Chapters Group AG - Capital Markets Day 2026 $CHG.DE

Thanks Jan. I'm going to try to keep it interesting. So what I want to talk through a little bit is what does creating an industry standard actually mean in terms of numbers? Where do we see the impact? Where does it come from? What does it mean for our EPS growth? And I'm also going to talk a little bit about the kind of early days of industry clusters in our VMS group. So starting with Lumra, obviously that is the industry cluster we have built. What we see for Lumra is organic revenue growth in 2025 was strong with 9%. We expect 20% plus for 2026 and some of that is volume growth with it which is great kind of people still love to come to Germany which is great. So that obviously adds to revenue growth but more importantly it's a review of a positioning of the three brands. It's a review of the pricing of the three brands. You are in a fundamentally stronger negotiation position towards all your key partners if you double the size. And we also see an increase in the conversion rates. As Jan pointed out, the key products still are blocked account and insurance. And the conversion rate of how many blocked accounts students actually also take an insurance product that is increasing over time because we are applying the best and breed approach we see in one of the three brands for all of the companies. So that is kind of the key drivers between organic revenue growth for Lumra in 2026.

And more almost more importantly this is what we expect in terms of growth. So as Jan pointed out we are putting this into one platform and that means that organic EBITDA growth for that segment we are expecting to jump from roughly 14% to 40%. A lot of this is reduced and focused marketing spend. Immediately I think three days essentially after signing what the team did shut down Google Ads. Kind of all of them were spending a lot of money on winning market share from each other. We stopped that and that is like hundreds of thousands savings per year just by not spending that money anymore. Paying commissions to partners, kind of all these kind of things where they just competed with each other. We can stop that now. And that adds a lot of margin. Then also we had to streamline the organization. There were some positions that were just doubled across the group. This is not easy decisions. This is not fun decisions but it's necessary decisions. The team took them quickly, executed on them quickly and also that adds to margin. And then in terms of suppliers, what we realized after clothing other than the accounting system, every third-party supplier solution out there, if there were two options, Fentiva went for A, Expatrio went for B for every single thing they needed. So also there we are obviously consolidating. We can renegotiate contracts. So you just have typical merger synergies that add to an organic EBITDA growth in 2026 of more than 40%.

Now what we want to do is to achieve strong organic growth in the long run. Not only with merger synergies but kind of really over time and that is where the not tip of the iceberg the bottom of the iceberg comes into play. If you can focus all of your resources and all of your thinking on how do I build the next cool product for my client, what you can actually achieve is long-term volume growth. And you can actually achieve long-term revenue growth. And I think we've built a cluster now with Lumra and we've built a team at Lumra that can actually do this. So creating industry standards is the ambition and I think for Lumra we're absolutely there. For our vertical market software segments it's still really early days. I still want to give you a bit of an idea what are we doing there, how we're thinking about it, what is the effect on the numbers. So many roads lead to Rome obviously. First one is PSE Hanscom and FA. Both of them in the mobility sector, software solutions for public transport, for bus, for trains, etc. What they also both had in common is that they were a little bit of the ugly ducklings in their previous corporate setting, a little bit unloved maybe by the previous management. So we acquired both of them in 2025, merged them into what today is Peak Mobility, and Peak Mobility is becoming the shiny, beautiful thread. Are we there yet? No. I think most of you have seen it in our numbers. Peak mobility, the combined businesses, they didn't have a good year 2025. We found a lot of things in terms of processes, in terms of culture, in terms of cost efficiencies that needed to be fixed and the team on the ground did a fantastic job in doing this. And I want to point out a few examples and give some numbers in order to give you an idea of what the extent of this can actually be. So the first one is a super easy one. We did a corporate merger. They moved into one office. We pay a lot. We save a lot of rent. This is a really easy decision. Everyone loves it. The new office is nicer. It's more modern. If you want to have a laugh, ask Jan about his first experience in the old PSE office. So that's the easy one.

The not so easy one is the pricing review and the value-based pricing project we are currently executing on. This as an investor actually at headquarters this sounds so easy. They haven't raised prices for ages, kind of just adding value and you just have these talks with the clients and you tell them how great they are and you go and then there are clients and they hate it if you raise prices and they have lawyers that actually say that the contract doesn't really allow for that and they push back and then they start complaining about what you didn't deliver in the last year and how the hell are you willing to raise prices? So, this one is really really hard to execute for the team on the ground because they are having these uncomfortable conversations. I'm seeing the numbers. They are having the uncomfortable conversations and therefore 2026 alone and this is not the end of it we will probably see an impact on EBITDA of almost a million euro. And then the most difficult one obviously is the part on team size. So as said both companies came from a corporate setting and there's a certain way of working in these old corporate settings that I wouldn't call efficient necessarily. And that also means that you have a team that is bigger than needed and that also the people on the team are not always the one that you actually need for the new era. So the team took the very tough decision on really downsizing the team. And again, this makes perfect sense but there are uncomfortable conversations, not just with the people but with everyone, a lot of lawyers involved. So this is really hard to execute on and kind of this is what the team did over the past 6 to 9 months and this is kind of tangible EBITDA effect in 2026 alone of 3.5 million, 3.6 million with additional measures adding to that. So we are expecting Peak actually to become profitable in 2026.

Now you might ask, isn't that what's industry standard about this? Isn't that just kind of your good old manuscript method playbook you should do with every company? Yes, it is. It is the groundwork that we need in order to actually build an industry standard. You can't start adding businesses or products to something that isn't running like a well-oiled machine. So this was needed in order to move to the next phase, which is strategic M&A in the mobility sector. This is four deals in the mobility segment. All of them are in due diligence. All of them with a closing date expected for 2026 and all of them have some link to Peak Mobility. They provide the same product but in a different geography. They provide an additional feature that would actually help. So all of this is really not just great deals on a standalone basis but it actually really adds to the value of the industry cluster in mobility. And then the second thing that we are going to do is become an innovative player in the sector. Kind of again your competition is they've all been around for a very long time. They all are doing things the way they used to do. And we now can do things differently and we now can actually think about what products are missing for our client. So what the team is currently working on and Leonard is here today and we'll give you a bit more detail on that as well is a software solution for essentially kind of fleet AI powered fleet management which is particularly important when you have electric vehicles like actually Hamburg has driving around Hamburg. So this is something that can bring us into a position where we become the product leader, the innovative leader in a very specific vertical.

