GenAI 2025
Of all the insurance and insurtech highlights of this year, the rise of generative AI (GenAI) has been the most compelling. As we look toward 2025, critical questions remain about GenAI’s long-term viability, cost efficiency, and in particular, whether it can replace its human counterparts in claims review, underwriting, and more. Special guest Alex Zukerman, Sapiens’ Chief Strategy Officer joins host Pat Ryan, Sapiens’ Marketing Director, to discuss what lies ahead for GenAI in 2025 in our latest podcast.
Pat Ryan|Alex Zukerman
Pat Ryan: Hello and welcome to the Sapiens Insurance 360 podcast! I’m your host, Pat Ryan, Marketing Director here at Sapiens, and I’m so glad that you’re out there listening. This is where we discuss the latest news, trends, and issues from across the insurance solutions and technology spectrum. Well, it’s hard to believe that another year is winding down, and as we look forward to the promise of 2025, we’re going to close this year out with one more podcast on the topic that’s probably been the topic of the year, which is Generative AI (GenAI). This week, we’ll be talking about what we can expect from GenAI in 2025 and how it will affect insurance and insurtech. What trends, innovations, and potential conflicts may arise as GenAI grows more sophisticated? And as GenAI grows more widespread and cost efficient, will its automation crowd out its human counterparts? Today’s guest is Sapiens’ own Alex Zukerman, Chief Strategy Officer, who’ll help us answer these questions and hopefully more about what we can expect from GenAI in the new year. Earlier this year, Alex spoke with our partner Microsoft about its partnership with groundbreaking AI research and development company OpenAI, which has helped bring innovative GenAI models to market and supports research into how these technologies can benefit businesses and society. I’m excited to continue the GenAI conversation with Alex today. Alex, welcome back to the program!
Alex Zukerman: Thanks, Pat, and it’s great to be back!
Pat Ryan: Great. So, so let’s jump right ahead, shall we? To start, how do you think GenAI will reshape the competitive landscape of the insurance industry in 2025? And in particular, underwriting, risk modeling, and fraud detection?
Alex Zukerman: Yeah, I think it’s a great question, Pat, and I’m quite certain that GenAI will be a dominant change agent in the insurance industry in 2025. I think the capabilities, look, when we talk about the insurance industry, the first thing, one of the first things that comes to mind is heavy on documentation materials, text, data pieces, that all needs to come together to form an opinion. An opinion can be how to underwrite a business, how to assess the risk, how to take decisions on fraud, how to take decisions on onboarding customers, how to provide service to customers. And the more the industry matures, understand that there’s no one size fits all and you need to be personalized. GenAI can be a great, great change factor in this. And I think that so far, we’ve seen only the top of the line, the tip of the iceberg, the incredible capabilities that are going into the industry. And if you want, I think it’s in those areas that you mentioned, the impact will be specifically critical. Let’s look at an agent trying to sell or to engage with a prospect or a customer as is exposed to huge amounts of data. GenAI can help him understand the data in a fraction of the time, understand who is the right customer for him to put focus on, help him to provide questions and answers to his customers. There’s no more the FAQs of the world and all these big databases of potential questions and answers. The answers can come in your language, on your terms, in the right moment. And same goes for with of course, with a little bit of a change when we look at the underwriting. So let’s look at an underwriting, especially when you go to these heavy underwriting activities on specialty lines, on ENS, on large commercial, those are complicated cases, typically not straight-through processing, not streamlined, and it’s heavily based on the specifics of the case, typically with a lot of documentation and the ability to summarize a document, to compare it to other documents, to extract the right piece of data from the document in a fraction of the time. But also to start querying about the customer or the risk object to get the history to compare the history. It always gives operation on steroids to the users. And imagine the similar logic is applied on claims as well. So the way I look at it, is it’s activity on steroids. It’s not replacing us as users, as people riding the industry, but definitely providing huge automation and enhanced capabilities to perform the task.
Pat Ryan: Great. That’s a great answer, really comprehensive. I appreciate that. So when you look at next year and really beyond, I mean GenAI is going to be around forever, it’s not changing. Will it get any easier or how will it get easier for insurers to integrate GenAI into their legacy systems or will it?
