Leading Digital & AI Transformation in Global Insurance

On this episode of the Insure Tech Geek Podcast, we explore AI transformation global operations and the future of insurance with Dr. Thomas Kunt, Chief Information Officer and Chief Operating Officer of HDI Global SE.

Podcast is all

About technology. The insurance world.

We're taking you on a journey through insurance.

This week's episode sponsored.

And we are back with another great episode of the Insure Tech Geek Podcast. If you can't tell if you're watching on video, I am live at Insure Tech Insights in London. It is currently March eighteenth, twenty twenty-six, and we are having a great time here with me and I promised him I would not say his official title after the intro. So it's Thomas Coo. Thomas, how are you doing?

Very well. Great to be with you.

I am so excited that we got to be here in this amazing studio. I wanna thank the production team here at the pod who is uh helping us record this and uh has a really brilliant place for us to record. But what a great conference.

Excellent. Yes, a lot of participants, stimulating discussions. I love to be here.

Is this how many years have you come to this?

To this one it's my first time. I went to Vegas, I've been to many conferences, first time in London. I have to try to see.

How many times have you been to Vegas?

Three times?

Three times too many. Vegas will wear you out. It's exhausting. It is.

And you see the same startups here. I mean it's kind of interesting, you have some pattern recognition

Yeah. This is a really wonderful conference, especially uh we got very lucky today. The weather's fabulous. It's it's early spring in in London. We're having a great time. We have I think six thousand attendees here at this conference.

Sounds fantastic.

So uh let's talk about you for a minute. Where were you born and raised?

I was born in northern Germany, a small town close to Hanover. I was raised there, then wh I went to university afterwards, went to Bonn, uh because I loved mathematics. Because when I grew up I wanted to become a scientist.

Well, math is the formula behind all of it, isn't it?

That's exactly right.

Yeah. You know what's interesting is the more because I'm fascinated by all fields of study. Um I think astr astronomy is fascinating and I think that physics is fascinating. I probably enjoyed physics the most because it was mathematics applied to the universe, right? And I I I enjoyed math, but it's exhausting. Um, but I just what I did find is that there's a formula for everything. At the end of at the end of the day, everything comes back to math.

And then I I my mom and my dad are both pilots and then I became a pilot. And flying is all math. I mean, and you have to be good at doing math in your head while you fly. uh if you're gonna be a great pilot. And so it was really interesting to to see um then I got into software and my best programmers were math majors.

Yeah, I I have to admit it's uh not that glorious. I also did not know the university before. I got a Fulbright grant back then and that was uh kind of they decided where you go. You basically s told them what you want to do and of course everyone wanted to be going to one of the big name universities in the US.

It it's it's wonder it's a wonderful town. I've spent a lot of time in Champaign and uh the restaurant scene's great, the bar scene's great, the c it's a wonderful college town. I'm glad you got to experience my my country and and that way. That's that's great.

No, I wanted not to well I mean I started academia because I wanted to become a researcher. I was really, really drawn towards research. And I also did my PhD in algebraic number theory. Very, very abstract. So back then I was proud that you can't apply what I'm working on.

Yeah, I mean it's the f finding the pattern behind everything is uh kind of at the heart of what we do in insurance is I I I I feel like as people who are intellectually curious, the search for correlation and causality is behind everything, right?

Yeah, exactly. And I think also today and you learn system thinking. If you th uh try to understand the geometry behind numbers, you don't you stop looking at the individual number, you looked at hey, what's the logic behind it? What are the patterns? And I think that's a lot of that you can apply even to technology.

Oh yeah. In in Spain. Especially with AI. We can talk more about that later.

But also in organizations, right? I mean if I think about it right is organizations are very complex systems. You have processes, you have people, you have technology. How do you connect all the dots, especially if things are not working well? I think I always try to manage the system. and move away from indiv kind of firefighting on individual topics more towards the system level.

So you get out of college. Uh, you're a a newly minted PhD mathematician. You've achieved uh uh a quite a quite difficult objective with as difficult as the dissertation and defense process is and Yeah, I I know that at U of I they were probably quite friendly to you in helping you achieve that goal, but it's still a multi year objective. You get out, what was the what was the immediate objective? Get a job anywhere? Or did you have something in mind?

No, it was a little bit um a period of hey what do I do now? Because the moment I decide hey maybe university career is not mine because there are very few pure research professorships, a lot of it is just teaching.