Many roads are leading to Rome. The other example I want to give is our ecosystem in the emergency response solutions. So in 2023 we did two acquisitions in Austria. Both of them were providing or still are providing software solutions for fire departments. Cybos is an ERP solution for fire departments. And Blau list is an alarming solution. So essentially instead of having that good old thing on your belt, you can get an alarm on your iPhone. And that is an additional solution with Blau having a super strong position in the Austrian market. So we acquired both of them in 2023, did a corporate merger. They are now called Zolaris. Again this is not that much of an industry standard. Now it becomes interesting with the acquisition of DVA in 2024. DVA is an alarming solution for fire departments and other emergency response providers that has a very strong position in Germany. So now we have a very strong position in Austria in the alarming solution. We have a very strong position in Germany with alarming solutions and we have an ERP software for firefighter departments. Three products all in the same industry. So, it's a little bit because we're in Hamburg, we figured we take a Hamburg examples. Hamburg is now a client. This is a project that started by both companies in 2022. That's four years ago. What Feuerwehr Hamburg wanted was essentially a combined product combining Cybos as their ERP solution with DVA as their alarming solution. Now you need an API and all the connections etc. You need a kind of real product. It's very difficult to work on something like that if you're still competing for alarming clients with the rest of your business. So it took ages. Now they are a client and that is what we really like. They have a prototype they presented at Intuitz 2026 as one team, one product. And can now we have something to sell to new clients. And the cherry on top, we saved probably 35k or so in expenses for the fair because we only paid it once.

Now what we are thinking here is building out an ecosystem, adding solutions for industrial alarming, adding solutions for dispatch centers. So anything that is emergency response related that could add to that industry cluster and we are looking for targets in that specific niche. So very short back to the jungle of numbers. I think and that has always been our complication. Two things are difficult for us is we know that M&A will happen. We don't know when M&A will happen. We just the other day we signed a deal. It's a fantastic deal actually. It's signed. We were ready for closing. And now the EU Commission has to approve the deal because it's so mission critical that the European Union has to confirm that we can be the buyer. That's three to four months until closing. I think it's great because not only the software is mission critical, but the end market is so mission critical that the European Union has to confirm. But it's three to four months adding to the timeline. So when M&A will happen is super difficult for us to predict and it's also and that leads to it being a bit difficult to predict what will happen in 2026, 2027, 2028. I think kind of on the long-term effects. We have a very good view on how the model can actually work. Now this is the M&A pipeline. These 11.4 million will stay on signing for a little bit longer than we expected. And we currently have 110 million in deal volume under due diligence that the teams are working on. Now, not only is that a lot, but we also really like the quality of these deals. We are looking at bigger deals than we used to look at. We are looking at deals that are adding to an existing industry cluster. We are doing very focused M&A in fantastic niches with really really nice companies that we are looking at that also meet our criteria on kind of revenue, recurring revenue that are ready for an AI featured world. So it's really really good quality of deal. So that makes us really exciting. But makes this one a little bit harder. So everything I'm going to tell you now is before M&A because frankly I just don't know what is going to happen. But a few effects of building an industry cluster on our numbers.

So what you probably have noticed is that our entry multiple or the multiple of capital invested compared to EBITDA jumped quite a bit from 2023 to 2025 from 6 to 11 and a half. That is a lot of that is driven by Lumra kind of that is a transaction where we paid more than our average 6 and a half to seven and a half. But what we do see kind of looking at the long-term plan is that we will very quickly go back to a multiple below six because we essentially operate it down. What you've also probably noted is that the share of adjustments in our EBITDA was fundamentally higher than it used to be this year. Again, a lot of this is driven by transactions like Peak, by transactions like Lumra. They have been dear bony to the people working at Lumra because this is a huge change for them. You need to make sure that the team on the ground is actually satisfied. There have been a lot of adjustments in relation to Peak because we need to turn that ugly duckling into a swan. So it has been a lot. It has been unusually high this year. We do expect this to normalize back to normal levels in 2026 again. And I'm going to make the big disclaimer of and whatever happens in terms of M&A because that might change things.

Now, we also had negative earnings per share in 2025. This is essentially down to the fact that while we adjust for all these one-off effects in EBITDA, they are in the operating results. We don't adjust for that. This is really a number where we adjust for all the accounting stuff like goodwill depreciation etc. But this is really a number that is net income of all the companies belonging to the group multiplied with our share in it and then we add it up. So 2025 has been an exceptional year. What I presented last year that for 2027 we expect a range kind of of 80 cents to 1.10 per share in 2027 and that expectation has not changed. Now what has changed is our view on organic growth. You saw we updated our guidance and last year I presented this

Sensitivity table, and I've put 10% organic growth in there, and I felt really bold. I felt really daring. I felt like okay. And we saw then that obviously organic growth is the oxygen for our earnings per share. And that hasn't changed. What has changed is that I actually dare to do a sensitivity analysis with organic EBITDA growth above 10%. And what you can see here, and this is five-year year 2027 to 2032. What you can see here is the organic growth we can achieve is so fundamentally important to our organic growth to our growth of earnings per share. Much more important than the entry multiple we pay. Much more important than the entry multiple we pay. If we can find great businesses with fantastic growth profiles, we can pay a higher multiple because that dilutes down very quickly. Now, you don't have to get worried that we'll get crazy and we start buying at 13 times all the time. Mark will never allow that. But I think it goes to show how important organic growth and hence how important building industry standards is to the entire model. Now, Jan gets the fun part of telling the story. I get the hard part of telling the numbers, I always say. Now, the really hard part is to get to the 20%. Because all of this makes sense in Excel and it's a complex Excel model, but it's Excel. Someone actually has to get organic growth on the street and that is not us. That is our platform teams and that is in particular Mark and the COO team and the CTO team that are supporting our platform teams where they can to actually get to those numbers. So over to Mark who's hopefully going to tell us how we get to 20% per year.