Alex Zukerman: So I think it’s a very safe assumption, let’s say, to foresee that it’s going to be dramatic to the way we write business in the industry. However, I question very much the ability to use it on legacy systems. It requires some capabilities that there is definitely a preference for a modern system with open APIs, with microservices, with a well-structured database that allows you to access it easily with ETS that allows you to access all the data in an insurance company on a single data warehouse or data lake, and thus provide a single version of the truth that you want to run your GenAI on. And here in the insurance business, we definitely want to use the GenAI on this as closed capabilities, not as publicly open capabilities, but [as a] closed capability that is limited to the usage of the insurance company data and then needs to be, and the right usage of it is allow access. The GenAI has to have access to the majority of the data in the company, and this is very hard to achieve with a legacy system. This is one aspect. The second aspect, at least this is the way we see it at Sapiens, is the GenAI is not providing only those capabilities, but it’s actually changing the way we think that the human and machine interface will be in the near future, to be honest. So the way that people interact with the policy admin system or the claim system, the way configuration is done, the way that launching new products is going to be completely different with the capabilities of GenAI. And I’m having quite a hard time imagining how can you do it with a legacy system. Let me give you an example. We are streamlining configuration of our call system using GenAI. We basically can take a requirements document from a customer, from a business person in a customer, and with a one click through GenAI, turn it into system configuration. This requires also not only the smart capabilities of translating requirements to automated configuration, but it also requires a fully configurable system flexible, well-documented, well-structured, and easy to use. And this stuff is very hard to have on a legacy system. And if you want in few years, what we envision is, and maybe I’m exaggerating a bit, but all the screens of the system will shrink down to a single field of a search or command. And through this single field, you can execute any task you want on your terms in your time using your own language. And again, this type of evolution, it’s very hard to see how it comes with a legacy system.
Pat Ryan: Yeah, that’s interesting. When you think about the digital transformation agenda that’s been really pervasive throughout our industry in the past five, seven years, based on what you’re sharing, I would submit that those carriers that are lagging on the digital transformation side are going to find themselves even further behind on their ability to leverage the capabilities of GenAI for optimizing their business. Is that a fair assessment?
Alex Zukerman: I think it’s a very fair assessment. I think you know, to quote a well-known phrase that GenAI will not be a lipstick on a pig. GenAI is a strong capability, but requires partnership from the other surrounding systems. Especially when we are in our vision, we don’t see it limited to managing documents, but actually going to be an automation tool, a configuration tool, and a tool that helps you navigate through the system, embedded into the UI and the UX of the system, and actually becoming part of the engine and the logic of the policy admin system. I think that as you very rightfully said, the digital transformation is not a silo type of approach. It doesn’t provide you only ability to communicate digitally with your customers or channels. It’s not only about moving to the cloud, it’s about positioning yourself as a carrier in the world of the modern technology. And there are piles and layers that are coming one on top of the other, where the baseline is beyond the digital transformation stage. It means be on the cloud, have a strong API connectivity, be able to extract data easily into a central location and those types of things that will enable also the smart and the real usage of the GenAI and not just on a simplistic endpoint use cases.
Pat Ryan: Yeah, that’s interesting. Your commentary about the need for AI tools to interact with other systems, I think probably answers this next question I’m going to share because I think I have an idea of where you would go with this, but do you think or do you see GenAI growing so sophisticated in the months or even years to come that it might supplant human underwriters and adjusters and people in those kinds of roles?
Alex Zukerman: That’s a frightening thought to be honest!
Pat Ryan: Yeah, it’s a loaded question, Alex!
