Insuring the AI future with insights from Allianz's Claudia Max

AI’s finally crossing the line from demos to day-to-day work, and the hardest part isn’t the model, it’s the operating system around it. We sit down with Claudia Max, Regional CEO of Allianz Benelux, to get specific about how a major insurer approaches AI implementation across core processes like claims, underwriting, and operations while keeping trust and data protection at the core. 
 
We talk about the shift from scattered proofs of concept to scaling real capabilities, including voice bots and enterprise knowledge assistants that help employees find accurate client and customer information faster. Claudia shares how Allianz builds a practical map of customer journeys, then identifies where AI agents can improve coverage checks, decision quality, and service speed. The big takeaway—generative AI creates the most value when it forces you to rethink the journey, not just automate a step. 
 
Because insurance runs on trust, we also get into governance and responsible AI—group-wide frameworks, local transformation teams, AI ambassadors in each department, and the non-negotiables like human-in-the-loop decision making and transparency. We cover the messy reality of vendor selection in a fast-moving market, plus why many leaders “see AI everywhere except in the P&L” and what to measure as adoption grows. 
 
If you’re building an AI strategy for a regulated enterprise, this conversation offers a grounded playbook for AI adoption, AI governance, workforce AI training, and customer experience improvements without losing control. Subscribe, share this with a colleague, and leave a review with the one AI challenge you want us to tackle next. 

Join us and listen to all episodes on www.pwc.be/aiunscripted

Transcript

Welcome and guest introduction

Patrick Boone 0:03

AI unscripted season two. And today again we're going to talk to a CEO and we're going to try to understand what AI really means uh for corporates, for companies, and of course then also for people. And so today we're going to talk about uh Allianz, and I'm here with uh Claudia Max, you're the regional CEO of Vinilux, I think. Yes. Welcome.

SPEAKER_02 0:27

Thank you. Thanks for being here.

Patrick Boone 0:29

Thank you. Maybe a few words of introduction, who you are. I think everyone knows Alliance, I guess, but still I think it's interesting to keep you overview.

SPEAKER_02 0:42

No, happy to do so. So I'm now uh CEO of Allianz Benelux. So we are organized around uh Belgium, Netherlands, and Luxembourg since uh January last year. Uh so a year in the industry or in in the region. Um I'm having an insurance background, so I I have kind of an my whole my life I have worked in insurance. First um at uh in sales and then in underwriting. And before that I also worked in consulting. So I still have a bit the conceptional background uh looking at problems from from a consulting perspective still.

Discussion around AI in today's world

Patrick Boone 1:20

Okay. And so AI is that then because that's the topic of the day, of course. How is that now already starting within uh alliance? Because of course you've got the efficiency stuff, you've got more the go-to-market stuff, maybe starting with the first one, using it internally.

SPEAKER_02 1:37

I would say uh it's a big topic, and I mean we've been talking about it for a lot of years now, and I think now we are really at the edge where it comes from, where it really is an implementation, and it's not the hype discussion any longer, but it's really about already how we can make it work, even how we can go from use cases to really implementing it at scale. So I think this is really what the last year showed of learning that we now reach that stage. And of course, you're still more naturally in this efficiency discussion. Um, but we are also now saying, we just had the discussion yesterday, next to efficiency, it should always go along with customer experience, also more qualitative decision making and so on and so forth, not to always play the productivity and efficiency card, which I think is also quite a bit of a narrow term. And and and I think, and happy to talk more about it, I think there we have a clear channel now set up.

Patrick Boone 2:36

So you say we started already enough a few years ago. How when did you then start and how did that was that because the others were doing it?

SPEAKER_02 2:44

No, I think I mean it it evolved over time. I mean, in the end, when you look at AI, in the end AI really was created in the 90s. Even in the 80s. Or even that. So the origin is pretty, pretty in the past. And then I think uh, you know, uh it was first the hype state, you investigate it, you get uh external know-how uh to review it. Um but now I think we are really at a stage, and I think this is the the interesting piece, is um where we really have for I would say for a clear map for each process within the company from nature to operations to underwriting, uh where and how we think at the current stage AI could make a difference. And for particularly for everything which is voice, you know, the voice board were the really early examples. Uh we have clear already implementations. So we already implemented, I think, uh to 20, up to 20 voice boards in the company that are live and running. Uh, we make more and more use of these enterprise knowledge assistance services so that in different areas, operations, underwriting, people can make use of getting faster qualitative information of client data, customer data. And now we are more and more really in the claims processes, underwriting processes, thinking about coverage checks and so on and so forth, so that we really have this map of journeys, plus where the AI agents can work into it, plus a governance and setup model around it, how to scale it in the organization versus having a little project here and a little proof of concept there.