Thanks a lot. Thanks a lot Marlene. I have three topics. One topic is transformability. One topic we already mentioned is industry standards and of course no VMS presentation without the topic of AI. So we heard from Marlene and that's the top part of this slide that obviously the manuscript method seems to deliver with the improved guidance. We also heard that we have also in the VMS space first indications that building industry standard does make sense. Now how can we actually do that in practice with a little bit more detail around a French example? I know they now also out of the World Cup but I still miss them and so we will still have them on the slide. And obviously what is playing the role of AI in that regard. So before I introduce a concrete example, I like to take a step back and look at why we all in this room like vertical market software from a first principles perspective.

You can see on the left hand side why we like investing in vertical market software. Obviously the well-known notion of high switching costs. So customers are locked in these B2B software solutions. It's a lot of effort for them to move from one vendor to the next vendor. You have to train your people. You have to buy new software. You have to do a lot of customization. It takes a lot of time and you're operationally depending on that software. So people typically don't do that. Even if I tell people to hijack the prices and move the prices up, they still remain clients. Another thing that we really like is although those vertical market software markets are sometimes really really small, remember we still have the market leader of orchestra management software doing revenues of €1.5 million and we are the market leader. So sometimes these niches are really really small but what makes it attractive is that in those very small niches you only have a few players and you don't have a lot of competitive intensity and all the big guys like Oracle, like SAP, they don't find it economically useful to develop software for such a small niche, so the supply scarcity is protecting our companies from a lot of competition. And lastly, but not least, most of the clients of our software business have what I would call a good enough inertia. They don't need the best UI. They don't need crazy additional functionality year-over-year. And frankly speaking, they sometimes are also pretty happy with okay service. And all of that creates a low churn profile, very predictable revenue streams and we end up buying those assets at pretty okay multiples. So it's a good business to be in. And if we combine what we can offer to these companies, we can offer much cheaper capital compared to those sellers of those businesses or a buy and hold forever approach of being a public company, not a private equity, not being required to exit those businesses is a very good fit to what those owners of those software businesses actually want. That is stability for the legacy they built over decades. Now the special thing that we bring in and I think we even more bring this in since the last two years is we collected a lot of best practices how to improve those businesses. So it's the improvability that we bring to the table. We call it the manuscript method. It is policy deployment. It is pricing. It is a lot of additional things that we collect and share in the group to improve the businesses and that eventually makes up the 22 plus percent guidance. So the classic power move is buy a durable asset, improve it, and everybody is happy. The key thing here on the slide is the defensibility of those assets is not something that we add. The defensibility of these assets comes when we acquire the businesses.

I think we will see later in the presentation why this is maybe changing with AI. Now a concrete example I always appear here trying to come back to last year's session. Last year's session I already talked about the French business that had some problems. It was this business somewhere in southern France that we acquired. It's a public sector VMS business, small business, 3 million revenues, bit more than 1 million in EI. So imagine your parents in France, you have to work and you want to have your kids being taken care of. Our software is managing this from a customer's perspective. We acquired the business and then after the acquisition, we realized oh the business is not collecting any new professional services bookings. What's going on? We use policy deployment. We did come up with some counter measure plans. We assigned some additional people to the sales side of the house. Nothing worked. And we used what we call the diagnostic manuscript ratios to figure out where is the issue of this situation. Issue was sales of marketing ratio was off, professional services ratio was off. The long story is either we bring in new professional services bookings or we need to conclude that this business is overstuffed in terms of professional services and we have to let people go. Right? So that was where I was at at the same time last year. And after this workshop, I was like traveling to France to sit together with the team somewhere in France, middle of nowhere like 50 kilometers outside of Marseilles. And that was like a very cozy meeting room with a lot of people with a very small window and it was really really hot but we had a lot of coffee. So what we did is we discussed what is the situation of the professional services. Is this business really not having enough professional services? Do we need to cut the team? And what we figured out is that it's not a problem of the business not delivering professional services. The problem was that there was no charging of this professional services delivery in an appropriate way. What they did quite often is they had some health checks. So customers wanted to improve the usage of the software and then they charged like €5,000 for a workshop for a one or two day workshop which is kind of okay. But the problem was all the change requests that came out of this workshop were actually included in this 5,000 euro and sometimes they spent work for like €20,000 and didn't charge a dime for that which is suboptimal.

And the other thing is they had a couple of larger cities as customers. And these larger cities, they were just having kind of dedicated support people. They called them like all the time. And we figured that some individual employees worked like three days out of the week for a particular client and they didn't charge anything for that person. So what we did decide, we concluded we wanted to introduce two manuscript method modules: one was a tool called Redspace, another was a value based pricing normalization. And the Redspace tool is a really nice tool. It basically gives you an Excel sheet, just visualize it a little bit: you have a column with the days of the week and then you have columns for each employee that should deliver professional services. And at the beginning of the week, everything is like red, hence the name Redspace. And the people, the individual contributors, then have to fill each day with professional services work. If they achieve a day full of work, they can turn it green. If they have no clear confirmation from the customer, but maybe they can do some work for a customer, it's yellow. If they have some vacation, it's gray. And if they cannot bring in enough work, it remains red. And the team is meeting every week and everybody in the team is looking at that sheet and everybody is obviously kind of a little bit embarrassed if all is red, right? So people are really calling customers, can I do this work? Can I do that work? And after a couple of weeks, the situation typically looks like that. So some people have it all green, some people have some red and some people have still a lot of red. And what then the team does is to understand what is actually this person doing all the day. Although it's not billable work, but then they figure out situations like I just mentioned that a customer is constantly calling that poor employee and the poor employee cannot defend themselves not to deliver services. So then the leader needs to step in and change that customer relationship that we either stop this or have the customer pay for that. And the business also did come up with a functionality in the software they call it the green button. So whenever a customer called they could press the green button and the green button was recording the time they spent for that particular customer. As a flanking strategy they also did a pricing normalization. So they introduced a new service level with the customer and as you can see there's a gold level that's plus 60% maintenance revenue charged if a customer was on gold level. And of course those nasty cities always overusing those professional services didn't have a choice not to take gold, right. And we did also some communication for the team because the team needed to understand why we are now treating the customers in a different way. The team first time ever realized how much revenue and how much profitability the business actually is making because the former owner never told that to the team. And all those initiatives led to a lot of individual accountability taking and you can see here these are the monthly professional services bookings that we did achieve when we started introducing this somewhere here to this and we increased that by 160% and all of those things is probably a combination of certain tools we share within the teams and the people on the ground that take those tools and execute those tools. And these are people that are different than me and I'm very grateful that we have a lot of such people nowadays in the no matter in France or in Germany. And that's how this whole manuscript method, call it 1.0, actually is working.

