Insurers face a world of liquid risk and great uncertainty. But they’ve also never had so much expert help to call upon.
Conference season was busy this year, but full of excellent conversations with brilliant folks. With a little time to catch my breath, my major takeaway is this: insurers have some incredible challenges to solve, but they’ve never had access to so much expert advice to help them solve it.
It’s cliché to say that insurers are facing challenges like never before (and, more to the point, we’ve been saying it for the last decade). But the nature of risk in 2025 is different from 2005, or even 2015. Risks are more global, interconnected, and difficult to forecast; they’re less faithful to historical trends and more grounded in rapidly shifting technological, environmental, and social conditions. In other words, risk is liquid now.
A World of Liquid Risk
The new nature of risk is fueled by a number of converging factors: climate upheaval, geopolitical conflict, trade wars, and a frayed relationship with trust and truth, thanks to deepfakes and social media algorithms. As these factors compound, the past is less reliable as a predictor of the future. Liquid risk factors are interrelated, constantly changing, and unsuited to traditional actuarial models. Insurers are being challenged to use new technologies to confront these risks in near-real time.
Wildfire insurance is an illustrative example. In Celent’s plenary session at ITC, we spoke with experts in modelling wildfire risk in southern California, particularly the changing weather patterns that have shortened rain seasons and dried out foliage. Understanding these heightened risks relies on forecasting how climate changes affect conditions on the ground—what places become more vulnerable and how, and how to model what may happen in the future based on what we know about the present.
In some ways, this is how insurance has always worked: we’ve never had full statistical certainty, even when risks were more faithful to historical trends. We model risk, understand it with input from experts, and use that understanding to make conjectures. Conditions are changing more rapidly, which makes things more difficult—but also, at the same time, we’re gaining access to data that’s fast enough, accurate enough, and precise enough to make these predictions in close to real time.
Enter the Experts
It’s a monumental challenge. But new tools and new data sources give insurers new ways to manage risk. A small army of data streams, analytics tools, and modellers are available to help. In almost every conversation I had at every conference, I was struck by the sheer quantity and quality of expertise insurers can access. Potential partners are really smart, they really know the underlying risk and how to model it, and—crucially—they also really understand insurance. That point is worth emphasizing. Prior innovation waves treated insurance itself as the problem to be solved. Disruptor insurtechs tried to disaggregate traditional insurance sales models, processes, products, and actors. Most of these struggled to reach profitability and some have exited the space or been acquired.
Today, the insuretechs that are attracting investment are the ones who have worked with insurers, brought their deep knowledge to the partnership, and helped carriers solve problems using better technology, more sophisticated analysis, or more modern tools than those insurers otherwise would have had access to.
Toward the Future
The future of insurance is more specific, not less—insurers solving discrete problems with very targeted solutions, bringing to bear the expertise of people who understand both these problems and insurance more broadly on a very deep level.
We see this starting to happen already, as insurers turn to underwriting workbenches to give them better tools to make specific decisions around risks, or use externalized rating tools to maximize profitability across a book of business. And we expect this will accelerate as agentic AI starts to dismantle what we think of as traditional software, inserting agents that can invoke algorithms, services, or other LLMs into traditional processes.
No discussion of expertise would be complete without talking about insurance’s ongoing talent issues, and I think AI can help here too. As expert underwriters and claims adjusters close in on retirement, AI can help gather their expertise into documented rulebooks and guidelines, then serve as a resource, trained on all that knowledge, for the next generation. This aligns neatly with one of the key challenges insurers have faced as they embrace AI themselves—it’s not hard to find AI talent per se, but finding AI talent that understands insurance as a business is very difficult.
These are challenging problems, and it will take smart people using smart tools to solve them. Luckily for insurers, our industry has a world of experts.
