The Agile Insurer: How AI, People and Strategy Are Reshaping Insurance | 8 Industry Leaders

Speaker A: Foreign.

Speaker B: My name is Juan de Castro and you're listening to Making Risk Flow. Every episode I sit down with my industry leading guests to demystify digital risk flows, share practical knowledge and help you use them to unlock scalability in commercial insurance. And uh, one of the initiatives you led probably three years ago was, I think you called it, how we worked. Tell me a bit more about what was it about? Like what was the trigger for that initiative? What were the pain points that you were trying to solve?

Speaker A: So Vince came up with how we work about the end of 2023 and I think what we realized was as we were looking to grow our North American operation, as I mentioned, standing up new products, areas where we were subscale, we needed to do that very, very quickly and very, very consistently and in a scaled fashion. Similarly, we're also starting to see in the international space, Lloyd space, a lot of facilitation of business distribution, changing, more MGA's, more delegated business. And so it became very quickly that one size fits all approach to the operating model would not be sustainable as the market evolved and in particular now as the market is transitioning to a bit of a softer marketplace. So this company grew up early 2000s, grew up in a hard market, insurance and reinsurance, a number of years ago, quite siloed. It grew quickly. Everybody created their own everything, their own ways of working and their own platforms, spreadsheets, et cetera. And so we realized that the way to take the company to the next level and truly scale and drive that shareholder value was to reduce the fragmentation, to operate more as an integrated team, to bring together these functions that had operated in quite sort of separate and disparate ways with single threaded accountability. So that was really the impetus and the driver. I ran how we work as part of my role as also CEO of re. But it became quite obvious quite quickly that this was a full time job and that the only way to truly be able to reimagine the operating model across these vast changes that only continued to come was to create this world.

Speaker B: When you took the case of issue and the turnaround, what were the pillars of like the investment thesis? When did you think you're going to out compete much larger insurers in the market?

Speaker C: Well, look, I think there are a couple of things that we anchored on quite quickly and uh, I guess the first one was culture. So I mean Esure was a ah, 20 year old company when I joined it. It originally been very, very innovative. It had been in the vanguard of companies that had moved from broking to telephony. So, so it was uh, like Direct Line at the time. It was a pioneer in terms of telephony based insurance and it was also one in the vanguard of companies that had gone into price comparison websites. So it had an innovative heritage but I think it had been pretty underinvested in for the period immediately before I joined the company for many reasons it hadn't properly adapted to digital and culturally it was really, really quite introverted. It was very proud, very introverted and quite slow moving. And I guess I had a sense of trying to create a company that was humble, but humble from a perspective of looking out at the world with a big ambition. So we did a lot of work initially on culture and also on mission and purpose. We set up purpose for the company of fixing insurance for good. A sense that we needed to do something that was not just about making money. We had a uh, transformational zeal to make the process of insurance better for customers. We felt there were a lot of slow moving competitors with monolithic complex legacy technology stacks and if we could create something that was fast and moving that would be a way that we could outcompete uh, so we really set ourselves the mission of becoming the preeminent digital insurer in the uk.

Speaker B: Two follow up questions, one on the technology stack and one on the um, agile point you just mentioned. So when you were thinking about the technology stack and making that brave decisions, right, even if they were the right ones, you were not following the herd in those decisions. What was driving those architectural choices? Were you trying to drive automation and um, lower cost? Were you trying to drive better insights and data analytics for pricing or a combination of both?

Speaker C: You've touched on certainly a couple of the reasons. I mean I'd grown up in large complex companies with large amounts of legacy and I thought there was a real benefit in being able to have all your data uh, in one place, dynamically updated on a real time basis. I'd never managed to achieve that because the complexity of the legacy and the complexity of the business that I'd previously been in just didn't allow it. And we felt that if you could do that you would just have huge insights in terms of how you price business, how you create propositions, how you tweak customer journeys, how you make decisions and claims. So that sense of having a uh, really integrated dynamic data set we felt was really important. The second thing was, and some of this comes from my days with Cytora, working with Fintech, we are conscious that there's Huge dynamism in terms of the kind of fintech type offerings. And we sense that that would continue to innovate. So we wanted to have a technology stack where you could plug and play different components.

Speaker B: I think you declared a really interesting ambition. Right, which is you want Markel to be the preeminent specialty insurance company on the planet. Which, I mean, like, which is, I think nobody would argue that should be your vision. And Markel is perfectly placed to become that. But, like, what does it mean? How would you measure? How will you know whether you're coming closer to that goal?

