Investors Are Stepping Up to AI

MATTHEW SCALLY: Very similar to the past, insurance and specifically insurance distribution remains an attractive and active M&A segment. Retail brokers alone accounted for north of 70% of deal volume over the past five years within insurance-backed PE. It’s being invested behind by financial sponsors and strategics, and there remains, we think, a viable pathway to IPO for select brokers.

But, as you may have seen or alluded to, there have certainly been variations in trends, variations by geography. For one, the U.S. being the largest and fastest-growing segment over the last five years in transaction value, and that remained true in 2025, with the U.K. following close behind in terms of CAGR [compound annual growth rate] by transaction value. Europe has actually declined over the past five years. So you see some of that variance by geography.

And then by archetype as well within insurance brokerage. In 2025 we did see a bit of a decline in total transactions and total transaction value. Most of that is driven by roll-ups: target availability, some element of timing, and of course changes in the credit cycle over the last few years. While several larger scaled transactions were closed over the past 18 months. And then in adjacent areas; think of the MGAs, for one, think of some of the TPAs, claim service players in the market. Those were actively traded over the last few years; and those continue to be a hot topic in 2025 and 2026 for many of our investor clients.

SCALLY: Yeah, there’s certainly been a bit of a pullback in valuations, some of it driven by AI chatter, noise, or potential impact. But the insurance brokerage space remains a very attractive component for any investor. It has typically lower capex expenditure, high margin, high recurring revenue, a component within the value chain that has been taking more and more of the economics. So the basic fundamentals of a strong investment remain.

Now, the complication around building a business that can report at a public level, the ability to be integrated, ensuring you have succession and perpetuation planning on your front line, all of that is critical. And I don’t mean to minimize those hurdles. But because of the foundation of the basic business model, certainly I see an IPO path for select brokers.

GRIER TUMAS DIENSTAG: AI is top of mind for every investor with whom we speak, ranging from investors in software businesses looking to understand how that will affect valuations to investors in distribution businesses thinking about how might agentic commerce change the purchasing journey, and investors in most businesses considering how the cost basis will evolve with the tremendous opportunity with AI.

These are just a few examples, but it’s to say that every one of the diligences that we are doing, as well as a significant amount of the portfolio company service that we’re doing, has AI as a major element. If we step back and think about how AI is evolving, we describe the AI staircase in our recent article—three steps that we see the sector progressing along.

The first one is around traditional AI. Think about the predictive data analytics that are already established, given the immense pools of structured and unstructured data that exist in insurance. Applications here might include fraud detection, pricing, risk modeling.

Then we have the next step on that staircase around generative AI. Here we’re thinking about those document-heavy tasks like policy issuance. We’re scaling demand letter generation in claims as an example. Many carriers are going on this generative AI journey.

And, then, the last step, for carriers as well as many of the others in the ecosystem, is this emerging frontier of agentic AI. We’re still absolutely in the early innings of understanding the potential here, but we recognize that agentic AI has the ability to transform end-to-end workflows. We are seeing some of that in claims, as well as in risk assessment, and in the purchasing journey in the direct channel, as examples.

SCALLY: Investors, in terms of their diligence process, have grown as sophisticated as some of the AI technology has. As AI technology has jumped leaps and bounds over the last 18 to 24 months, so has the process and the use and the really nitty-gritty that a lot of investors get into. If you go back 18 months or two years, many investors would have a simple AI checklist that they would go through when assessing a company, bucketing things into risks and opportunities. Now, many investors have a very detailed, AI-driven value creation plan that they want to engage with the management teams on. It’s critical those management teams have perspectives and even at this point have some initiatives ready to support it. Our view is, management teams certainly don’t have to have AI solved and their strategy perfected today, but they certainly can’t bury their heads in the sand and assume it’s not going to disrupt elements of their business, be disruptive to their industry, their client base, and their workforce.

SCALLY: Well, investor interest I don’t think has waned at all. It continues to be exciting. But if you’re asking about the AI impact and investors’ perspective on how that is going to interact with or change the business model, that is certainly evolving and developing as we speak.

I would say some of the successful brokers we’ve interacted with have a clear view of where they want to create value with AI, whether that’s generative AI on a lead generation standpoint; whether that is large language models evaluating [and] comparing policies to ensure compliance, to ensure risk structures are solid, to ensure that exclusionary criteria isn’t off base. All the way down to enhanced consulting services for loss and risk mitigation that brokers should be in the space for. So all of that has been moving.

