For years, insurers have invested in digital transformation, yet many still struggle to realize the full value of AI. Today, AI is not just another layer of technology; it is an opportunity to rethink how you operate, make decisions, and deliver value.
Many insurers are under pressure to accelerate AI adoption. However, organizations often move forward without addressing the foundational constraints that limit impact. The challenge is not the technology itself. It is the persistence of siloed data and systems, outdated processes and legacy operating models and mindsets.
When AI is applied to existing processes without rethinking them, it may deliver incremental efficiency, but not meaningful transformation. In some cases, it adds complexity rather than reducing it. A similar pattern is emerging across industries. Organizations adopt AI quickly, expecting it to resolve long-standing issues rather than using it as a catalyst to redesign how work gets done.
Early AI use cases often focus on efficiency, including automating tasks, extracting data or supporting underwriting workflows. These are valuable steps, but they typically deliver incremental gains. The greater opportunity lies in intelligence-driven modernization. This is where AI:
For example, rather than simply accelerating underwriting, insurers can rethink decision-making, combining human expertise with AI-driven insights to improve outcomes. This shift requires careful consideration. It introduces new questions around trust, governance and the appropriate level of autonomy for AI, particularly as agentic AI evolves.
In many organizations, AI ambition is moving faster than the underlying readiness required to support it. We often see insurers deploying AI without:
This creates a gap between ambition and execution. A common pitfall is starting with AI as the answer, rather than defining the business problem first. Without a clear view of the value you want to achieve, AI initiatives risk becoming isolated experiments instead of drivers of transformation.
A successful modernization approach balances vision with pragmatism.
This portfolio-based approach helps you balance quick wins with longer-term transformation.
The starting point for insurers is not technology; it is the problem they are trying to solve.Taking a structured approach can help:
Thomas Raushen emphasizes that data remains the launchpad and the foundational layer. Without accessible, high-quality data, even the most advanced AI solutions will struggle to deliver meaningful outcomes.
AI should be a core consideration in any transformation, but not the sole focus. Like cybersecurity or operational resilience, AI must be integrated into broader transformation efforts. At the same time, it introduces new considerations, including:
Taking a holistic approach helps organizations ensure AI strengthens their transformation rather than complicating it.
One of the most common risks is losing sight of the “why.” Large transformation programs can become focused on milestones, timelines and technical delivery. As a result, organizations may complete multi-year initiatives without achieving meaningful business value. AI adoption risks following the same pattern if it is not anchored in clear objectives and measurable outcomes.
Insurers that embrace these shifts—rethinking how they work, not just what they use—will be better positioned to unlock the full value of AI and deliver meaningful, lasting outcomes.
- Chapter 1: Why do many AI initiatives fall short?
-
Thomas Rauschen:
In today's episode of AI Uncovered: The Future of Insurance, we will discuss how insurance organizations manage the challenges between legacy transformation and modernization. Great to have you back.
As you can remember, in our last podcast, we established that AI momentum is real, but the value hasn't fully shown up yet across the insurance industry. Also, I think we discussed that AI won't reshape insurance at once either. So, that means that some areas, like distribution, software, and unstructured data, will move fast, while trust-heavy, complex insurance, like commercial reinsurance, will change more slowly.
But I think the most important takeaway that we discussed, Daren, was that the biggest limits aren't algorithms or tech when it comes to AI, but rather the foundations and the way we work. And following up, let's dive a little bit deeper into the modernization challenge and how the industry can move from legacy to innovation and intelligence.
And the first question I have is really as you know, across the industry, we hear the term legacy a lot, you know and sometimes it's all about like legacy tech, legacy data, you know, stopping companies to modernize or using AI in the most efficient way. So, it almost feels like that AI basically amplifies the weaknesses in legacy environments. But I want to step away a little bit, Daren, from talking about legacy systems and legacy data, but we also could talk about legacy operating models and processes or legacy mindsets. So, what is your view when it comes to legacy and why is it blocking companies in driving their modernization initiatives?