Yeah, that's uh i it'cause it allows you to this is the same thing I did. I actually did two internships and consulting with a large global consulting firm and then, you know, was Found very quickly what I wanted to do. And I think that's what's nice about consulting is that you can work in a lot of different industries. Did you work in insurance immediately or did you did you spread the the work around?

I started initially in banking. Yeah. But then I realized I mean I was firstly driven by hey, let's crack the hardest problems. But after one or two studies I learned hey, if you work day and night, it's not the problem that keeps you up uh or keeps you happy, but it's the people you work with. And then I had a mentor that uh then moved towards insurance and I followed him and ever since I've done fifteen years of consultancy in insurance.

Wow. What was your first gig? What was your first your first thing you ever did in insurance?

Oh that was ha that was optimizing the tight agent channel, so how to boost sales and profitability in uh tight agent uh networks.

Wow. Yeah. Mine was automating and optimizing the order flow for underwriting inspections for property policies.

Oh wow.

It was it really, really, really interesting. So when you write a property policy you have to get th at property inspected and we had to rebuild. Then I was twenty four at the time and had to re engineer the process for processing orders and it was really fun and it exposed me a lot to the to the property industry. Uh so that's that that's interesting.

It's kind of funny how that came. I mean because I started as a CO and a month later I got the CIO role because we had a large scale IT project that failed. And I will not forget the day when our supervisor board uh member said, Look, I don't want IT and business blame each other why the project failed. I want one person that I can hold accountable.

Yeah. More so than probably any time in my twenty five year career. A lot a lot of promises of of radical change, but when you actually get I mean, do you go sit with underwriters and adjusters? Yes. I do too. when you actually sit with them, things aren't that different for them. You know? They're they're they're still slogging it out on the phone with with cli with

So we are global uh we do global corporate business and specialty lines. Uh we're roughly ten billion euros. We are the leading one in Germany, we're pretty big in Europe, we're small in the US so far, um but we write all lines of business, but only corporate and uh specialty business. Five thousand people roughly.

Wow. All in Germany?

Oh no no no all across the globe. So Germany right now repres and it's about a f a quarter of the business. Twenty to twenty five percent of the business is Germany. Um a lot of still uh we do uh we are big in multinational programs. So that's our heritage. So we followed the German industry when they went abroad. So we are over a hundred years old. So we did all the multinational programs where we ensure a large corporate across all the geographies they're active in.

So like a large German manufacturer, they decide to expand to a new country and you expand with them.

Exactly.

So they're gonna write premium in that country, you're gonna have an underwriter who understands that risk.

No they they still underwrite most centrally, so we would do the master policy in Germany and then we would issue local policies consistent with the master policy.

You'd go get license there. Yep. Then you collect premium back in Germany and you'd you'd issue it there and then you obviously have to adjudicate the claim locally with license adjusters that are licensed in that country.

it's even more complex, you have to collect the premiums locally as well, pay local premium tax. So it's quite complex to manage those multinational programmes.

And then y you know, at the end of the day that company probably wants one renewal date and one master policy for everything, right?

Exactly. And then they want to make sure that local policies are of course legally compliant locally but also consistent with the master policy. Sounds easier than it's

It's not easy. I've I've been dealing with master child policies for twenty five years and it's not easy. It's not even remotely easy. 'Cause you have different billing schedules too, you have different issuances. Your software, of course, has to support all of this. The people have to understand it and uh equally apply it. It's very challenging. So um

Yeah, it's even more foundational. They founded us. So we are at the core styling mutual. We were founded by the German industrial companies when they went abroad and didn't find an insurance company that could support their needs when they grew internationally. That's why in nineteen oh three they founded us as a mutual insurer. Right now we are still we do we are s publicly traded, the stocks is still owned by the mutual.

Yeah. So they it's it's almost like a gr a group captive. Yes. Except it's a public company. Yeah. And you'll write non member owned business, right? So yeah.

Yeah, exactly. So we once we have the local the the German source linked business in uh in the international entities, we start expanding into writing local risks. Because you have the operations, you have the setup, yeah, you don't have the underwriting yet but it's easy to then branch out into writing local

You might as well just write the business.

Yeah, exactly.

Oh, that's fascinating. What a great what a great background and and and understandable. So let's go to the next topic. You know, you you're giving two talks here, right? One you're talking about agentic AI. And on another the COO compass for Global Ops, which are very different but connected topics. So what are the core ideas you're hoping leaders in the room walk away with in these two talks?