My perspective is retail brokers and insurance as a whole have probably been slightly behind other industries but [are] catching up very quickly. A majority of the top 20, top 25 brokers are private equity-backed. Those board members are getting pushed by their partners and their investment committee to ensure a strategy is in place, and that’s flowing down to management teams. So we see an acceleration on that going forward.

SCALLY: I think there are certainly differences. I mean, one, you can talk about the amount of delegated binding authority or MGA business that’s now embedded within wholesalers, and that takes on a completely different agentic underwriting lever that needs to be pulled and needs to be evaluated.

On the traditional wholesale side, I think there is going to be a really important piece to ensure capacity, capacity relationships, are secured and new ones developed over time. And AI is very much going to have a role in that.

Whereas on the retail broker side I think we’ve seen some softness in the rate environment, some headwinds as it relates to organic growth, and on that side of the coin they’re really going to have to use digital lead generation and gen AI to supplement and drive growth.

So, I do think there are certainly differences between those two business models. But a lot of the back office and some elements of the servicing, efficiency gains and accuracy, those certainly apply to both business models.

SEBASTIAN KOHLS: I think on MGAs what’s more interesting is going deeper on the underwriting side. Especially with agentic technology gaining more maturity, a lot of the processes of intake, segmentation, triage, etc., all of that can really be increasingly automated.

For sponsors that own multiple MGAs and think about platforms on which to onboard to gain scale in the investment, there’s a lot of attractive opportunities to gain efficiencies. I think it does also mean, though, that the focus will increasingly shift even more on retaining your competitive edge and really understanding a particular class of business. What is your insight into how precisely can you price the exposures in that segment? Do you have loss data to do that? So, I think we’ll see a shift, but MGAs that know how to price business that others don’t will remain valuable.

KOHLS: Yeah, I think you’re going to see a lot of disruption. It goes back to what is actually your competitive value proposition. You see certain models where they’re built around the digital flow more than anything else, that I think can be disrupted now. If you had an easy portal, an easy way of submitting, and you had a workflow that was built in a very efficient way, that’s quite easily disrupted through AI today.

Other moats will remain defensible. If it’s, “I understand a certain type of risk better than anybody else. I have access to the capacity that is interested in that risk. I have access to the distributors that get me to the customers that need to place the type of risk,” then those types of core competencies are not really subject to disruption through AI. I think we’re going to see a separation between those that really are risk specialists and those that may be less so.

TUMAS DIENSTAG: When we talk about software providers, there are absolutely differences as we think about the range of players that exist in the market. The impact will vary significantly for a core tech provider building these solutions on their platform, an estimatics player further extending into underwriting, and a cloud native startup trying to displace legacy players, to name a few examples.

Recognizing these differences, one major is from a monolithic, legacy system focus—think about going from wanting everything to run on my legacy system to a platform that is much more open and modular. Clients are getting very excited building an agentic mesh where multiple solutions and eventually multiple AI agents can interoperate via APIs.

So, as we think about the winners and losers in this, we consider who is going to be able to create that connective tissue that allows carriers, and also brokers, to be able to plug various specialized best-of-breed tools into these systems without much of the time and energy that goes with the integrations of the past.

KOHLS: We’re also hearing from a lot of carriers in that space. That means, of course, that the bar goes up on what you need because increasingly the value proposition of, “I am just the person who can stitch different things together for you and I am a single provider,” that goes down in a world where carriers have a greater capacity to just really pick the best of breed. So I think you’ll see that pressure on being truly distinctive, whether it’s in claim servicing, whether it is in underwriting.

SCALLY: I would say first you can glean some learnings from other industries that are a little further ahead, many of which have tried to make bets on things that they fundamentally believe will differentiate their service or their product for their clients, meaning being able to build and own that capability themselves. While on the flip side of that coin [others] have looked for things like process efficiencies to potentially be outsourced, to understand that it can be table stakes or just good enough. And there’s a view that they’re not in the best position to potentially build those capabilities and solutions.

I think intuitively that makes a lot of sense for the insurance industry and specifically the insurance distribution space. If you’re an MGA with the best access to very specific data on your programs, your industries, and your risks, and the overlap between those, it probably behooves you to build out the agentic AI underwriting copilot or solution that supports your people. If you are an insurance broker, I think you want to make sure your go-to-market strategy, your lead generation is unique. Your marketing efforts tied to generative AI are very specific.