Alex Zukerman: I think this question is in debate across also all the companies and the generators of GenAI. And I think here, the place of regulation is critical. And I think here regulation is our friend and not our enemy, and we need to be very careful in how we implement it. And I think the way we look at it is the GenAI, at least at the moment, we focus on it being a copilot and not an autopilot. And together with the copilot, it means that it’s a tool that assists the underwriter to execute this job and not replacing the underwriter. Now of course, that what we foresee is such a huge efficiency, automation, and “work on steroids,” as we like to call it. Then with the same team of underwriters, you can multiply the business you can handle. But together with that, the responsibility lies with the human, the responsibility eventually to approve a final decision if even if the AI and the GenAI are actually doing all the legwork, the responsibility is on the human. And this is why we have also those approaches of a white box rather than a black box, meaning understand the steps of the GenAI. And by the way, it’s also right for machine learning and more traditional AI calculations running an algorithm that provides a risk score. You want to understand the steps and understand what was the reasoning for the machine to provide a risk score. When you get responses from GenAI, especially when it comes to risk scoring, underwriting, claim management, you want to see the location. Where did the GenAI take the logic from? Where did they find the answer? Point me to the right place. So as an underwriter, if I feel this is a critical point to my decision, I have the capability and probably also the responsibility to go and check for myself and ensure that the machine gave me the right response so the machine can do all the legwork, the calculation, the finding of the locations, bringing me the data in a structured way, et cetera, et cetera. But still, I have the responsibility as an underwriter to validate it. The more we get coherence from the machine, better results, less hallucination, less bias, we can give it a higher rate of confidence. But still, I’m very much in favor of this copilot approach that we are not replaced yet by the machines.
Pat Ryan: Yeah, I love that little nugget you threw out there. AI should be considered as a copilot and not an autopilot, which I think, if people kind embrace that notion, it eliminates some level of fear around human replacement. Right? I think that’s one of the big challenges with AI today is this notion that AI is going to take over, but that’s not the way it works. And I don’t think that’s the way it’s going to work. So I think that response is really strong. I appreciate your perspective there. I’m going to ask you one final question. As we look ahead, do you think GenAI could lead to an insurance ecosystem that is more proactive than reactive? Not to say that our industry is primarily reactive, but from a balance standpoint, do you think it will put us in a position where we’re focusing more on loss prevention rather than loss compensation?
Alex Zukerman: Yes, I definitely think it can contribute to that. And on this subject, and it’s [a] very interesting subject path to touch base and actually to finish this discussion. I think what we see in general, it’s [a] very, very powerful point that the role of the insurer is shifting from being a pure risk taker and protection provider into the additional spaces of prevention and service. So we see more and more, the desire to provide a one-stop shop that I can underwrite your risk, help you prevent it and provide a service, not only a compensation when stuff happens. And I think that the GenAI and AI in general can be a very strong contributor into this approach. To give you a couple of examples to this line of thinking, by better analyzing our customers and prospects who they are, where are their risk points, we can better align the type of insurance we provide them. When we better understand their targets, their motivations, and what’s behind acquiring this insurance, we can provide a better response to the need between those three pillars of protection, service, and prevention. And again, this starts in identifying where there’s a higher risk for leakage and asking customers to put a water detector. Similarly, a smoke detector, this is just one aspect of that. And then using AI, GenAI to really map where are the right places you want to focus on, when is this obligatory, when it’s nice to have, when it doesn’t have a huge importance. And in general, being able to understand better the data for our customers and then make sure we provide them the right answer. [It] can be a protection, [it] can be a prevention as well.
Pat Ryan: Really, really thoughtful feedback. Alex, I really appreciate it. I know we’re probably running a little bit long here on time. You’ve got so much to share. I know I certainly learned a lot from you today, and we all know that GenAI is really at the forefront of today’s insurtech ecosystem and it’s going to continue to raise the bar in terms of next-generation technologies. Thanks so much again for your insights today, and I know I speak for both of us when I say that I can’t wait to see what the next GenAI innovations will be and what’s in store for us in 2025.
Alex Zukerman: Absolutely. Pat, I want to thank you again for having me on the show. It was a lot of fun and I really join your passion in looking into 2025 and seeing this great progress of technology into our industry.
Pat Ryan: It was my pleasure, Alex. Best wishes to you and of course to all our listeners, a very healthy and happy holiday season and happy new year. As always, we love hearing from you. So if you have any comments or would like to follow us on social media, please reach out to us on our channels. And don’t forget to subscribe to the podcast. We’ve got so much more coming in 2025, so stay tuned to our Sapiens Insurance 360 podcast. Bye now!