So for me it's all around um people process technology. So on the AI slot I talk uh I talked about the AIP so technology. My key message is don't wait. I mean start to experiment, iterate. I mean this is maturing so quickly. So I think this is not gonna go away. We need to figure out how to do how to work with it, capture the opportunity. And then of course the on the CO lens it's now what does it mean for processes and for people? Especially in a global organization, how do you

It is a ridiculously exciting time to be in computer science. I will say that. As a lifelong pro I started writing code when I was eleven and I'm forty six. So I'm in my thirty fifth year of writing code. I'm in my twenty fifth year of being in business. And this is the first time in my whole career when I don't have to say no anymore. I used to have to tell clients, insurance companies, that there were certain things we couldn't do.

No, I think that captured it well as an individual agent. Yeah. I think where It gets even more interesting if you then combine multiple of those agents with speci uh specialized roles. I think we see more and more and that's making me so confident if you look at software development which we are pretty close to. If I look at clawed code and our some of the identific things, yes wipe coding I would not apply to the enterprise environment.

It's it's amazing and uh and people said, Well d does this spell the the end for software engineers? I said no. Uh not at all, because I've watched non-engineers trying to operate cursor and claw and it goes awry pretty fast because they don't really know what to ask for. They don't know how to validate the results. And it's hard for them to think about how to architect the solution. I said, but what it does mean is the end of backlogs.

I think it's probably the simplest use case is again input management. Um because our input channel mostly is outlook. And to process that, I mean, yes there was already certain type of input management based on rules, but whenever something didn't fit you had those exceptions. And we found that applying agentic AI, you can really improve the performance of input management significantly. And nobody likes to open those emails, read it and then you can continue with triage.

They're an actual absolute total pain. You know, and that's also why a lot of underwriters have to make a decision to just not quote business. Because they're too busy.

Yeah, agree.

It's it's the it's the end of the underwriting backlog, it's the end of the claims backlog. It doesn't mean it's the end of the underwriter Uh you know there's there's all the the the doomsday's

Yeah, I and I think that will be a big challenge ahead of us. I mean I I completely share your perspective on underwriting claims. Yeah. Right now and we have a pretty aging population. I mean hardly anybody is now coming from university and saying, Oh, I want to enter insurance, I want to become an underwriter. I mean it's we really struggle with recruiting. So on the pure underwriting profession I'm quite positive. I mean we have a third of our population retiring in the next ten years.

Yeah, and and th th this is the message I think people need to remember is that the reality is there's a shortage of skilled workers. And and so e even with all the AI, um, you know, we're not gonna be allowed to have l AI licensed of underwriter bots and AI licensed uh claims bots. I mean, we're gonna have to have people involved. And and and I by the way, I wouldn't want AI to make final decisions. I'm I'm a massive believer in human in the loop.

That's our main IP. I mean,

And then and then, you know, potentially let's add in loss prevention and control, right? Like the actual risk engineers that go out and try and prevent losses from ever happening. So let's say that m is likely the three big legs of the stool.

Because of because of that, like we've got to find ways.'Cause m uh so many of their tasks are manual BS tasks. I I've sat at their desk, so have you, I'm sure. They they they s Outlook it they shovel documents through email. I mean it's just insane. I call I call Excel the cancer of business. Uh do you do you do you have a do you have any strong feelings about Microsoft Excel?

I love it and I hate it. I mean it's so versatile, it's very hard to replace Excel. I mean it's about yeah. But on the other hand it's also very hard to standardize or even automate Excel. I think if you want to do even w even with AI, even with ingenting I harmonizing boardrows you get I mean you can do so crazy things in Excel. Even an agent can't harmonize Excel sheets yet.

Exactly. They can be like they it can be quite difficult if you take uh go into Claud Code for example and say here, take my very complex pivot table and Excel spreadsheet and turn this into a web app. It'll actually struggle with that. Still.

And the other opportunity I see and that was one of our first most exciting use cases for the underwriters is the whole external research piece. If you look at farmer liability underwriting or financial lines, they do have to do background checks on the companies we're about to insure.

Yeah. Yeah, I I don't I don't see it and I don't actually you know, I've I've spent twenty five years trying to automate menial tasks. I've also spent twenty five years trying to kill Microsoft Excel. I've been successful at automating a lot of tasks. I've been unsuccessful at killing Excel.

I do think that the technology is progressing extremely fast and I'm quite optimistic that we will be able to support many more activities in the future than what we can support today. Yeah. I mean if you look at the exponential growth of the capabilities of some of those models, I mean it's just amazing. Even the models we've seen this year compared to the quality the beginning of last year. I mean technological progress is is rather rapid.

