model low will have no file to produce. There is no record of what the model said and no comparison against the forms in effect at the time. 

What it will have is a self-attested risk assessment concluding that the model was low risk. Plaintiffs' counsel look for exactly that pairing: a carrier that never checked, and a document explaining why it did not have to.

How the rating works

The rating is set early, in an inventory. Each carrier lists its AI models, describes what each one does, and assigns each an inherent risk level. Only the models rated high move on, and there the questions turn specific: were the outputs tested for accuracy, how is performance monitored on a continuing basis, and how is the model reviewed against unfair trade practice and claims settlement laws? 

Models rated moderate or low do not reach those questions, and a regulator reading the inventory has discretion to stop at the rating and request nothing further about the model. That is sensible design, since regulators cannot examine everything. It also makes one self-assessed rating the single point of failure for whether a model's accuracy is ever examined, and the carrier is the one holding the pen.

How a customer-facing model goes unexamined

Two judgments decide whether a customer-facing model is ever examined. Does its output reach a consumer, and how much risk does it carry before controls? A carrier can get either one wrong without meaning to.

Destination settles the first question, not who speaks the words. The supplement's background section covers decisions and actions “made or supported by” AI, and a coverage answer that reaches a policyholder is one of them. The pitfall is the supplement's support category, defined as a system that provides information without suggesting a decision or action, which is not counted as having direct consumer impact. A model that drafts a coverage answer is not just providing information. It is supplying the decision the representative delivers, and the support definition excludes exactly that. But because a representative delivers the answer, the model looks on paper like one that only informs an employee, and that resemblance is what puts it in the wrong box. Filing it as support keeps the model out of the count regulators use to decide what to examine, and out of any question about whether its answers were right. Those answers still go out to policyholders, unchecked.

A model that is counted can still be rated low by mistake, and the same representative is usually why. Most carriers put two guardrails around a model's answers. The representative is expected to check it before it goes out, and the platform runs its own internal check against policy language. Both are controls, and because they are built into the process, crediting them when rating the model is an easy mistake to make. The supplement asks for inherent risk, the risk the model carries before any control is applied, and the word "inherent" is easy to read past. Under the supplement's own terms, the representative and the platform check belong in the residual rating, after controls. A carrier that credits them in the inherent rating has rated its model lower than the supplement intends, and taken it out of the testing questions in the process. Both judgments are already recorded, one line per model, in the carrier's own inventory.

Why the cheaper rating costs more

Rating a model low is the cheaper answer. It closes a line on the inventory, raises no question the carrier has to answer, and creates no obligation to find anything out. Rating it high creates recurring work: finding out whether the model's coverage answers were right and continuing to do so while the model runs. The pressure runs one direction, and nothing in the supplement pushes back. The saving lasts a quarter. The gap it leaves is permanent. That work cannot be backfilled later, because the evidence exists only while the conversations are happening, and controls do not close the gap. A representative who catches a wrong answer fixes that conversation and leaves no record of it, so even a carrier with an excellent review process cannot separate an isolated error from a model-wide one. It also cannot see drift. Models do not hold still after launch. They get updated, the policy forms they draw on are amended, and the questions customers ask shift. A test run before deployment describes the model that was deployed, not the one answering calls this quarter. What a low rating costs is a year of conversations nobody measured, and nobody can go back and measure.

I cannot tell you how often AI models give wrong coverage answers, and neither can anyone else, because almost nobody is measuring. The draft is open for comment through Sept. 29, with a vote on a later version expected in November. The questions are being written now, and the evidence they ask for can only be gathered before they are asked. Rating a model low does not change what it told a policyholder. It only changes whether anyone was looking.

The general assumption about innovation is that it's top-down. Someone invents something complicated and expensive. As it achieves scale, the innovation becomes simpler and less expensive, making it accessible to broader audiences. Rinse and repeat until the innovation becomes global.

In insurance, that trend has typically meant an innovation in the U.S. or Western Europe that finds its way over time into developing economies.