Daren Rudd:
Yeah, it's a really good perspective. And I think we talked about it a little bit last time, as well. When we look at AI, while it can potentially force us to really rethink things, it is just another technology. So, we've got to avoid layering AI over the top of those legacy systems and processes again. And I think it's not a lot of difference from some of the other digital transformation initiatives we've seen, at least in the near term, when we are probably looking at those efficiency-driven modernizations, trying to do what we do now but use AI to do it better, there's a heavier rethink that needs to be done to really drive the value out of AI. When we’re really looking at it from an intelligence-driven point of view. If all we're really doing is saying, okay, I'm going to do what I did before with a person and try and replicate it, and I do mean try and replicate it with a piece of AI, maybe going to make it a little bit more efficient, but I'm not really addressing why those systems look like they do or why the processes look like they do.
And, we've obviously got, as you say, a lot of legacy, and a lot of those systems today are difficult to get into the data, difficult to extract the data, difficult to actually join up those processes to make them work properly. And I don't think you know, AI is not going to ride in on its silver horse and solve everything. So, I think it is important that we think from a systems thinking point of view, not just silo. And the worry I see at the moment is that people are not really addressing the legacy or the way of thinking about it, they're just trying to layer over AI in terms of the way we do things today to solve a problem.
- Chapter 2: Efficiency vs. intelligence: Where should insurers focus?
-
Thomas Rauschen:
No, I do agree with what you see, and I just want to deep dive a little bit into the efficiency point, right? A lot of AI initiatives that we discuss with our client are really driven by efficiency-driven modernization. Do you think it needs to be more pivoting towards intelligence-driven modernization? Really using data, adjusting your processes, because as you just said, AI is supposed to help humans make better decisions, make their life easier, instead of only applying an efficiency lens.
Daren Rudd:
It's not a binary decision. I think there's stuff that's going to happen where you can see some immediate benefits in leveraging AI. Let's just talk about, and we're seeing this pattern quite a lot within different insurers at the moment. How do I make my underwriters more efficient? How can I use AI to get the data out of a document that's arrived so that I give that information more effectively to an underwriter rather than having to dig around and look through the document itself? So, I think that's a key win and an advantage that's but it's low level in terms of efficiency.
The bigger part, but the tougher part, is then rethinking how do I really make that underwriter more effective by buddying up with the AI. How much of the decision-making do I pass across to an AI to go down that route, and then how do I really emphasize the human value? And there's lots of debate at the moment around agentic AI and whether you would really allow that to go forward or not. There's a huge amount right now while we're recording this about Multbot and the ability to give these agentic AIs the right to go off and do stuff, huge security holes around that. So, you've got to be really careful about how much you've changed there, but it's harder to rethink through those current processes. So, I think there's advantage in going with AI for immediate efficiency gains, but I think the way you're really going to drive this out is to really rethink the harder point about how do I make people superhuman almost in terms of leveraging the AI, but to enhance what a person does.
- Chapter 3: Is AI ambition outpacing modernization readiness?
-
Thomas Rauschen:
And when it comes to what we are seeing globally, but also as you're located in the UK, in the UK market, in terms of what you're seeing in reality with your clients, and I'd be a little bit provocative now. Is AI ambition running ahead of the modernization reality? Basically, do we try to apply AI before we just said thought about the legacy challenges, be it people, be it data, be it organizational model? You know, do we apply AI too quickly?
Daren Rudd:
I think there are some organizations that are moving very fast and are maybe not fully thinking through the implications of AI. I worry a little bit that those organizations are assuming that AI is going to solve a lot of the problems that have been around for the last 30 years, at least since I've been in the market with old systems, legacy data and silo, it won't fix that. A lot of areas that we're working on at the moment are much more in terms of how do I free up the data, how do I make the data available, how do I break those silos down to feed AI initiatives, rather than running on the other side of it.
But I also think if we're not thinking more holistically, then the AI won't, as a single initiative, deliver the value. We already know that you know big transformations struggle. So, if I'm not really clear on why I'm applying AI, and again, there's a lot that I see in the market at the moment where people are assuming that AI is the answer now. What's the question? I think we need to be asking ourselves a little bit about what we're doing and why. It's very easy to get board-level interest and you've got to have something in AI, otherwise it's not going to happen.