Patrick Boone 4:26

So you have then dedicated governance models, you have teams focusing specifically on the use cases, or is it important?

SPEAKER_02 4:35

So I would say um what we have in Allianz Group and apply across the countries is that we have a governance framework that that, of course, you know, we we build it on trust. So we need to get all the data protection regulation, et cetera, in place. So I think this is a global framework that is applied to each country. And then locally, what we have as a setup is that we have a transformation team reporting into me, where we have a data and AI office. And then they have a kind of in the different departments a counterpart that is kind of the SPOC or ambassador for AI in that department. So you cannot centralize everything. And with that, they have this roadmap of initiatives we want to do that are centralized in the transformation team, and then if they do it in the local departments. And of course, I think the important piece is um we can use what what we do in a global uh from a global perspective, the AI platform, the AI agents, that it can be replicated and deployed in the countries so that we can kind of benefit of the colour.

Patrick Boone 5:45

To create scale also exists. Because there's a lot of trial and error, of course, in AI? Because no one really knows what the end game will be.

SPEAKER_02 5:52

This is definitely the case. I mean, it's it's a lot um how the data AI platforms, how you build it up and so on and so forth, that you can definitely reap the scale. But then there's a lot of trial and error is also how to work with external parties, with whom do you work with, with whom don't you work. And this also changes very rapidly over time. So somebody that is really good in voice bots on might be not the case anymore a year later or even a few months later.

Patrick Boone 6:21

And what is your view on on return on investment of all this? Because that's a very complicated.

SPEAKER_02 6:27

That's a good question. I mean, it's always the discussion uh you see AI everywhere except in the P ⁇ L.

SPEAKER_01 6:33

Um the cost we do.

SPEAKER_02 6:35

The cost we do the investments, yeah. No, uh look, uh I think it's a bit of a natural transformation how we do our business. And I think it's for me clear that there will be value uh had. It's a bit like when we I mean it's a stupid example, but when we introduced the internet, it was not so tangible in the beginning, but you've seen it, there's a vast future to it. And I think we are on this journey, and the quicker that we can educate, adopt, learn what it can do, and then integrate it in our system, which is a bit the most difficult part sometimes, and the data, I think there will definitely be more speed, more customer experience, even severity gains from an insurer's perspective. So I I really believe in that.

Patrick Boone 7:22

And maybe expanding a bit on that user experience case, because of course you're in a in a business where some people will say, will AI be really inclusive or is it will will it be exclusive? So I think you also on your website you mentioned, of course, uh positive uh aspects also for the customers. Um how does that go, in fact?

SPEAKER_02 7:42

No, I I I think uh uh and as as a lot of insurers and all insurers based on trust. And I think we are a company that is uh more than 135 years there. So I think we will never um put that into risk. So this is the most important piece, and this is why we need to make sure how we apply it, that it is uh inclusive, that we uh keep uh customer data protection as the the rearm and hence make sure with all the the diffic the normal things, human in the loop and so on and so forth, data governance, that this is that people can rely how that we apply it in a secure way. And and that is important, and then I think a lot is about learning education and communication around it, that it's really transparent what is happening there.

Patrick Boone 8:33

And also on the insurability of risks, I think it can also have a positive impact.

SPEAKER_02 8:38

Yeah, this is this is the thing that we often um I mean when we s talk about application of AI, we often talk from uh we as a company how we apply it. Respectively, on the other hand, from an insurrect perspective, it's I mean, it's in cyber products, it's in um professional identity products and product liability. So AI plays a lot a big role from a uh risk perspective. We even see in it in our risk barometer that AI now jumped to um place two. So it's a huge topic. And I think also they are the journey to educate our clients in in what risks they are facing there.

Patrick Boone 9:17

Yeah, yeah. Which of course is key, yeah. Because that's that's typically then a question that some CEOs and previous sessions also mentioned. A bit the tendency that we basically say in Europe, okay, are we not over-regulating uh compared to others that are maybe doing more business uh with respect to AI? Because you're in a specific sector where there's not a negotiation aspect. You trust is core of everything that you do. So what what's your feeling on that? Are we over-regulating or it's always a good question.