But there's also an interesting trend known as "reverse innovation" that I've watched play out for some 15 years. If you have a multimillion-dollar medical device, you can take cost out, time and again (and again and again), and still not make it cheap enough to be economic in a poor, rural community. So some smart folks in India working within the severe cost constraints of a poor area invent a version of the medical device — and sometimes that device works well enough that it can reverse the normal flow, migrating from developing areas into the major economies.

To learn more about what reverse innovation is happening — or could be happening — in insurance, I turned to George Kesselman, the founder of InsurTech Asia. He grew up in Canada and went to school there, before spending the last nearly two decades working in Hong Kong and now Singapore, so he has a thoroughly global perspective.

His short answer: Pay very close attention to China.

His longer answer follows.

I met George through a sister organization, the International Insurance Society, for which he has been hosting a working group on innovation that I'm part of. (George will present the findings from the IIS's global survey on insurers' priorities, as well as insights from our working group, at the IIS' annual gathering, the Global Insurance Forum, being held in London at Lloyd's on Nov. 15 and 16.) Here is what he said about innovation in Asia, edited slightly for length and clarity:

KESSELMAN: This year, I had the pleasure of doing a coast-to-coast tour over two weeks in the U.S. and Canada, spending a few days in New York, Toronto, Vancouver, and San Francisco. My read is that innovation is really quite different in North America than what's happening outside. 

The last wave of innovation, from roughly 2014 to 2023—let's call it a 10-year span—was very much an Asia-first wave. It was all this digital insurance that started in China, then Southeast Asia, with a lot of new models when it comes to embedded, small insurance. Then I think the innovation went a little bit more to Europe. In North America, it hasn't really taken off in the same way.

This time, the wave is kind of inverted. With all the AI innovation, all the investments, it's North America first, then Europe. Asia is seeing bits and bobs but not nearly as much as the other two regions.

My theory is that the scale is massively different in the regions, and the problems are also quite different as a result. Asia is very much about the growth of insurance. Insurance wasn't really popular here because the region was a lot less developed. The last 10 to 15 years have seen tremendous economic growth, so innovation and insurance started to come in. Insurers couldn’t scale the same face-to-face distribution, so they found different ways, including digital distribution.

But the scale still remains drastically different. When I was in North America and Canada, everybody talked in terms of billions of dollars. In Asia, people talk about millions, maybe sometimes tens of millions. 

CARROLL: What kinds of embedded insurance innovations emerged in parts of Asia that have filtered out to other parts of the world?

KESSELMAN: In e-commerce, we saw quite a bit of innovation in warranties and the return shipping space. Now we're seeing innovations around trying a new product and getting coverage if you don't like it—you experience an upset stomach after eating something, or face delayed delivery. These are very region-specific and emblematic of the logistics and distribution challenges in Asia.

Travel is another ecosystem that saw significant innovation. Mobility is the third. Then fintech-related insurance is probably the fourth. Each has its unique characteristics. In fintech, for example, we've seen a lot around cyber incidents—if your wallet gets hacked, you get paid—or bill payment protection, where if you get sick, the bill gets paid on your behalf. These are embedded and bundled propositions that are very continuous to the core offering.

These innovations started here because digital leapfrogged in a lot of countries where the infrastructure didn't exist before and really accelerated adoption. Last week I talked to one of the big e-commerce players here, and they estimate about a billion dollars in these micro embedded insurance propositions being conducted yearly now. The growth is definitely there, and the trend has reached scale across the region, though it's still relatively small compared with an average MGA in the States.

As for how this is spreading, some of it is now landing in Europe. Warranty products, lens insurance, hearing aids—these are probably more specific to Europe because the needs and market development are different. In the mobility space—scooters, car insurance, and ride-sharing—I'm hearing that's happening there, as well.

CARROLL: I assume China has played a big role.

KESSELMAN: Embedded and digital insurance really took off from China. The big players were Ping An and ZhongAn—both kind of protection-focused players—and both of them really aggressively invested in innovation. China probably remains three or four years ahead of the rest of the region.