I think really we should be coming back to what's the value I'm trying to deliver, what's the change I'm trying to make, and then look at the initiatives and could AI be a part of that rather than just assuming that we're going to it's going to drop in and solve the problems with underwriters managing stuff or adjusting claims or you know supporting clients. I think there's a question that needs to be brought back first before we get too excited about the tech.
- Chapter 4: How should insurers approach modernization?
-
Thomas Rauschen:
No, I do agree, and I do think it's a perfect segue from what you just said into our organizational readiness, or as you just described, future capabilities. So, I just want to give you a message that I hear across the industry. It's like modernization must be a big-bang program, or we modernize first and innovate later. So, what I'm saying is, what you just described can be, on the one hand, we need to look at more holistically, or it's more evolutionary. You know, we do it gradually. So, both approaches are valid, I guess, but I think the companies need to have a clear picture of what they want to be in the future, right? Basically, what I'm saying is you need to have a capability model almost for the future in order to make the right decisions. What is your perspective?
Daren Rudd:
Yeah, and I think you are right. We don't see big transformation programs go well a lot of the time, or they don't tend to hit the value necessarily. I mean, we obviously work with lots of successful programs, but generally in terms of the market, there are lots of challenges on that.
I think the way I'd look at it is that AI is a disruptive technology in whichever way you look at it. You still need to have your north star. We talk to clients a lot, and they have this idea that I need to know roughly where I'm going and what I want to be, but that's quite hard to predict in the long term. So, I think you need to be an organization that runs smaller, iterative tests and learns with the technology and tooling today, particularly by getting it into people's hands so they can see what it looks like.
We have done some interesting work with a couple of our partner vendors looking at the different types of roles that are more likely to be impacted by AI, and then almost doing like a geological survey across all of the different roles and responsibilities, which ones are most likely to be impacted by the current generation of AI, and then do we do deeper dives in there to see whether there's value and benefit rather than trying to do it across the whole organization. So, you then do the deep dive and actually draw the mining into those particular areas. In fact, we can use AI to help us do that a bit more quickly and faster.
But I do think you've got to have a view on where you're going, but you'll need to be ready to adjust that. But then the way you learn about which direction you need to go is through that iterative test and learn. I don't think trying to either wait until you've decided everything or having a very rigid structure and working to a five-year or even three-year plan is going to work.
- Chapter 5: Building the foundation: Where should insurers start?
-
Thomas Rauschen:
Yeah, I think it needs to be a hybrid, right? Where you have to have a view: what does your capability model look like in the future, be it from sales to underwriting, product development, claims, always easier supporting functions, but equally, be sensible about what you just described. Be quite elusive and agile in making your decisions, but also test and try and before you roll it out more widely.
When it comes to where to start, right? So, really, how to modernize for AI and intelligence, you know. We can talk about all you need to build in trust from the outset. We talk about process redesign, we talk about composable modular systems, data first, you know. What is the starting point for you from your experience or in your advice that you provide to your client? Does it start with the data in order to have an intelligence-driven modernization, or is it the process or the people? Where would you start?
Daren Rudd:
I think I'd start with the problem. So, when I'm looking across the organization, what is the challenge that I'm trying to solve? We know there are, from applying different types of AI technology to our own organization, as well as working with clients, there are stronger value drivers than others. So, I think that the bit here really is to look across your portfolio of change, your areas where you're looking to improve or optimize, and start to look at those ones as the areas where okay, I've got these things that I want to do.
Second side of it then is to, and we use a sort of an assessment framework to say which are the ones that are more likely to actually work if I'm looking for an AI push, and that will have the impact in terms of you know, how ready is that team to change? You know, are these people bought into it? Have they just been through a massive change program anyway? And they're very unlikely to want to do anything more? Back to you know, what's my technology estate looking like? Have I got lots of legacy? Can I get to my data? So we tend to look at it holistically.
