The proving ground
Insurtech has never lacked ambition. Over the past 15 years, successive waves of technology companies have promised to modernise insurance, digitise underwriting and transform the industry’s operating model. Billions of dollars flowed into the sector as investors backed platforms designed to streamline everything from policy administration to distribution.
Many found niche success. Few fundamentally changed the way insurance was done.
Martin Henley, founder and chief executive officer of mea Platform, told Bermuda:Re+ILS that he remembers that period from the other side of the table.
Prior to leading mea Platform, he spent years as a chief information officer responsible for buying technology for insurers. He watched the first generation of insurtech emerge, evaluated countless products and saw an industry searching for solutions that the technology of the day often struggled to deliver.
“The 2010s were more around point solutions,” he recalls, but processes, he said, “pretty much stayed the same”, while many businesses attracted significant investment before proving they had genuine product-market fit.
“As a buyer, I was never quite convinced of how many of these really were going to get proper traction in the market.”
That experience shapes the way he views today’s AI boom. While comparisons with the previous insurtech cycle are inevitable, Henley argues they overlook a fundamental difference. Earlier technologies helped insurers complete existing processes more efficiently. Agentic AI, however, “does the work itself. It’s a very different set-up”.
The capability is no longer theoretical: “We can very quickly demonstrate to clients, or potential clients, just how impactful this could be on their businesses. We can do that with a demo in a way that you just couldn’t even have thought about seven or eight years ago,” he explained.
Commercial specialty insurance has presented one of the hardest proving grounds for new technology. Policies are bespoke. Documentation is extensive. Risks are highly complex and often span multiple jurisdictions. The first generation of insurtech frequently promised to tackle that complexity but lacked the technology to do so.
Now agentic AI that understands the insurance market is closing that gap. “The technology can just handle things that, frankly, the industry, especially the commercial specialty industry, has been forever crying out for. These types of policies are complex and carry large risks, and therefore so too do the documents that come with them; we can now manage that with AI very, very reliably.”
For Henley, that represents something more significant than another technology upgrade. It marks the point at which insurtech begins to fulfil the promise on which the sector was founded.
The real AI advantage
From there, the conversation moves beyond automation and into something altogether more strategic.
“The AI becomes a way to permanently change the unit economics of your business because it allows you to do all of those repeatable tasks, the things you need to do every time, instantly with no human effort.”
Much of the industry’s discussion around AI has focused on efficiency and reducing expense ratios. Henley sees those benefits as secondary. Faster processing and better data allow insurers to respond more quickly to brokers, strengthen relationships and increase underwriting capacity. “Framing it all as making some cost reduction savings is one way of looking at it. But we don’t believe that captures even a tenth of the opportunity,” Henley said.
The distinction between generic AI and insurance-specific models sharpens his argument further.
Models that understand insurance
One of the industry’s emerging misconceptions, Henley believes, is assuming that increasingly capable general-purpose AI models automatically understand insurance. “The industry is only just starting to realise the gap. Generic AI understands language. The insurance-specific AI, certainly the AI we’ve built, understands the language and the language in context, and that becomes very important in insurance.”
Context, after all, sits at the heart of underwriting and claims. The same phrase can carry different implications depending on the line of business, the wording around it or the jurisdiction in which it is applied. Replicating that judgment requires more than a model trained to generate convincing language.
This was where Henley’s passion for the distinction between generic and insurance-specific AI really came to light: “Some terms in the same documents can be very similar but mean very different things in different contexts. That’s what the insurance-specific AI allows you to manage, and as it learns, it can also understand and follow the logic of an insurance policy, or an endorsement, for example.
“Context is everything. Whether you’re writing a policy or deciding to pay a claim, context can even be different by line, market or jurisdiction, and the insurance-specific AI is able to deal with that in a way that an underwriter or a claims adjuster would.
“We’re seeing auditability and the confidence scoring, which is critical to regulators, particularly for underwriters, and those are the aspects the insurance-specific AI is able to do with a much greater level of fidelity than the generic.”
Those capabilities begin to move AI beyond automation and into decision support. Rather than simply extract information from documents or produce draft text, the ambition is to mirror the reasoning that sits behind underwriting and claims, while providing the transparency and confidence measures that regulated insurance businesses require.
This is notably steps ahead of the origin stories of insurtechs, no longer a tool but a toolbox and production line for insurers. That puts mea a generation beyond the first cohort.
Addressing the sceptic
The industry’s natural scepticism remains understandable. Claims that AI can perform as experienced insurance professionals inevitably invite scrutiny.
Henley does not attempt to persuade through abstract promises: “My typical answer is, let me show you examples, and we’ll show you that it works – fast.” mea is live with more than 30 clients across 20+ countries and has processed more than $450bn in gross written premium.
The ability to demonstrate real-world performance, rather than theoretical capability, reflects another difference between this technology cycle and those that came before it.
It also changes how insurance organisations should think about implementation. Rather than embarking on multi-year transformation programmes, Henley believes insurers can introduce AI incrementally, redesigning workflows one process at a time.
"You don’t need to implement a huge platform to do everything. And it doesn’t have to take years. Those days are long, long gone."
Bermuda as the proving ground
AI will embed itself in insurance, but the full transformation will require proving time in an environment supportive of innovation.
For Henley, Bermuda possesses several advantages that extend well beyond its size. “The decision-makers for some, if not all, of the world’s most complex risks all sit in Bermuda. And everyone walks Front Street,” he said.
That concentration of expertise creates unusually short distances between carriers, brokers, technology providers and regulators. New ideas can be tested, refined and adopted within a tightly connected marketplace.
Henley also credits Bermuda’s regulatory environment, identifying the regulator as “both pragmatic and sophisticated”, while the market itself has “a very good record of adopting new structures and set-ups fast".
Just as importantly, Bermuda has repeatedly shown a willingness to rethink traditional business models. The development of catastrophe bonds and the expansion of the ILS market demonstrated the island’s ability to create entirely new forms of risk transfer. Henley sees parallels with today’s AI evolution.
“Bermuda’s obviously got a very strong track record, and reinventing the business model is what I would say Bermuda’s known for. We domiciled headquarters in Bermuda for all these reasons,” he said. "We see what Bermuda’s managing; what runs through its waters are some of the biggest, most complex, difficult risks in the world. That’s what we’re targeting, the tough stuff.”
The vision
Ultimately, Henley believes AI should not become another distraction from insurers’ core strengths. Instead, should push insurers deeper into what makes them different.
Drawing on his experience as a former CIO, he made a clear distinction between what should and should not remain proprietary: “There are some things that are strategic to your business. It’s the pricing, the underwriting, the adjusting, the client relationship. You get them right, and you’re winning.”
Alongside those capabilities, however, “There are some things which are not competitive advantage: running the back office process.”
His conclusion is straightforward and the driving force behind the vision: “Build where it creates competitive advantage as that’s your secret sauce. Anything that isn’t that can be bought, it doesn’t have to be built in house.”
For Bermuda, that philosophy feels familiar. The island has rarely sought to compete by doing everything itself. Its success has come from identifying structural shifts in global insurance, creating an environment where innovation can flourish and bringing together the expertise needed to turn new ideas into functioning markets.
If insurance-specific AI follows the trajectory of previous innovations, it might prove to be the latest example of Bermuda doing what it has done for decades: providing the place where the industry’s next chapter moves from possibility to practice.
- - -
Martin Henley is founder and CEO of mea Platform. To find out more about mea Platform, visit www.meaplatform.com
Read the full Bermuda:Re+ILS Annual 2026 here.
Did you get value from this story? Sign up to our free newsletters and get stories like this sent straight to your inbox.