From China, innovation propagated to Southeast Asia. I think India is now also quite a significant market for it, but China was very clearly the first because of the combination of digital payments penetration being super high there, the development of super apps that put everybody—all the distribution—in the same place, and then e-commerce and all the logistics. Everything developed at the same time pretty rapidly.

CARROLL: What are some examples in Asia that other parts of the world could learn from in terms of insurance innovation?

KESSELMAN: Ping An and ZhongAn remain interesting examples. They have been investing very actively in AI, and I think they're still very much focused on distribution and servicing, and developing propositions that are very targeted.

In India, there are some insurers that are starting to do some really interesting stuff. Acko is one that comes to mind in the mobility space that has been quite active. The rest of the region has pockets of innovation, but because the countries are generally smaller in scale, it's very hard for insurers to really grow very rapidly. You know, you start in Singapore, then you need to expand into a neighboring country, and then you need to go and redo the whole growth journey again.

CARROLL: What sorts of interesting startup ideas are you seeing at Insuretech Asia?

KESSELMAN: There have been a couple of waves of startups. The first was really about trying to disrupt distribution. That wave has kind of passed now.

As part of the second wave, we have a couple of really interesting startups. Two are unicorns. Bolttech is one. They focus on distribution and have a couple of different complementary models. They have a very big presence in the U.S. and a global presence, but their headquarters are in Singapore.

Then we have InsureMO, which is like an infrastructure for insurance distribution. They have very big operations in India and Southeast Asia, and now they have growing operations in Europe, the U.S., and Latin America, as well, but they’re also based in Singapore.

CARROLL: What makes those companies different?

KESSELMAN: I think it’s exactly the fact that they started in Southeast Asia, so they needed to do everything very cheaply and fast even though the platforms that are here are very demanding in terms of the operational support that’s needed, in terms of technology. Being incubated here, I think, gave them that natural foundation of really building a scale-up in a demanding environment. They tell me that when they go to the States and talk to insurers there, insurers are very surprised that they can do things for a fraction of the price and at double the speed.

CARROLL: What has surprised you about how AI is being adopted—or not being adopted?

KESSELMAN: The part that surprises me is that it is still very much a technology conversation rather than a business conversation. To give you an example: Recently a couple of insurers in the region came here, and they have flagship programs, but they're all kind of stuck. They don't really know how to solve the problem of creating business value with AI.

This is the same theme we saw in the IIS survey.

Everybody kind of saw this shiny object and had experience with really interesting outputs of this technology, but there's a disconnect between the solution that exists and the problems that insurance companies need to solve. I also think there's a disconnect between the leadership level of the organizations and middle management organizations, which is not as bought into this whole idea of "let's bring AI everywhere, let's augment every job with it."

CARROLL: That's such a common problem—bosses demand change, and underlings nod their heads… but they also find ways to keep doing things the way they always have. Peter Drucker once told me that people don’t change. They just die off, and the next generation then does things differently. But we were talking 25 years ago. I hope we’ve learned a bit since then.

This has been great, but any final thoughts?

KESSELMAN: There are interesting developments in health insurance, again starting with China. Everything is getting so integrated. Hospital data is getting a lot more integrated with insurance data, and it's getting linked up in the hospitals to make it much more proactive—managing health, managing well-being. So rather than putting strain on hospitals, with insurance being really at the back of it, the loop is starting to get closed.

The rest of the world will eventually need to tackle that integration issue as populations age and healthcare systems are probably redesigned. Insurance right now is still very much on the receiving end.

That integration in health insurance is probably one of the most interesting things happening in this part of the world, and, again, I’d use China as the reference market. 

CARROLL: Thanks, George. This is super helpful.

Cheers,

Paul

A teenager's first driver's license represents freedom, responsibility and, for parents, more than a little anxiety. For insurers and independent agents, it also introduces one of personal auto's most challenging exposures: a driver with little experience, no established record and a statistically higher likelihood of a crash.