That would be the way I would recommend it. Look at your problems first, then do a bit of an audit around the maturity before you make your choices, and then work out a portfolio of those things that are possible, and there'll be some that are easy to do to prove that you can make these changes, some that are a bit harder but have got bigger, bigger value at the end of it, and you you've got to think of it from a portfolio point of view, but no different from all the experience we've had is years and years of running technology, that it's just the same application, the same way of thinking. I don't see AI being treated any differently from a technology, but the way that you potentially have to think about the impacts of it might be a different way.
- Chapter 6: Should AI be embedded in every transformation?
-
Thomas Rauschen:
That makes sense, and I do think, as you just described, I mean that that reflects what we see in a lot of transformation and initiatives, as you say, that you're running not only AI but platform modernization and others. I think one of the key areas for me that I want to highlight here is that data remains, it was in the past and will be in the future, it remains the launch pad or the foundational layer.
The one thing that I want to ask you, though. We can talk about operational resilience as well, or cyber, right? Do you think that AI should be almost like a workstream in every transformation program going forward and in the future? So, you basically treat AI as operational resilience or cyber, you know, you need to make sure whatever you do in your transformation programs that you run, it could be across the whole organization. Just make sure you have an AI layer to build it in from the outset.
Daren Rudd:
I think it should be a consideration, but again, as we've said, it's part of an overall technology strategy; it shouldn't be the only conversation. So, like you say, data has to be a key part of that. Operational ability has to be a key part of that. So, I think yes, we should be thinking about it in terms of operational resilience or cyber, and there will be applications of that, but again, it should be just one component overall.
But flipping it around, the impact on cyber and resilience leveraging these technologies really has to be thought through. There's a lot of reliance on third parties for these technologies to come in, and the attack landscape around AI is quite exposed, so that it has to be really thought about in terms of what you're doing, particularly with these new technologies and how quickly it's moving.
So, again, I think it has to be in the round, thinking about it as just one part of it. But you are right, we should be thinking about, could be leveraging this type of technology around the change that we want to make, but it's going to iterate very fast, and I think you've got to keep sort of quite an open mind to test and learn with it.
- Chapter 7: What do successful modernization journeys have in common?
-
Thomas Rauschen:
Perfect. And before we wrap up the podcast, I have one question, it's actually two questions in one, but I'm cheeky here, right? So, what I want to ask you, based on recent line projects or conversations, what do successful modernization journeys have in common? And on the other hand, where do organizations typically underestimate the complexity?
Daren Rudd:
Good questions. I would say successful programs, and there's lots of factors, but I'd say, in terms of what I see, some real clarity on the problems that we're trying to solve and why we're trying to solve them, a real consideration of the human side of that change in terms of the viability and the value in it, and a willingness to constantly review and recheck where we're going.
What I see fail a lot is, if we set a program running for three years, because it's a big piece of change, and everybody kind of forgets to come back and constantly monitor and check why we do it. Not around, you know, what did the project plan say, what milestone should be hit, but why are we here, what are we doing, is it still relevant? And I think that's where some of the big transformations come off the rails a little bit. Everybody gets so focused on hitting the plan and doing the milestones, they kind of forget why they're there. Bring everybody back to that. That's the bit I would argue is important.
- Chapter 8: The path forward: From systems to intelligence
-
Thomas Rauschen:
That's a great summary. We experienced it quite a bit in the past, where we ran these five to ten-year programs, you know, platform modernization programs, and then after five years, you ask yourself, why didn't we achieve the business value? Why don't we make the most of it?
Thank you so much. And I mean there are a ton of takeaways from today, but I summarized it in three takeaways or key takeaways from me. First of all is: modernization is no longer about systems, it is about enabling intelligence. That's my first one. The second one, you cannot bolt AI onto legacy foundations. You really need to think or re-rethink your organizational model and how you embed it more deeply. And the third thing is for me, sustainable AI requires a new operating model, not just new technology. Does that resonate with you, Daren?
Daren Rudd:
Yeah, absolutely. Absolutely. I think that's a really good roundup.
Thomas Rauschen:
Perfect. That's great. Thank you so much, and Daren, and I believe we see each other in the next podcast where we talk a little bit more about data.
Daren Rudd:
Fantastic. Thanks, Thomas.
Thomas Rauschen:
Perfect.