Now what happens with AI and what happens if you would just continue with implementing the manuscript method. I think what happens with AI is that AI is not attacking directly the different moats I mentioned on the first principle sides but it basically attacks the underlying factors that lead to those moats. So why are there high switching costs? So one big answer to this question is that it's just bloody painful from a migration effort perspective to move from solution A to solution B. Now with all the automation that AI is offering, that migration effort, ladies and gentlemen, is decreasing. So that will erode. Will high switching costs ever go down to zero? Probably not. But they will erode. The supply scarcity: having a small niche and nobody coming into this niche because it costs so much to develop for that small market. The orchestra example, I think that will break because with every frontier model we'll see the build costs just go down and down and down. And we can see that and you'll see that later with some examples that Toby is sharing. It's actually amazing how quickly you can come up with a fully fledged ERP solution and we did that internally as well. Now the good enough inertia, the satisfaction with the status quo, that will also depending on the particular businesses hold back for a while. But when you have a situation where the AI startups or the existing providers of competitive solutions are providing more and more value using AI in their products, at some point the delta between our solutions if we don't do anything and the competitive solution will be so high that this delta kind of makes sense if you combine it with the low migration effort. And what we also should not forget is that often times our solution has been introduced to the customer organizations maybe 10, 15 years ago. The guy that has been introducing this software from a customer side is getting older and older and at some point that guy will retire. And then a new guy comes in without all that history, without all that relationship, and the new guy will maybe say oh I want to have a cloud solution. Oh I want to have... Oh yes and I should have the microphone on my lips. Exactly. So these are kind of breaking points and if we are not the one that is delivering top AI enabled services and solutions at that point then we're losing out. And if you become like a lazy incumbent, there is another option. We have the same possibilities as the AI startups using the different tool sets, the agentic tools that are at our disposal. So if we can bring our existing companies to now move quickly because we have several advantages, we do have the distribution to the customers. We have these customer relationships. We are basically a create from a procurement perspective. It's much easier for us to deploy new software versus a new player. So we have a lot of advantages but we really need to move quickly and that's why we did come up with this RAV that you might have heard of. This stands for red alert for opportunities, kind of our AI strategy that should help our existing companies to move quicker and to AI transform those businesses. Now if I go back to my first principle slides, you might say, 'Oh, this is red. Switching costs go down. The niche is no longer unattackable and you have a yellow on the good enough inertia.' Am I in the wrong movie here? And my answer is no, you're not. And why is that? Because these red situations also create something that I told Greg, one of our investors from Sedor just before the session started. Often times in these little niches, you do have just a couple of players as I mentioned. It's like the four tiles here on the carpet and that market structure is very hardly ever changing because of this low customer churn profiles. Now AI is changing that and those players that don't move will basically be washed away and there's also a possibility and opportunity for our companies to actually expand their market share not only against these AI startups but also against the existing players. So there's a great opportunity for us if we can bring our existing companies to move quicker than the existing players. And I think what we also should take away from this slide is that the AI through these red areas is actually breaking the division of labor. These companies no longer can protect themselves because those things break. So there's actually a responsibility and opportunity for the holdco or the platform in our situation that we help those companies to transform and that implies that the manuscript method needs to transform itself from just improve businesses like as I just have explained by the example of this French business to actually transform the companies.

Now, what does that mean? If we're talking about transformation, I think it's good to remember that most of our companies actually have quite some time until this red situations come in. And why is that? Because they're mission critical. Now, mission critical is a very high level term. So if you go down one level and you ask yourself what actually makes mission criticality of these software businesses, I would like to introduce a concept that I call operational anchors. What are operational anchors? Operational anchors are data and functionality in the software that if you remove that, the customer really has operational problems to deliver. So often times they are critical parts in the main workflow of the customer. They accumulate a lot of valuable data. Often times these operational anchors are also connected with each other. By the example of this French business, all of them are examples of that French business. So for instance, our software is collecting the attendance data of the children in these nurseries and this information gets sent to the city and then these nurseries they get subsidies. They don't get paid if this software doesn't send this information. They use it for the canteen forecasting to know which child has which dietary preference. And either you have enough meat or not enough vegetable. I mean this is an operational issue for those customers and the mayor of that particular city has a legal obligation that every child is enrolled in school. Guess who is actually managing that source of truth? It's the software of our business. And if they don't provide that regularly, the mayor of this business will violate the law. So these are basically networks of operational anchors and that's one of the reasons why customers shy away to replace those businesses. Now what happens if we acquire and collect businesses in the similar industry and in this situation we actually have that I already was talking about this business for early childhood caring. We do have when we are taking the vertical as citizen in France in social need, being it parents that need to take care of their kids while they're working or a little bit older children that might be disabled or they might unfortunately have lost their parents. There's software to manage that. We have two businesses. We call them Heart One and Heart Two. And we do have two businesses that take care for elderly people. So some elderly people in France they cannot pay their taxes because they have a little bit dementia and they are guardians that take care for these type of people. They use software and that's King One and King Two. All of these businesses have operational anchors. Hence they have low churn profiles. But all of these businesses will at some point be disrupted if we don't move and AI transform those businesses. And we're doing that with the RA strategy. Toby will talk about this in a second or in a minute. And I think that's great. We can create additional AI moats if we AI transform these businesses. But what we also can do and this is something that we actually did today. So I learned from Torson who is heading up the group in France that we acquired and closed and signed a deal today. No joke, we call it the full house. Now the full house is a so-called CAS software which is managing the registration process of when there's a citizen in France in social need they have to register themselves that hello I have a problem and the hello I have a problem software is full house has about 20% market share across France. And the beautiful thing apart of yes they also have some operational anchors, they are in the process of replatforming their solution. And why is this relevant? It's a great opportunity for us not only to help them to replatform their solution with what we learned over the last 12 months around agentic coding and also thanks to Toby and team, but we can actually now coordinate similar replatformings in some of the other businesses and while doing that we can synchronize the data model and we basically can allow that a person coming in via the full house is then basically assigned to an appropriate order of our businesses and we can do that in an automated way. We can also when we synchronize the data model basically have a platform that always knows where the individual citizen is stuck in the process and that allows us to provide a lot of automations, a lot of analytics, and a lot of cross-selling into this space and essentially will allow us to create an industry standard in this particular niche. And therefore the acquisition of this full house will make the whole system much more stickier and therefore essentially make the whole companies more valuable. And I think we acquired this business for like five times ITA. I would argue we could have paid probably a bit of a higher multiple for that and it would have still made sense. The key ingredient that we need here is that we can actually only do that if we can replatform some of those businesses. That requires that we level up our R&D departments to really apply agentic coding and to become much quicker in developing software and that's something that Toby will speak to in a minute.