The financial effect can be dramatic. Bankrate data cited by CNBC estimate that adding a 16-year-old to a family "full-coverage" policy can increase the average annual premium from $2,671 to $5,910. The human cost matters more. AAA reports that 2,636 people were killed in crashes involving teen drivers in 2024, underscoring that the consequences extend beyond young drivers to their passengers and others sharing the road.

For most families, insurers address this exposure within the personal auto policy rather than through a separate teen-driver product. They adapt through inexperienced-operator classifications, vehicle-assignment rules, underwriting requirements, and pricing that reflects years licensed, vehicle type, garaging, mileage, and use.

What Agents Should Do

For agents, the process should begin when the teenager receives a learner's permit, not after the first solo trip. The first obligation is accurate disclosure of everyone the application or policy requires to be identified, including household residents and customary operators. How a disclosed driver is treated for rating purposes is a separate question governed by the insurer's filed rules. This early conversation also provides an opportunity to discuss formal training. In some states, waiting until age 18 may eliminate a mandatory driver-education requirement, but it does not make the newly licensed driver experienced and may mean missing an available training discount.

That distinction becomes especially important when parents live separately. The ISO Personal Auto Policy uses "family member" for a person related to the named insured by blood, marriage or adoption who resides in the named insured's household. The Massachusetts Auto Policy uses the term "household member," with a substantially similar definition. Neither definition designates a single primary residence or makes the address on a license or registration controlling. Depending on the child's actual living arrangement, the child may qualify under policies issued to both parents. Agents should therefore review both households, both parents' auto policies and umbrellas, and every vehicle the teen may operate.

Ownership creates another potential trap. Some families place the policy in the child's name while the parent retains legal title to the vehicle. A carrier may accept that arrangement, but acceptance does not change the policy definitions. Being listed as an operator does not necessarily give the parent the rights or insured status of the named insured. The mismatch can affect liability arising from ownership, permission for other drivers, automatic coverage for a replacement vehicle, physical-damage proceeds and umbrella protection.

Coverage Vs. Price

Coverage should receive at least as much attention as price. Liability and uninsured or underinsured motorist limits, physical-damage coverage, deductibles and the family's personal umbrella should all be considered. A resident teen relative generally qualifies as an insured under the ISO personal umbrella, but forms vary and the underlying auto insurance must be coordinated.

Affordability still matters. Depending on the carrier and state, savings may come from good-student or low-mileage discounts, completion of approved driver education or advanced crash-prevention training, and telematics programs that evaluate speeding, hard braking, mileage, time of day and cellphone distraction and provide financial incentives for safer driving.

Insurance cannot eliminate the risk that comes with a new driver. But thoughtful underwriting, properly coordinated coverage, safety-focused incentives and an informed independent agent can help families manage it—giving teenagers room to gain experience and parents a little more peace of mind.

$117 billion: the insured average annual loss (AAL) estimated for natural catastrophe events in the United States, according to Verisk's 2026 Global Modeled Catastrophe Losses Report. This figure represents roughly two-thirds of the $171 billion global AAL benchmark. The AAL is not a forecast for next year; it is the level of loss the insurance industry should be prepared to absorb over the long term. North America remains the dominant contributor, accounting for $124 billion of the global total.

A benchmark built for today's risk environment

Global insured losses have topped $100 billion for each of the past six years. Since the report's first edition in 2012, estimated global modeled insured average annual losses have increased from $59 billion (in 2011 USD) to $171 billion (in 2025 USD). The increase reflects several factors, including exposure growth, expanded model coverage, advances in catastrophe science, and continuing refinement of the industry's view of risk. The 2026 AAL benchmark also reflects the latest generation of model updates, so it captures current science rather than relying on the assumptions of the last decade.