Speaker D: Well, I'm lucky enough to have been brought across here by our CEO, Simon Wilson, whose vision this is, to be this preeminent specialty insurer on the planet. And he's already kindly, he's defined what that is to all of us. And he talks about four things when we're trying to build that. And we are not there yet in being the preeminent special issuer on the planet. We are building it right now. And four things he talks about is one, we must obsess over our customers. Insurance has this incredibly convoluted value chain where sometimes you sit quite far removed from your customer. The best specialty insurer, uh, will absolutely think about the person buying the policy and the person we are helping out when things go wrong. I think it's so often we forget the brilliant role insurance plays in society. Our economic role is to help people out when things go wrong. So when you are building the preeminent specialty insurer on the planet, you've got to obsess over those customers and remember that they're the ones who pay you for that promise to help them out when things go wrong. So really, having that obsession around the customer is important.

Speaker B: And you would argue this is important in any industry. But I think specialty insurance, where you are, uh, dealing with really specific and complex risk, it's even more important, right, than in any other business.

Speaker C: Yeah.

Speaker D: And I think that is the most important thing a business can think about is their customer. When I was a consultant, you always start a problem, start with a customer, and if you think about the customer, then you'll often make the right decision. But I think you can't commoditize specialty insurance in quite the same way you can do in other PNC classes. And so therefore, by definition, you are having to think about the niches and the specifics of your end customer perhaps more often than in other products in that insurance ecosystem. And then linked to that is the second point really is around the Best expertise. So really making sure that in these specialty classes you've got the best minds thinking about how to underwrite it and thinking about how to handle the claims. The third thing that Simon's talked about, which I think is kind of the most exciting right now, is speed being critical and exciting, uh, right now just because of what these new technological changes going on across the world right now are allowing us to do, we are able to move quicker not only in the way we transact our business, but also in the way we develop those solutions. But speed of answer, um, whether it's to a broker bringing us a submission and they want to quote quickly, or it's a customer needing a claim, get paid, we pay it quickly. That's the best kind of answer. And so to be the preeminent player, you have to have speed at the forefront of your service offering.

Speaker B: So we're also data aging on the first one, on the real case studies of how AI uh is actually driving value. Have you been able to answer this question? What have you heard in the last couple of days around what are the ingredients of actually driving value at pace?

Speaker E: What's the right way of approaching an A.I.

Speaker B: uh transformation?

Speaker E: Yeah, so I think the most interesting one that I saw was the one that was announced yesterday with Travelers and OpenAI. Right now they have been able to take 90% of all of their first notification of loss calls done for auto claims and use an AI voice agent developed in partnership with OpenAI. And uh, they picked auto claim specifically because it's a high volume. Travelers, uh, deal with 1.5 million claims a year. And so there's a lot of data involving to go out there. So that was the reason why they picked that. And then they've worked very much integrated with the business, getting the business leaders involved in this process. And that is a case study where I thought that was interesting to see how they've been able to at scale now. And they started out in eight states and then they've now after just a couple of months of testing, they was able to roll it out nationwide.

Speaker B: And ultimately all these initiatives require change management.

Speaker A: Right.

Speaker B: Require the clinic adjusters, the claims teams to feel comfortable with the recommendations from those AI driven decisioning. So I guess the involvement of the business is not just to best define the operating model, but so like to feel comfortable with how it's working. Right?

Speaker E: Yeah. And then I think there's the two other things which came up when one of my conversations with another tech company on stage here was a the notion and the person I was speaking with had a background from Facebook and as she started off saying, well, Facebook was very famous for, you know, move fast and break things. And many people don't know this actually, but they actually changed that statement to was it move fast on stable infrastructure or something, but not breaking things anymore. And the sexy thing in the industry is just let's, let's go quickly, let's move fast. But when you look at the big projects that gets done, it's actually a bit the opposite. It is you spend more time upfront thinking about, okay, what is it that we need to get done here? Who are the people we need to have in the room, what is the data we need? So you're actually thinking slowly and then you have a much more rapid execution because you've just gone through and mitigated for all of the things that you really don't want to come up once you're sitting and deploying and then that's really then holding you back.

Speaker B: One last question, just building on that because most players today are thinking about like, okay, how do they prepare their workflows, technical architecture for that future you are describing, right. And I think one thing I see is, I think that there's a risk that in this world of AI, ah, we are still building workflows that are human first. So how do you advise your clients in thinking? Okay, well most likely for the next months or whatever period of time, whereas you get more comfortable, you still want an underwriter to sniff, test or to look at every risk. But over time the trend you cannot negate, it's going to increase the automation. So how do you think about how they should be thinking about their, I mean at a very high level, that's their thinking in the architecture.