I think the probably last slide here is finishing off where I say where we should go forward. On the left hand side I think we should continue to underwrite to acquire networks of operational anchors. We should do that in the verticals we're already present in to become a more dominant player in those verticals. This will obviously give us much more pricing power as you can have seen that house for me. I mean this is like pricing like this is paradise pricing and I think that should be something that we really should do. And on the other hand, when we're doing M&A we should have open eyes for transformability. Now transformability transcends to we need to understand for each acquisition what is the headroom, meaning what is the current situation and how can we move this business into a transformed version of themselves. What does it mean in terms of team size? What does it mean in terms of processes? What does it mean in terms of pricing? And we're currently developing checklists to better assess that. It also is important to understand what's the tractability. So tractability means how easy we can do this transformation or how hard is this transformation. For instance, we found that in some of the businesses where we have leaders that really leaning in into the AI transformation, everything is much more easily done as opposed to you have someone that just doesn't believe that AI is meaningfully impacting his business and that will also instruct what we're doing with this business respectively with the leadership team. But also the tractability is concerned when it comes to customers. So we have some customers that basically tell us go away with AI. We don't want to do anything with AI. And such a business is for me less attractive as opposed to you have a sector that is really appreciating AI. And so tractability goes kind of both ways. And I think going forward we really need to invest in the repeatability of all those transformations because the more often we're doing that, the more proficient we are becoming. And I think if we develop muscles in that regard, that will become a key differentiator in the future. And we already seeing it right now that some of the companies we're talking to in the M&A process, when we're talking about what we're currently doing in our companies in terms of AI, it already becomes an attractive feature of us because they can see that we can help them to manage the AI transformation. So the new power is keeping the focus on making this transformation happen quickly, more quickly than the potential decay. And by compounding this across eight platforms, we're basically leveling up the manuscript method from just improve stuff to transform stuff. And the key ingredient that I currently see is that we need to double down on making our R&D teams into agentic coding factories. That's the groundwork we already started doing and Toby will show a couple of examples. And the next area that we tackle will actually in parallel is to do the AI transformation in other departments as well. Okay, that's it.

Yes. Hello everybody. I am Toby, a new addition to the team compared to last year. So Mark already gave me two questions to answer in this talk. But let me start with some general remark about software and AI. So every company today is an AI company and everybody comes with a cheap narrative and that's not what I want to do today. I don't want to show you a vision slide with some agent symbols on it but actually show you what we are doing in the portfolio, how we track it, and what we do in the broad sense of the portfolio to really move the companies toward the AI transformation we are aiming for. Let us take a step back and look at what has happened over the last 18 months. So in Q1 of 25 we started the first AI working group, got people together from all the opcos that were interested in AI, found the enthusiasts, started something that later turned out to be our AI champions league. Then later in the year in the manuscript days we already discussed how we can identify in our companies the right positions where we can apply to add value for our customers. Eventually we noticed okay we need central capacity, central knowhow, and Mark started to hire the first parts of the core AI team. Then in Q4 I joined as the CTO and we decided two things: we first need to do is formalize our view on AI and the way we see AI in our portfolio and that's the AI maturity framework we developed in Q4. I will talk about this in more detail in the later slides. And we also came to the conclusion okay we have to change something in M&A and the first level of adjustments to our M&A guard rails was done already in the fourth quarter of last year and there was another refinement in March of this year. Also in the fourth quarter of 25 was the Opus moment. So when Claude Opus came out, we quickly noticed okay something has changed and the way we have to look at software development really changed and what was possible previously in years is now possible in weeks, months or sometimes even days. And we took a step back and decided to really accelerate our efforts, put up the velocity, and that was the birth of the Rafo framework and our Rafo initiatives that were ready and already starting to be rolled out when then eventually the market also understood that there is something happening and the stock price reacted. But at this point we were already ready and rolling out Rafo which we will see in the rest of my talk. In quick succession we started all these initiatives. We started momentum competition AI hubs. We started the coding training and now today we already see results in the companies where we see more and more opcos where agentic coding work has taken over human work. We see green field projects that were really out of scope for Chapters companies before AI. And I want to explain you in the next couple of minutes how we got there.