SPEAKER_02 9:46

I think now uh the regulation is set and and it's about how we implement it. I think often uh it's how we implement it. Um I think it's good that we have a secure framework where every can can rely on and some legal security. Um and at the same point in time, then it's now rather um versus we should now really get in the execution mode. I think we are a lot in still in the stage of talking about it and and often create fear in people without having even applied one single topic. So I think we should rather now put it into action. We have the framework in place that needs to be in place, and then really endeavour and learn on the way um how to bring people along, what it means, what risk it creates, and and be very mindful about it. I think this is something it's also a strength of Europe, but at the same point in time not holding us back in pure discussions versus the can-do mentality sometimes.

Patrick Boone 10:50

Alliance can alliance. Alliance can play in that uh yeah, debate.

SPEAKER_02 10:57

Yeah, I I I think um we can definitely play a uh I mean it's a lot about communication, uh also about sore leadership. I mean, as an insurer, as I as we just talked about, we we see the insured risks and the the claims that also happen. So also, you know, really create that as a as a knowledge base. Um and then of course um we are in Europe a huge employer also um and and also hands bringing our people along and showcasing it in in in in also in in our workforce what it can do in a positive way.

Patrick Boone 11:32

So how many people do you have?

SPEAKER_02 11:33

Uh in Bennox, 1500.

Patrick Boone 11:35

1500. Okay. And so how in general, how do you then because yeah, you have the same d dilemmas as everyone, you need to bring your people along, uh, and and some functions are of course more impacted by AI than others. How do you make sure that yeah, everyone is positive?

SPEAKER_02 11:52

Yeah, um I mean it's it's a lot what he said, uh communication and education. I think um the fear in uh is is resulting from often not knowing what's what is there because AI is not really tangible. And then you have it in the media, in the news, everything about uh job layoffs and so on and so forth. So I think we create a fear before we even have applied it. So uh what we try at Alliance Bandalux is now that we really make people engage with the software. So just next week uh we have an AI week where every employee can participate and in very easy ways uh um use AI with prompts for creating pictures, for getting little tasks of uh um just just applying large language models in a very simple way. Um I think this is an important and what we also do as an example to make it a little bit more tangible. We anyways have for each employee a year around six, seven days that they can spend for learning. And we now say of these uh 12 hours a year, so it's more or less an hour or a month, should be really focused on data and AI. Because I I really believe in when you know what you're talking about, it becomes um more tangible, more you know, you're just used to it, and in that way it becomes less m much less of being a potential fear.

Patrick Boone 13:16

And what about your leadership team and yourself? You're also part of that.

SPEAKER_02 13:20

Yeah. Yeah, yeah. We we participate, and I think it's it's actually the case. I was myself, I mean, it doesn't do the huge thing, but I was last year one week in uh LA in San Francisco. And um, I mean, just because you spent there a week, you're not now the AI genius. But uh, I think you really when you experience it and see where these tech companies already are, you really realize uh that you need to engage in the entire topic because it will just come sooner or later. And and this is what we try also as a leadership team to lift. That we also um we have Allianz GDP GPT that we use, um you use it privately, Gemini and and ChatGPT and you name it, so that that you can also that it's not like this artificial management talk on AI, but that you know what you talk about and you also engage with it as as as your employees do as well.

Patrick Boone 14:14

And it's also told from Otopa, I think. We need to play ball as everyone. So but talking about then the investments are important, um and it's also an area where indeed scale is important that already you talked about, but third-party alliances. Is that something that you basically say? Yeah, the the make or buy type of decisions are of course specifically relevant in AI.

SPEAKER_02 14:38

Yeah, it's it's a combination. I mean, we we we do definitely have external partnerships. We also announced a few in the past weeks. Um and and it's however, what I said, it's it's not an easy, easy question. And it's what's we also make use of the combined group because you have so many external parties out there that also as a management team you almost don't know how to evaluate who is now delivering additional value to your company and expertise and who doesn't. And you could spend a huge amount of time in visiting different providers and so on and so forth. And so what we try to set up or we uh have set up and are developing further on the way is that we have a team that really on a global level reviews the different providers where we really go in nitigrative who can do voice, who can do document uh extraction of data, who can do text to voice or different uh around, and that we for each have a category. What are the providers that that makes sense to work with? Because if each local team or even department needs to engage in that, they spend huge amounts of time without uh without really creating value. And at the same point in time, in that course, you then also decide you do it go external, you do it internal, which provider you're applying, what fits with your current AIA platform, and so on and so forth.