Yeah, when I when I my first big thing to roll out was med record summarization on the medical component of claims. It was the first time that I've had a software release and hundreds of users wrote in to say this changed my life. Like it's massive. Like this is more accurate than what I could do. This is better. This is amazing. Thank you, thank you, thank you. Like it was a big deal.

Exactly.

All the rest is B. And BS meaning things that actually don't really add value but are necessary tasks to complete to to achieve the final objective, right? Either close the claim out, uh indemnify the claimant, right? Whether it's property, auto, GL, whatever. um or or underwriting. So I I think there's a I think it's exciting. I think it's a brave new world.

It's always a challenge with the enterprise versus a personal environment. Um because of course enterprise grade IT is always more protected, so personally I I do use uh cloud code. N I'm not really a coder, but it's finally an environment in which even I can code and I love to really learn and understand. I think nowadays, even as a leader, you have to get your hands dirty. To really understand what's happening, you have to work with it. As painful as it sometimes is.

Interesting. Yeah. I think co I think co pilots like exposure therapy. It gets them used to it. It shows them what's possible. And it gets it done inside of a Microsoft enterprise agreement that's uh excluded from training and abuse monitoring, right? Exactly. Those are the two things you have to opt out of. Abuse.

They can try it out, they lose the fear, they they get used to using it.

Yeah. But it you have to admit it's a bit underpowered compared to the other tools. It is. On a personal level, I'll tell you, Gemini, Gemini, Gemini. Oh my gosh, Gemini. And its abilities through Nano Banana for image generation, video generation, Veo three point one. I mean, I my mind is being blown right now by Gemini. Um it's deep research mode. writes some of the best research papers.

Yeah, and they really need to work together. And that's what I like. I mean it's we've talked about business IT alignment for a long time. But I think with AI I mean business becomes IT and IT and technology is business.

Yeah, well the process has to follow quickly. Ideally you define process before you implement tech because if you apply amazing technology to a bad process, you just suck faster. Right. And no one wants to do that, right? We actually want to be better. We want to be more efficient and more effective. We don't want to make a bad process more efficient. So but sometimes we have to lead with the tech and then embed the process consulting engagement inside of it, right? That's

And we do the same.

And then while you're doing it, the whole time you're doing it, you're evaluating the people and determining are these people that need retraining or do we need new people? Right. So it's a it's it's a it's a very People process technology, they're not separate. It's almost like you need to always say it together, right?'Cause they're evaluated together.

really needs the business to lead that transformation because it's not a tech initiative. It changes so much. I don't want them to take the S-Is processes and just do AI in three sub steps. They need to rethink the processes.

Yeah.

That's not a tech child.

Yeah, the fur no it's not. The first question should always be why are we doing this? And that's an ops question. All right, I'm going to finish with this. personal like something interesting, unique, different, or a personal hobby that you're really into.

Well, I'm a long distance runner, so my my big hobby is long distance running.

All right. So it's the farthest you've ever run at one time.

Marathons. I mean I I actually uh when I lived in the US I learned to uh my or I found my love of running. I did a lot of those cross country races in the Midwest, always sponsored by a brewery, a lot of fun, right? You race and you have some fun afterwards. Yeah. Once you finish and then

True German style drinking beer after working out. Of course, yeah.

But hey uh the Midwest they are doing it all over the place.

Yes, they do.

And then I uh started doing marathons in Chicago. The first three marathons I ever did was uh Chic Chicago marathon three times.

What was the most recent one you've r you've run?

Oh, that's Berlin. That was three years ago.

Okay. Have you ever run anything longer than a marathon?

I've walked, I've walked for fifty kilometers. Wow.

That's a long time

Yep.

Yeah.

It took us seven hours I think to finish.

That's like walking for a whole day. That's that is actually painful. You need to soak in the tub after that.

You know what the fun piece is? It feels good till you reach the finish line. The moment you sit down, half an hour later your body shuts down.

Yeah, it's like it's like I was fine while I was moving. We're done now. Yeah, that's amazing. I I I like to walk for an hour at a time. I think it's it's it makes me feel really good. I get I get the Get that out. But that's that's remarkable. Okay. The Marathoner, Thomas Kunt. Thank you so much for joining us on the Insure Tech Geek podcast. Appreciate it. And thank you to all of our listeners for tuning in today to Geek Out live in London.

The Insert Geek Podcast is all about technology that's transforming and

So enjoy the ride.