All right. The AI maturity framework. If you have 60 companies, all of them are really individual, they have their individual markets, different sizes and so on. How do you assess if your portfolio is moving and if it is moving in the right direction? What we did there is the AI maturity framework. We separated our companies into 27 distinct capabilities. And for each of these capabilities, we developed a transformation lane as we call it. Here on this slide, you see as an example one capability from the support area where the company moves from beginner to assisted, enabled, and transformed. That's the same for each of these 27 capabilities. So as a beginner in support, you maybe have a first chatbot pilot. It captures some ticket information, gives a little help. On the assisted level, the AI already drafts answers to common questions, routes requests to the right employee, still with a human in the loop. On the enabled level, that's where we think a company already has a knowledge layer and has a clear knowledge base in the company that is fed by different sources and is the basis for agentic help that already end to end cares about the most common cases, gives direct answers to customers. And then in the end there is a transformed level where agents really have responsibility in the support flow and take care end to end of the majority of the support requests. So this is one example how such a transformation lane in one area of a company can look like and we did the same for 27 other capabilities. We developed one lane that we call foundational lane and the rest of the areas are related to the standard operations and functions of a company. What is the foundational lane? Those are the basics. So this is do you have AI policy? Do you have an AI representative in your company? Do you have basic AI training for all of the employees? Do they know the EU AI Act regulatory requirements to work with AI? All of this is the foundational lane and that's more or less done in most of the companies by now. So what you see here on the slide overall is the result of a self-assessment of all of these companies which was then challenged by the platform representatives. We are closely aligned with our platforms. They have the same view on AI and these levels as we have in AI core team. So this overall gives us a clear and comparable assessment of the portfolio over all of these capabilities.

And as you can see, foundational changes are the clear winner and after that R&D and support are the areas where our companies invested most and reached the most steps in terms of this maturity. However, this does not mean that our accounting departments don't care about AI, but that our opcos took a deliberate decision to focus on support and R&D because they see the most immediate advantage there. Now, that is a decision that is taken at the edge and that should be taken at the edge because the opcos know this better. What we can do as a holdco is to enable them and provide them the tools like the maturity framework, like the training and so on to take smart strategic decisions today and to give them the tech and the training that they need to actually decide is it rather support, professional services, or whatever that brings us forward most. Last point about the maturity framework. You can use it to build your strategy because you have development lanes and can plan how do I want to go along this path. You can use it to measure the portfolio. And the last point is you can use it to set goals and that's what we did. We plan that until end of the year all of our opcos have reached the assisted level for everything foundational. They have to reach the assisted level in R&D because we are software vendors and that's certainly the biggest point for us in AI. And then apart from that, each opco needs to choose at least three development lanes where they see the most advantage for their company but we demand that they move at at least three points. So this is how we track this, this is our result. You heard our goals and that's also something you can track me against for the next year.

Let's come back to Rafo. So you have seen where we are. You have seen how we measure this and now let's come to the details of what we are actually doing to drive the progress. On the left side you see what protects us today. So we have systems of record, regular moats. We are embedded in the transaction of our customers which are rather slow moving. All of these are parts of the iceberg that is melting. And now Rafo comes in and helps us transform these companies to harden these moats and turn the iceberg into stone. Again, what is often the starting point of this AI journey or has been in several opcos is Mark coming to the opcos doing something we call the future back workshop where we take a larger part of the company and discuss with them a scenario where it's the year 2030 and they have been disrupted by AI. Going from this point back we see what happened on this wave. So you are disrupted and from the starting point we build up again a positive vision how you can be part of the AI winners in 2030. This format has proven to be really good to drive adoption and to get the people started. But it's not just talking. AI is also knowhow and tech and that's why we have developed the AI hub, an open-source-based AI solution that gives you software but also GDPR compliant infrastructure, something most companies would maybe struggle to set up. We found one solution that we apply in all of the companies. We roll it out so there's no central AI hub from Chapters but every opco has their own individual AI hub, can shape it to their needs, but they don't have to reinvent the wheel. And the AI hub gives us the opportunity to take use cases that work in certain opcos, take it up to a central repository and then ship it into the whole group and that really speeds up because we are not just sharing ideas on the level of blueprints but actual implementation.

Then there's agentic coding training. We set up a bunch of trainers. We went to AI events. We searched for people that are AI and agentic coding maniacs and have done nothing else in the last two or three years. Then finding out how can I create my code with agents. We convince those people to come into our companies and tell our people how can you do it because what we learned is most people in our R&D departments understand that AI is great. They understand that there's a huge opportunity but there is some uncertainty if it really works in their 20 year old stack and that's the gap that the trainers can fill because they come into the companies, they stay there two or three days and after that everybody has seen it works also in our 20 year old stack. And yeah, that's when movement happened. Then there's Chapters Momentum. At some point we thought okay our companies have long-term plans, they have migration plans, they have a road map with their customers and we don't want to disrupt that. We want to keep them moving along their strategic paths. However, we don't want to be left out with all the cool new AI features. So we set up the Momentum process where companies can combine a really great business case together with hard customer commitments to get funding from the holdco structure to make sure that external forces can help them to have AI features in their product now and today while still keeping their strategic plan. And this has been quite a success and we will hear more of something that came out of this later at peak. Then there are people. So we said the change needs to happen on the opco level. So we assigned a champion, an AI champion in each company, and we also decided every MD needs to be an AI leader. So we take every MD into training sessions where they not only learn how they can transform their company but how they can be a more effective leader using AI every day in their work as an MD. And that's something that we also found really useful and beneficial because once the MD has seen in his everyday life what difference AI can make, they are ready to push it into the opco and then there's not so much that we have to do.

Okay, let's go on and come back to the second topic, the agentic coding factory. So what we have seen before was mostly related to how can we transform the rest of the companies. My first example and I will show you a total of three agentic coding factories we have built is Software 24. So this is maybe in some sense the least glamorous place where you can try to use agentic coding because Software 24 offers Vinca. This is a more than 20 year old Delphi based solution for property management.