Patrick Boone 16:08

Which of course indicates that you already have, as a group, reached a certain level of maturity, of course, in AI. Maybe m many people that are listening in or watching are not necessarily at that level yet. So what would be the key learnings that you basically would say, okay, if you're starting from scratch or just start to to play around with AI? What are the steps that you would say, okay, these are the things that you maybe the two, three points that you need to focus on? Maybe also based on the stuff that you have done and at the end proved to be not successful.

SPEAKER_02 16:38

Yes, it's it's a good question. So I would say in the beginning, uh you need to have a certain map where you want to focus on of journeys and really think from the customer perspective. I think what often people do wrong, and we might have done wrong in the beginning, is that you think about you purely automate simple processes. But for me, AI is not about automating simple processes, it's really about rethinking journeys. It's often when you also in the past applied technology, you just automate it as well. You put the same process in a digital system. So I think this is important that you do some groundwork looking at your journeys, what kind of value is attached to these journeys by different KPIs, and then say you focus and I would really focus. Don't focus on the tiny nitty-gritty value uh um journey, really focus on a big one. And I would risk that to do that because then you really can showcase you there, and you can showcase uh that it really brings value versus another tiny pock that brings your 10k in benefits.

Patrick Boone 17:46

Okay, so doing less but more than that.

SPEAKER_02 17:48

Yeah, focus, focus. So we we really in the beginning spend a bit of time of really identifying what is a core process that reaps value versus you just do a tiny pock here and there. I think this is what we started with, but uh, I think now AI is already at the stage where you can really apply it to the bigger things, and you need to prove it that it's that's working there. So that is, I think, a first start. And then I think what is important to set up a team where the core team working on the processes, and it just needs to be experts that really know how the process works, and then experts in AI and journey redesign and so on and so forth. But they, I think what is a core learning is that they have a team around that removes all the blockers, all the cloud approvals, workers' councils, data protection, legal questions that come their way. Because if the team that is should core work on this journey improvement has to figure out all these blockers that come on their way, you'll never progress. You will never you will never be finished. So you just give them the room and freedom. I mean, this is also a management topic, they can work, they and all the blockers they just outsource to somebody else to solve it. I think this is a core topic, and then yeah, it's communication, communication. And I think also a lot that we can still learn on it. But I think uh we really need to showcase the progress that we are doing because else AI becomes a bit of this nobody can hear it anymore topic. And at the same point in time, you don't showcase that you're actually moving along.

Patrick Boone 19:26

Because that's the risk that people are saying we're a bit disappointed about the return on investment today. Whereas, of course, we know that we're all typically overestimating the short-term impact and underestimating in long term. So the real impact is yet to come, in fact, and it's going very fast. So that's I think also with the dilemma that that that you have as a bit of an early adopter, that there's a maybe a bit of a fatigue in the heads of some people that here we go again. AI.

SPEAKER_02 19:50

Totally, totally. And and I think this is however why we need to move from pure conversations and uh to really uh showcasing things. And it can be small things. So when I just say focus on the big ones, it doesn't mean don't do the tiny stuff, but also dare to address real business topics because else it always is a nice show it what it really is. Exactly.

Patrick Boone 20:11

So do you then uh do they have different profiles on boards than you had maybe three, four years ago?

SPEAKER_02 20:18

I would say it's a devolving. And it's the intr really interesting questions about AI. Um, how does workforce role evolve over time? Um and I I would say this is a con as you said, as you do you don't see it yet in the numbers, of course the profiles are not completely changing yet. But certainly we assume uh and and we need to monitor what AI can do and how the uh the work is evolving. I think now we are really focusing on that on the learning so that our workforce is really at the forefront to know what AI can do and what not. And this is a journey, so we are not yet there. But as I said, we need to get it started because else once it will be there, and then uh we we want to be at the front and not at the back.