Why the average is only part of the story

The insured AAL is a long-run expectation, not a prediction. Especially in the U.S., frequent perils such as severe thunderstorms, wildfires, inland floods, and winter storms drive the average, while severe perils, including hurricanes and earthquakes, drive the extremes. Losses in any single year can land far from the mean, which is why the full range of modeled outcomes matters more than any one number.

The difference between average and extreme outcomes illustrates why insurers focus on more than the mean. Against the $171 billion global AAL, the same models estimate a 100-year industry loss approaching $477 billion and a 250-year loss exceeding $600 billion. Catastrophe models are designed to look beyond historical experience and quantify plausible events the industry has not yet experienced.

A single year of lower loss activity doesn't mean lower risk

Catastrophe losses were below average in 2025, but a below-average year reflects a favorable draw, not a change in the risk landscape. Despite the absence of a landfalling hurricane in the continental United States, global insured catastrophe losses still exceeded $100 billion. Wildfires and severe thunderstorms demonstrated how the accumulation of frequency-peril losses can generate significant industry losses even without a headline catastrophe event. The first half of 2026 further illustrated the point. The U.S. p&c industry's combined ratio improved to 92.7 from 96.5 a year earlier, and its estimated net underwriting gain nearly tripled to $31.7 billion. Those results show where the industry landed on the curve this year, not a shift in the underlying risk. The $117 billion expectation for the U.S. doesn't shrink because the first half of the year was kind.

A favorable year can improve financial results, but it does not alter the underlying distribution of catastrophe risk that insurers must manage.

Better to be prepared than surprised

For risk leaders, preparation starts with the AAL, the anchor for catastrophe risk appetite and capital planning and the number with the deepest data behind it. Catastrophe models rely on increasingly detailed exposure information, including building characteristics, reconstruction costs, and claims experience. The quality and currency of that data influence how insurers evaluate risk appetite, reinsurance needs, reserving assumptions, and capital allocation decisions. As exposures evolve and catastrophe science advances, maintaining an accurate view of risk becomes increasingly important.

Three considerations stand out in an environment where catastrophe losses regularly exceed historical expectations. First, evaluate risk appetite against the full range of modeled outcomes rather than recent loss experience alone; the tail of the distribution is often what places the greatest strain on balance sheets. Second, consider how reinsurance programs respond to the perils that drive extreme loss scenarios, not just the events dominating recent headlines. Third, reassess capital assumptions when exposure profiles or catastrophe models change materially, rather than relying solely on periodic review cycles. The purpose of catastrophe modeling is to look beyond what happened this year and quantify what could happen across thousands of plausible future scenarios. The industry's challenge is not preparing for the year it just experienced, but for the years it has not yet seen.

Sources

2026 Global Modeled Catastrophe Losses Report: https://www.verisk.com/resources/campaigns/modeling-insured-catastrophe-losses-a-global-perspective/

1H 2026 underwriting results: https://www.verisk.com/company/newsroom/

When a representation letter lands on a workers' comp claim, an experienced adjuster does roughly the same things every time. They acknowledge the letter, reach out to the claimant's attorney, adjust reserves now that an attorney is involved, and update the claim summary. Ask 10 adjusters for the sequence and you'll get the same list in slightly different orders, with a step or two of local flavor.

The same holds when a return-to-work note comes in, full duty or with restrictions, or a demand package. It even holds for silence: a claimant who hasn't heard from anyone in three weeks is its own trigger, with its own response. These sequences repeat from file to file, and mostly from shop to shop. An adjuster with 15 years of experience has run some of them a thousand times, and could run them half asleep.

Workflow orchestration gets defined in a lot of ways. If we keep it really simple, it comes down to this: when something needs to happen on a claim, the right response follows.

Done well, the routine response happens on its own, and the exceptions land with a person quickly. The way to get there is a system that learns the actions taken on past claims, when they were taken and what they led to, and then executes them automatically. In other words, it's like a new hire on the claims floor, one who happens to have read every claim the company ever closed and can be trusted with the motions everyone already agrees on. The calls that matter stay with whoever owns the file. And like any new hire, the system gets brought along in phases. None of this was practical before, because most of a claim file is prose. But language models can finally read it.