Your ability to drive loss or risk mitigation controls through AI are specific. Your ability to control FTE costs, to develop efficiency solutions in the back office, or potentially do things like process claims—probably not as differentiated. I think that’s how the tree will shake out over time, over the next several years. It’s certainly in line with the early to midterm conversations we’ve been having with many of our own clients and others within the ecosystem.

KOHLS: On TPAs, there’s a lot of potential in principle to think about claims processing, where you could imagine significant productivity improvements both in the intake and the processing. I think similar things that would apply to a carrier apply there as well.

The open question then remains, are the incentives actually there for the TPAs to realize these efficiencies? It’ll all come down to how carrier relationships will evolve. Because conversely, carriers are also worried about the TPAs effectively agentifying their own systems and the carriers not gaining in that value.

Today, I think a lot of those structures we do see, it’s more compensation based on unit costs, some version of FTEs. I think if you take that lens, the incentive isn’t really there for the TPA. We’ll need to see new structures emerge so that you see some version of sharing of the value that’s being generated so that incentives are aligned.

The other angle you could take is, as a carrier, it may be that the old value proposition of a TPA is just no longer quite as valuable as it used to be. If I can agentify more of my own processes as a carrier, I need a TPA less as a pressure release valve. If I outsource something to a TPA because they have an advantage in understanding complexity, that may be less so if I again can bring more of that in-house through the use of AI. So that’s another open question on the TPA aspect.

TUMAS DIENSTAG: Today, TPAs play an enormous role in our industry. There are hundreds of thousands of employees associated with TPAs, and they’re critical for carriers as it relates to extra capacity that’s needed, expertise in specific areas or types of claims, as well as in specific geographic areas. Those TPAs are even more important for self-insureds who rely on them oftentimes for all their claims.

As we go forward, there first and foremost is a cost imperative for these TPAs. We all recognize that in businesses that have tens of thousands of people and rely in many ways on document review, summarization, customer service, there are many potential use cases. There will be an opportunity for AI, and specifically agentic AI, and that will reduce the cost for these TPAs to serve their customers. We feel that the high-performing TPAs will go on that journey.

They’ll be able to take that cost out by reshaping entire journeys and workflows, and the economics will adopt accordingly. The next step, though, will be to say where is the differentiation and what sustains the industry over time? There, we will see TPAs use the expertise and data that in many cases they already have aim to exceed their customers’ performance on quality, i.e., the accuracy at which claims are settled, and customer experience. There are also new service lines to explore. In summary, there is lots of potential…and a need for these TPAs to play both defense and offense in the coming months and years.

SCALLY: I think the differentiation point, as we talked about a little earlier, is certainly key and topical for many of our insurance clients, and frankly our firm’s broader client base outside of the insurance industry is thinking about the future. And the future is six months and 12 months from now.

Grier alluded to the AI staircase. I think the technological bar has certainly been met by many AI tools, i.e., the efficiency gains that can be provided, the ability to access data, the ability to impact your own product and services. Agentic AI, generative AI, large language models, and processing, I think we’re there.

We’re going to start to evolve very quickly with winners who can manage their workforce to adopt and engage with these solutions. Thinking through the incentive base for your own employees to utilize this, reducing the barrier and friction in their own workflow so there is not a large, massive change versus an embedded solution that they can access, I think that is going to differentiate winners from those who are at least stalling in terms of getting real gains from AI adoption.

KOHLS: If you were to look today at carriers, you look at brokers, you look at other members in the ecosystem, I think everywhere you would say there are maybe one, maybe two examples that have not only demonstrated there’s value in theory, but they’ve actually scaled it and have really transformed. Clients, most come to us and say, “I know it’s possible but I haven’t been able to do it.” I think as you look into the end of 2026 and the end of 2027, we’ll probably be in a place where it’s no longer one or two, it’s four, five, or six. And if you’re not one of those four, five, or six, you will probably really struggle.

TUMAS DIENSTAG: Hopefully you have heard several real challenges, as well [as] several really exciting sources of differentiation that we are already seeing emerge in the industry. What we are counseling our clients on today is very much the need to have the AI strategy, the roadmap, and the path to get going and also scale out of those early-stage pilots. Time is of the essence. We recognize that we don’t know everything today, but having that clear roadmap and building the capabilities that are required are really important right now.