Thanks, Yan. So I'm Andreas Philippe of Altamount, and we actually started Altamount two years ago in this room. We introduced it. I think some of you folks have been here at that point. Two years ago in July '24, and I think today our ambition of Altamount is really to become the leading public sector software company in Europe. That's what we work for every day and I think AI is a big part of that. In the last year we acquired six companies. This year in '26 we expect 20, 45 to 50 million revenues plus five, six, seven more acquisitions this year which will obviously increase the revenue number for this year. When we started we had some hypotheses, to be honest AI back in '24 was not one of them, to be very honest. But we believe AI is super important and we're all super pumped about the power of AI and I want to show you that in a second. So as I said, in the last two years we built a platform, we built a great team, we had a lot of hypotheses that we mostly have been validating, added some of them like AI, and we're really ready to accelerate the growth. And I think one important part to just show you how important AI is for Altamount: we set up a dedicated entity we called it Altamount Ignite with the objective to ignite AI in our portfolio companies. And this unit Altamount Ignite is run by Leonard. Leonard, where are you here? And Leonard will be showcasing next door one of the use cases that he deployed together with Peak Mobility. So we had it a couple of times here before: Peak Mobility, public transport provider in Germany. It's the market leader, I don't want to say the world market leader but one of the global market leaders for depot management. And it's a mysterious word, depot management. You go out on the street, you see all the buses, the red buses driving passengers from here to anywhere else in Hamburg. In Hamburg, there are 1,200 buses driving around, you know, during the day. And at night, all these buses, they have to park somewhere. They come to a depot. And in these depots, they are being maintained, cleaned, the wheels are changed, and you know, everything that needs to be done. And this is a pretty traditional process. In Hamburg it's software-based, but globally we're below 15% digitization in these steps, and this is a global phenomenon. So this is something where Peak is really, really strong. And now you know, if you have a diesel bus, it's pretty easy: you fuel the bus within 7 minutes and it's ready for the next day. With the e-buses that are coming more and more into the cities, it's much more difficult. It takes 6 hours, 7 hours to fully charge a bus. And sometimes this is happening during the day. So if he's doing a ride, he cannot just easily get recharged. So he needs to go to a depot. And this needs to be integrated in the schedule and the planning of the buses. And this is what Peak is offering. And we're the first on the global market who are currently developing a solution where we have a dynamic route planning the next day where this whole charging and maintenance of these buses is fully integrated for the depot. So it really becomes a dynamic depot, and Leonard will be showcasing that in the room next door. Pretty exciting case. A second use case that we have: it's a company, it's also a portfolio company of Altamount called Ecomdias, run by Christian here, Christian Eard, very nice company. It's a secure workflow automation for critical infrastructure but it's mainly for the police in Germany. So what Ecomdias is doing: they provide the so-called online police station in Germany. This is where the 80 million or 60 million adult citizens in Germany can file a police report. 'My handbag got stolen, my car has been crashed, somebody killed somebody or whatever.' Just the filing of a police report. You have two options: you go to the police station or you do it online. Those are the two options you have in Germany, and we are providing the online channel. So there are a lot of police reports coming into the system 24/7, and somebody at the police station has to read all the reports and prioritize on what to do first. And this is a very manual process. And then at night, if there's a police report coming in at night, there must also be somebody at the police station who reads that report, which is quite expensive to have people sitting there 24/7. And what Christian and the team of Ecomdias have now developed and are just about to deploy is a fully automated prioritization engine. So it reads all the different police reports, and if there's a murder or a bomb threat or whatever, it gets fully prioritized, and some of the others are just held back until 7 a.m. in the morning so the night shift doesn't have to deal with that. So it's a very powerful solution that's also decreasing the cost for the police, which the police I think is very happy about, and Christian will also showcase you on the live portal later on.

Yeah, thank you very much, Jan. And indeed Par was a great school for me to learn everything that could happen in a software company, but this is not the topic today. What I'm going to show you today is an example of an ongoing Chapters Momentum initiative at our opco CSB. So CSB offers an ERP, or often called dealer management system, specifically for motorcycle dealers, and we are market leader in this market. We have about 900 dealers online, which means that about 1 million service appointments are going through our system each year, while the booking process is still quite manual. Most likely the dealer has a customer on the phone and types everything manually into the system. So there's a huge amount of digitalization potential, and this is why we decided to set up with the Chapters Momentum initiative a project, and we are now building an AI-based digital service assistant for CSB. The idea is that the service assistant can be implemented into the customer's website, and then end customers — our customer is a dealer — the end customer books an appointment directly into the system by using natural language. And what did Momentum specifically enable about it? I will keep it short from a top-level perspective. First of all, we gained a lot of speed. The idea of Momentum is that we take external capacities into the projects. Those were recommended by Tobias and financed by Chapters, and those external guys were able to ship the feature very fast, so there was no disruption of our existing roadmap. And secondly, we have a way more capable product now through the use of AI, which from a high-level perspective means that we found a solution that the AI is directly connected to the on-premise ERP system, so it can talk in real time to the dealer and is always up to date, and this makes it possible to set up an appointment whenever a service assistant and mechanic is available and the customer wants to have it. Finally about the timeline and some economics. We started that in May 2026 and it took us only 3 months and about 20 days of internal capacity to build the first MVP that is fully tested, running internally at CSB now. And we are planning to deploy this to the first batch of customers. We have about five customers that committed already to pay for the solution; they are waiting for it. And once they are satisfied with the solution, we plan to further push it into the overall customer base. Based on some estimates we made before, we assume that we are able to create an additional revenue stream for CSB of €350,000 per year for subscription fees for only that small add-on. That's it so far. If you are interested in more details, feel free to stop by later. I'm somewhere over there in the back in one of the rooms, and I'm glad to be there for your questions.

Thank you very much. I'm going to talk about the scorecard for M&A purposes. Before AI, we used to have to dig through all sorts of information, just from the website, finding out whether a target can be interesting for us was not that easy. And basically we wouldn't have the idea whether that target would be a good fit to our guard rails and whether it would really pass or sustain in a shark tank. So we came up with an AI solution, with an AI lens as we call it, which will find out whether a target is really interesting to us. But just using AI isn't always the goal you want to use because it would give you randomized output. So we had to train the AI. But we did it for a specific sector. So we need to train the AI what is a key definition to us in that sector: whether it's a core platform or whether we could treat it as an add-on. And most importantly, the AI also needs to understand how software is bought in that segment, what do our customers really need in that segment, and of course the AI also needs to know what are the current market trends to actually score the companies in an initial setup. So that's how one scorecard could look like. Basically telling us what's the owner structure, can we actually buy it, does it check out all the quality gates which we have in place, and gives us a lot of context of that company in a matter of seconds instead of just digging through it and having first meetings with targets who are not really interesting to us. So if we do it on a bigger scale, for example for logistics software, we did it before actually entering the market. So we can actually see up front whether we want to go into that market. So we see on the top right-hand side that we have lots of core platforms available, and which of those can be really addressed. And highlighted up here are the ones where we were actually in talks with them. As you already see, there are also some optimization layers which we could add on at a later stage. And finally, what that score gives us is fast analysis and a good personalized dedicated outreach with high response rates and a high conversion rate. And of course it gives us the market overview which I've just shown you and how we can complement those targets in the long run. Maybe for some cases actually build an industry standard in the long run. I'm happy to talk to you later on and looking forward to your questions. Thank you.