Where to Start

The place to start is simple, getting the work itself into view. What's happening on each file and what the adjuster is doing about it, gathered in one place instead of spread across the diary, the inboxes, and the core system. Today the diary is the closest thing to that picture. A file comes up because 60 days passed, not because anything happened on it, and the first half hour goes to finding out whether anything did.

An overview like that is worth building before a single routine runs on its own. It shows what changed overnight and what's waiting on a decision, and it starts the adjuster's morning at the right files. It also means the state of a file no longer lives only in the adjuster's head, so nothing stalls when someone is out for two weeks, and a handoff doesn't start from zero. And it gives a claims operation something it has never really had: a clear picture of its own work. Everything else, the suggestions and eventually the actions, builds on that clarity, because work can only be handed to a system once it can be seen. The phases are how that picture gets built.

Phase One: Analyze

Every one of these sequences starts from a trigger. Sometimes the trigger is an event, like a letter arriving or a claimant going quiet. Sometimes it's a scheduled action: a state deadline on the first payment, or a follow-up that was promised and hasn't happened yet. The first phase is about proving those triggers get caught. The system doesn't act on anything yet. It reads everything that arrives and reports what it sees: what changed on the file since yesterday, and which of those changes should set a sequence in motion.

The same reading, pointed backward, is where the insights start. Once the files are readable, questions that used to require a file review project can be answered directly from the data: how much subrogation actually went by unnoticed last year, or how long attorney letters sat before anyone opened them. Each of those numbers is a count of missed triggers, and together they make the case for orchestration from the operation's own files.

A system that only watches already earns its keep. Nothing important sits unread anymore, and leadership questions stop being answered by sampling and gut feel. It's also the phase today's models are ready for. Reading a claim file and noticing what matters is exactly the kind of work language models have become reliably good at, and the cost of a miss is an alert that didn't fire, which is no worse than the status quo. Every week of watching also builds a record of what the system noticed and what the adjuster then did about it. That record is what justifies the next phase.

Phase Two: Suggest

In the second phase, the system starts making recommendations. The acknowledgment letter shows up drafted. The claim summary is updated and waiting. The reserve review appears as a proposed task with the reasoning attached, and the wage statement request is ready the day a claim turns lost time. So is the status update to the injured worker, who mostly just wants to know that someone is paying attention. Plenty gets corrected before it goes out, and some suggestions get ignored, which is useful information in itself.

Reviewing this output is real work. If the system surfaces 40 suggestions a day to an adjuster who already has 30 diary entries due, nothing has been automated; a backlog has been created with extra steps. The review process has to be staffed and sized like any other part of the operation, and it has to be worth the adjuster's time. The corrections matter most. Every fix an adjuster makes to a suggestion is a lesson, and a system that keeps those lessons gets better every week it runs, while a system that throws them away will be making the same mistakes a year from now. The way out of this phase is consistency, motion by motion: suggestions going through untouched often enough that the approval becomes a formality.

Phase Three: Act Where Possible

The last phase is the one vendor demos lead with, and the one that should come last. Some motions start running on their own, but only the ones with a record behind them, and that record is task by task. Sending an acknowledgment letter unsupervised is a very different proposition from touching a reserve unsupervised, and phase two produces the evidence for that call: how often adjusters approved each kind of suggestion unchanged, and where the overrides cluster. Settlement authority for a new adjuster grows in the same way, with the track record. Autonomy should expand at the pace the record justifies.

The AI models will keep getting better. That doesn't change the design. It just changes how quickly a claims operation moves through the phases.

With several states now allowing all employees, not just first responders, to file mental health-related workers' compensation claims, industry professionals are pondering a potential new challenge. What happens if an employee becomes so concerned about AI taking their job that it has an extraordinary effect on their mental well-being?