Speaker F: So I mean there's the business architecture and then there's the technology architecture. So fundamentally AI will over time shift the way that people think about technology architecture. At the moment, technology architecture is predominantly structured around processes and around human LED processes fundamentally. Actually, if you look at genuine kind of pure end to end automation within kind of services based businesses rather than product based businesses, there is a huge amount of, actually a lot of it is about decision making. And what generative AI can do is help make decisions. It's not just about automation, it's not just about taking a boring process and doing it quickly with a computer. And ideally more accurately, it can make decisions, it can make choices. Again, are we there today in a way that you would just let go without thinking about it? No. Is that of are we getting There rapidly.

Speaker E: Yes.

Speaker F: So one thing is around the technology architecture is actually the technology architectures will shift to being process aligned, to being decision and capability aligned. Much more so. And that is a very different architecture from today. And I'm very conscious that people are starting, they're not even starting in today's world, they're starting two generations back mostly. So that's not only do we help people look at where the future is, but go, okay, but where are you today and how do you get there?

Speaker B: Great. You mentioned that there were two use cases. I stopped you, interrupted you several times on the first one. What was the other one you had in mind?

Speaker G: Yeah, I think this one is particularly fascinating because one of the things that we've seen in the market is that the method of fulfilling a request. So let's say a new business submission is tightly tied to the channel. So for example, email is, you know, associated with human underwriting. Straight through processing is associated with completely automated digital ratings. And I think one of the big unlocks here will be the decoupling of channel to the method of fulfillment. So for example, with agentic AI, what you can do is you can actually assess, irrespective of the channel that the risk came in, the most appropriate path for it to flow down. So even if it comes on the email channel, you're able to assess is this digitally tradable. You're able to assess the data points required for that digital trading route and then you're able to extract that data from the unstructured documents and fulfill the requirement there. Similarly. Actually, one of the main challenges with straight through processing is, well, there's actually two main ones with STP which is often spoken about. One is it's like a one shot process. So there's a set of data points that you need and if you have the data, you can get a quote. And if you don't have the data, you can't get a quote. So imagine a situation whereby that back and forth that we spoke about earlier could actually be enabled for the STP channel. So Juan sent something in on email and there's one data point which is missing, which is critical. Instead of saying computer M says no, we can't give you a quote, Juan, what we could do is we could say, Juan, just give us this one data point and we'll get you a quote. And because we have this state or memory, we can actually update the risk object and get the quote for you.

Speaker B: So we think five years out, it's 20, 26 now we think, let's say by 30 30. What will the kind of standard workflow in underwriting look like?

Speaker H: So I'm not sure if this is a disappointing or exciting answer, but a lot of what we're talking about, the potential of AI, should be realized by then. And what I mean by that is an underwriter's day to day should look fundamentally very different than how most commercial underwriters are spending their time today. So what does that mean? They are no longer receiving an email for a submission and having to review it before sending it off for kind of centralized processing and intake. Instead they're going to just be receiving an output from the workflow up front of the underwriter, where it is a, uh, fully assembled submission, it has been enriched with data that was supplementary to the actual submission received, and even a recommended action for the underwriter to consider or next steps to take should be clearly outlined and visible for the underwriter. So the underwriter is fundamentally shifting from data assembly and retyping of data into really thinking purely about pricing, nuances, exceptions, and then broker relationship management outside of the transaction itself.

Speaker B: So one final question to wrap it up, Nick. Obviously there will be a number of people listening to this episode who have set up or are setting up a broker. Uh, they are thinking about a potential exit in the next year, 2, 5, whatever to a private equity firm. What advice would you give them now? Or almost like what are the things you're going to look for when considering the deal?

Speaker I: Yeah, I think it goes back one to a couple of the fundamental tenets. Defend your book. That's the demonstration that you're adding value. So really double down on account retention. And at the same time, I think what's important if I were running an agency today is to identify one or two or three areas on the servicing side where AI over time will have a meaningful impact in my cost structure. And so not necessarily to drive down the cost today or tomorrow, but to have the flexibility to adapt to an environment where the revenue profile of a broker might come under some pressure. So those are the two things that I think are I would focus on if I were running an agency and I wanted to be resilient and I wanted to demonstrate that I'm forward leaning.

Speaker B: Making Risk Flow is brought to you by Cytora. If you enjoy this podcast, consider subscribing to Making Risk Flow in Upperclass Podcast, Spotify or wherever you get your podcast so you never miss an episode. To find out more about cytora ah visit cytora.com thanks for joining me. See you next time.