Perfect. Thank you so much. I'm very glad that I'm a human in you too. So, maybe first of all, I was a professional athlete. Now I have a company Fuksum. Where does Fuksum come from? Fuksum comes from 'future of XM.' That was our initial idea. Five years ago I was based in Berlin, there in a university, and I thought why we are still writing exams with pen and paper. That was my initial idea. And now we built a much larger solution. We built a full Z solution for universities, the complete operating system. And you are thinking now, why does this guy do that? Because I recognize because we did market research, that was the only thing I as a student could do, and we saw a paper that 120 universities use 1,800 different software solutions. And when I recognized that, that was the moment where I saw that we need an all-in-one solution, and that is exactly what we're doing now. We have now 30 customers all over Europe, we have six modules. We are building now 90% with AI. So we were very glad that we were, I would say, funded born in the age of AI because it helped us very good to build an all-in-one solution. And now we have six main products: Fuksum Apply for application management, campus management, learning management system, we have a mobile version, we have the examination tool which was our initial idea, and our alumni management. We are doing a lot of other very, very cool things. I'm glad to see you maybe and talk to you maybe later, and we'll see us there in the back corner. Thank you so much.

Yeah, thanks for the introduction, Yan. A little bit disappointed that you did not introduce us as the Kälte Helden, the coolest cooling company in town, as you used to, but I'm fine with it. We can talk later over a beer. So we are the Kälte Helden. We are 23 people in the HVAC market in Germany, in Hamburg. And I think AI is one of the biggest opportunities in our industry, that maybe ever happened. And that's why we started to look at our problems and search for solutions. One of our biggest problems is that our technicians need to write a service report after every service call. After a long day on a rainy rooftop, they need to type it in on a small tablet keyboard. And to be honest, a lot of those reports are incomplete, and a lot of information is missing. That's a heavy problem for us because on that information we build our invoice. So we probably lose money when our technicians don't write the report properly. So we started with the AI Hub to create our service report buddy. It's a GPT that works along with the technicians. They don't have to type everything now; they just talk about the problem they had. It can be completely chaotic. AI structures this, and the most important thing: it asks. If AI sees something missing, it will follow up. And we did something that the GPT also has technical knowledge. So it's not a master craftsman, but it's kind of an experienced journeyman. If there's a young technician and just saying 'okay problem is solved,' the GPT will ask 'what was the problem? what was the root cause?' If the technician doesn't know, the buddy will give him hints where to look, where to search for the problem. Afterwards, we get a clean report and just put it into the ERP system and build our invoice on it. So more complete reports, that's pretty important for us. And to be honest, it's just the beginning. We have a lot of problems in our company and industry that can be solved and be more efficient by AI. So all people can focus on keeping the future cool. Thank you.

Yeah, thank you very much. And last but hopefully not least, I've got something very operational for you today. So basically, what is it we do? I asked AI to do a picture as well, and I think it's telling you exactly what we're trying to achieve in Germany: public safety. So our software solution helps the authorities to keep criminals and other bad people away from guns, and as well we do that in terms of nature. So we help Germany to look green. In terms of numbers, this means we have around 30 employees, around 450 customers, and that's around 90% of the relevant market in Germany of gun, fishing, and hunting licensing control. So this means around 5 million firearms in the hands of 1 million people. And that's basically our business. We like to talk to our customers, and we have something which most companies have: customer success managers. Those people need to prepare for their customer calls. We found out that this is roughly 1,200 cycles a year at least, and at least 18,000 minutes spent in our small company just gathering information from different systems. So you know what's up with the customer: what tickets did they raise in the last couple of weeks, what opportunities do we have, what do they talk about with our support teams, so is there an opportunity for an upsell or something like that? And what we did: we needed something which reduced those 18,000 minutes because it was around six systems you need to look into, you need to do manual research, and if you're not that seasoned it may take a while. We used the AI Hub for that. So we have a compliant opportunity now to link all of our internal systems to the AI Hub and create a very easy report within seconds about everything which is happening at a customer for the last couple of weeks, months, a year, whatever you want to see. So this enables us on the one hand to have very, very good conversations with our customers. So it makes sure we think about the right people and the right things, and it makes sure we have a very clear logic applied to that. It knows our products, it knows the questions the customers raised in the last couple of months. We know what's happening in sales, which conversations there have been, and so it gives the customer success manager a good idea of what is happening at the customer at the moment. And this takes less than a minute. So it gives us as a very small company a lot of hours. So 300 hours or more just on time back in terms of better conversations with customers because we don't need the time to research, we have the time to talk to the customer. It's a very small operational example, but it's in the AI Hub. So it means it can be shared across the community of companies we have on the one hand, and we have a couple of other things coming out of that because we don't use it in this specific use case. I actually use that before every conversation I have with a customer because it takes me less than a minute, and then I don't need to log in to different systems. Nothing like that. I just have one prompt that looks like ChatGPT. We all know that. And I get a report with very easy to read traffic lights: green, yellow, red, for different categories like sales, support, all the other stuff. And if there's something I'd like to know a bit more in detail, I get into the report and have all the data ready at hand in just a couple of seconds. So this for us is the biggest opportunity apart from AI, agentic coding and all the other things we've seen today in an operational area between customer support and sales.

Um, so there are fridges with drinks. The bathrooms are at ground floor and a bit hard to find. Don't get lost. We have who's in here? Folks and Kit in here. I'm going to be in here. You're going to be in here. Mark's going to be in here. Toby's going to be in here. Andreas, where are you? Down the hall. But down the hall. Yasha, where are you? Also down the hall. Okay, but mingle around, ask all the questions you want. Whoever has a dot has something to say. So, thank you very much.