While it's a fascinating question, the reality is that we haven't seen any evidence of this type of situation occurring in case law thus far. Instead, most workers' compensation professionals are focusing on the positive aspects of AI, using it to streamline the multitude of time-consuming, behind-the-scenes tasks that have slowed down insurers, companies and lawyers for years.

This refocus on the positives of human-supported AI in business combined with recent rollbacks on AI-related layoffs could actually lower stress on employees, as well as reduce the potential for those AI-rooted mental-health related claims.

Job replacement fears allayed

While early employment trends stoked fears of AI replacing human workers, we're now seeing a vastly different scenario play out as companies start to realize that AI, while powerful, has its limitations.

Major employers like Ford and IBM, which conducted AI-related layoffs, rehired employees across engineering, human resources, and customer service this summer to fix quality issues that AI could not address, according to a recent CNBC article. Another report shows that 55% of business leaders who made employees redundant after deploying AI now admit they made the wrong decision.

As companies roll back AI-related layoffs, the renewed emphasis is on creating a workplace where AI and humans co-exist. That reframing is reducing stress for some employees rather than adding to it. Workers are using AI to complete more tasks faster. The end result is that some employees can now spend more time interacting with humans and less time writing emails and staring at computer screens.

Yet while the "AI is taking your job" storyline has subsided in some instances, the case law regarding AI and mental health in the workplace has yet to be written. Workers' compensation boards in most states are very claimant-friendly, and with AI changing so much about how employees do their job, it's within the realm of possibility that we may see stress-induced claims related to AI pop up in the coming months and years.

Benefits Outweigh the Concerns

Despite the negative headlines related to job replacement, AI presents benefits to all parties involved in a workers' compensation claim.

To see how, consider the common denominator in any claim: paperwork. Claimants, claims professionals, and attorneys must comb through thousands of pages of incident reports, medical summaries, specialist notes, imaging studies, surgical records, physical therapy and rehabilitation notes, return-to-work forms, and disability status certifications. Doing so manually typically took months.

AI now expedites the process by reviewing and summarizing reams of digitized files and pulling out the most pertinent information. Lawyers and claims professionals can search for specific terms within those documents to refine their case. These types of efficiencies accelerate the entire litigation process, potentially leading to faster resolution.

AI also benefits injured workers. Some states require workers to show proof of their job search after they've recovered from an incident. With AI, workers can enter their skills, experience, and medical restrictions, then run queries to find jobs that match their unique needs. Carriers and third-party administrators are doing the same, using AI to assist claimants in the job-search process.

For corporate risk managers, AI insights can help develop stronger workplace safety programs. An employer can review its claims database, look for repeat patterns and determine whether injuries are more common at a certain job site, in a specific role or at a certain geographic location. They can then dig into the root causes of incident reports and adjust protocols. If data shows, for instance, that workers are getting hurt due to improper lifting, risk managers can enhance their safe-lifting training program by providing more specific and frequent education, thereby improving worker health and well-being.

AI is also shaping litigation strategy. Attorneys are analyzing historical claims and litigation outcomes with AI to identify trends in settlements, dismissals, and case resolutions. These insights provide additional context that claims professionals and legal teams can use to evaluate their past results and tweak their case strategy going forward.

Human Judgment Remains Vital

While AI can support many steps of the workers' compensation process, it does not replace professional judgment. AI is most effective at analyzing data and identifying patterns. Claims professionals and legal teams should remain in the loop at all times, review all AI outputs for accuracy, and make the final decision whenever data conflicts. The strongest results come from combining AI-driven insights with human expertise.

The fear around AI and job replacement is real, and it may someday lead to a mental health claim from an injured worker. Fears, however, have a way of working themselves out over the long run. We saw this when states first began legalizing medical marijuana. The fear was that it would forever change workers' compensation. The reality was that the legislative changes had a more measured impact.

It's too early to know how much AI will change workers' compensation, but as the fear of job replacement starts to subside, professionals should focus on the positives. When AI increases efficiency, we can give workers what they deserve: a safer workplace and a faster resolution to their claims.