Patrick Boone 21:06

Maybe a philosophical question, but it's vital because um yeah, you are focusing on trust, we are doing the same at PWC because it's the only thing that we have, in fact. Uh so critical thinking of people is key. Um how do you deal with that? Because it's not always right what comes out of the machine, of course.

SPEAKER_02 21:24

Yeah, uh so how yeah, that's a Yeah, and and I I mean this is also I mean a lot a lot is said. So in the decision making, we clearly apply the principle of human in the loop, irrespectively in the end, that is uh no decision goes out or is applied without uh a human uh looking at it. But rather having AI as an enabler for making better informed decisions. Um so I think this is important, and I think also this core principle will stay.

Patrick Boone 21:53

Also with agents.

SPEAKER_02 21:55

I think also with agents, because let's face it, humans have a lot of tr trust and even. Even uh they are fine with human making error, but we usually don't have a huge span of uh flexibility for a machine that makes an error. Um so I think that principle will stay, but we will need to learn on the way how we apply it exactly and in which exact attorneys respectively which exact agents.

Patrick Boone 22:20

Okay. Maybe moving to a number of shorter questions. Okay. So as we call it the rapid fire uh segment. Um a big word for just basically saying short questions and uh don't think too long about it, just okay. You can think long, but it doesn't matter. But the usage of AI of yourself, you already said don't from the top, but in in reality, yeah. In reality, what's the AI that you use?

SPEAKER_02 22:49

Um really the typical I use Gemini or Chat GPT. I I really have to say I don't Google that much anymore, I realize. Uh or I use that I in in in Google I look on the Gemini AI generation, not the the the real results. So this is what I really do. In Allianz myself, I use Allianz GBT. Uh and that's more or less it. I have to admit, a lot of my colleagues do. We also have licenses for it, but I don't use so often um um um or what is it called, even copilot. There you go.

Patrick Boone 23:26

Okay. Flashback. Assume that you are 18 again, and you need to decide what you would study in the context of, of course, today, so AI all over the place. Would you have uh chosen something else than you have initially to what have you chosen to study?

SPEAKER_02 23:43

Uh I studied uh economics, so pretty easy. Um so the what you usually say when you don't know what you studied, you study economics, nobody. Um, I really I I I even already had economics in school as a as a kind of deep dive subject, and hence I always, I think, after a certain time wanted to go to business. Did I know that I will end up in insurance? Definitely not. So this came by accident. My first project in consulting was an insurer, and since then I never did something else.

SPEAKER_01 24:14

Okay.

SPEAKER_02 24:14

Uh, but I think it's it's good where I am, and so I I would not do I would chase choose most likely the same again.

Patrick Boone 24:21

But you're convinced that knowledge workers still have a future, even with AI?

SPEAKER_02 24:25

Yeah, yeah. I I think they did what you see also in in science now, that the people that apply AI plus have a good knowledge are the ones that really excel. So Okay.

Patrick Boone 24:37

And even indeed science, looking at the Nobel Prize, uh, I think it's it's it's a good example. Yeah. Um yeah, maybe the biggest mistake that you have made related to AI so far to be positive.

SPEAKER_02 24:51

I I the the the funny part related to AI, I think we don't even know which mistakes

Rapid Fire questions

SPEAKER_02 24:55

we all made. That's also true. Yet so we will uh find it out, but it it it I think we need to dare in order to do the mistakes and learn from it. So I think this is really not now talking about failure culture, but I think uh sometimes we what you said before, we shouldn't think sometimes too long, but sometimes we need to do it. Just do it in order to find it out.

Patrick Boone 25:16

Yeah. And maybe ending off with the stuff that really excites you about AI in the context.

SPEAKER_02 25:23

Yeah, that it's a a potential. It has a huge potential. I mean we s we might not even know what we said before that we which potential, but I strive for daily improvement of the company and and the service they do for our customers and brokers, and AI just is a huge enabler to do that.

Patrick Boone 25:41

Okay, great. That's positive. I th I see that you are an optimist. So that thing that's definitely what we need. So thanks

Closing

Patrick Boone 25:47

a lot uh for being here, for sharing your uh ideas, your thoughts. Much appreciated.

SPEAKER_02 25:52

Very welcome. Thanks for having me.

Patrick Boone 25:54

Thank you. And um yeah, thanks for tuning in again uh to another episode of uh our AI unscripted series. Um, tune in for more, uh I would say, and see you uh back soon. Bye bye.