Good morning. A warm welcome to day three here at the DSRG Annual Congress 2026. Looking around the room, there are some aftereffects from yesterday. We estimated that we poured out around 20,000, 1000 glasses of Kölsch yesterday. Yet you all look fresh and cheerful. Very good.
Gaffel Kölsch, which you drank yesterday, by the way. This company had already been thinking about AI two years ago. They didn't get very far with it, only as far as the term "delicious intelligence" — which does have something to it.
Artificial intelligence also played a role in Carnival here in Cologne. We had a brilliant start the day before yesterday with the Tanzchor Luftflotte. Yes, and AI also made an appearance at the Rose Monday Parade of the Cologne Carnival, in the form of a decorated float, and it looked like this.
So AI has us firmly in its grip there too, and it will continue to do so throughout the day today. On Tuesday it didn't take long and we went from Carnival to the keynote by Jens Hungershausen in the Wild West. And that's still the case now. We've made ourselves comfortable on stage. We've set up a little saloon, and I'd now like to welcome, for a fireside chat, the head of SAP, Christian Klein, and Jens Hungershausen.
Good morning. Good morning. Make yourselves comfortable. These are the kind of armchairs you never want to get out of once you're sitting in them snugly. Christian, great that you made it to the DSRG Annual Congress — you were just in the USA. Your calendar is full, not just with AI topics. Yes, but, yes, a very warm welcome. I'm glad so many of you are here on day three of SAPire — it's always a bit of a tough one. Yes, after a bit of a celebration the evening before.
Yes, but indeed, I really enjoyed it — SAP Connect in Las Vegas. I arrived late yesterday, couldn't sleep, had a bit of jet lag, so I put on a hoodie and walked around near the cathedral a bit. I also thought, Cologne has so much more to offer than Las Vegas — although [laughter] — and no Kölsch.
I have three topic blocks I'd like to chat with you about. It's about artificial intelligence, about the Autonomous Enterprise, about transformation, and about collaboration. Christian, at SAPire you presented the topic of the Autonomous Enterprise and developed a vision in which agents support business processes, but also carry them out partly on their own. Which business processes do you think will be the first to run fully agent-supported?
Yes, indeed, it's only been a few months, feels like years, and since then we've been very busy. I believe quite a few developers at SAP had a very work-intensive summer, and this week we've released over 200 agents in total. It covers many HR processes, finance processes, procurement processes, supply chain. We've fed over 7 million data fields into our knowledge graph by now — so to speak, into the agents' brains. We've connected many public APIs, and of course the point now is — and this is a path — to keep making the agents smarter and smarter. But I can only recommend it to everyone: great work. I've already heard that some people are using it, also on the DSRG board, dear Thomas, and I can only recommend it to everyone, because you can already notice it in my daily work. We just had our quarterly close. In the past, I would always go into my BW and look at the numbers again, or my communications department, my investor relations department, would present the numbers to me. And today, all of that is handled by voice. I ask Joule, give me the SAP numbers, give me the analyst comments, give me the analyst figures — and it prepares all of that nicely. Then I ask what I should say so that the share price also goes up — and even Joule has an answer to that. [laughter] And you can already see how work is changing. We've made great, big progress there, but yes, it is a journey.
How does DSRG, how do the customers see this journey, Jens?
Well, I think when the Autonomous Enterprise was announced, we first said it's good that there is exactly this vision, and I believe that as a tech company, SAP is doing exactly the right thing — to have a vision of where you want to go. The question in the end is — and this also has to do with transformations, and I already spoke about this on Tuesday — the challenge that comes with it is how can I actually make it usable. And that's where we look at it from the perspective of existing customers, from the perspective of the customers who first have to create the prerequisites so that they can use it at all in the end. And it's fascinating what is possible technologically, and the technology is there, it works, I believe. But I think we in the companies simply have to look at whether we have the prerequisites for it. We still very often discuss in our working groups, our committees, the topic of how do I do my transformation to S/4HANA, how do I perhaps do a cloud transformation, is the cloud transformation the right thing for me to do. And I think we have to make sure that we bring together the speed and the expectations that SAP has on the one hand, and that the customers on the other hand simply cannot fulfill, in a better way. And I think that's where we as a company and the members of DSAG are still heavily dependent on getting more support from SAP on how this actually works. I believe AI has enormous potential. It will change business processes, it will also change business models. We notice that in our daily lives too, in how we go about it. But it is a real challenge to actually make it usable. And in the end, of course, it's always the topic that SAP naturally wants to monetize it. That's a big point. I can understand that too, at first. I also work at the company. We also like to have a positive result at the end of the year, but in the end, companies really have to look and say, how do we actually map this so that we also have added value from it? It has to pay off in the end. It has to bring added value, and I think that's where we have, especially regarding the Autonomous Enterprise, a great vision — it's good — but the path to it will really still be a challenge, I believe. And I think that's where we will have to coordinate very well with each other in the coming year, but also in the coming years, on how we can simply implement it, how we can do it, where the benefit really is, and that has to be proven. So I think those are really the challenges in the vision that I perceive, also in the conversations we've had here at the annual congress.
I'd like to pick up on two points. First, the support from SAP. What are you doing there?
So, I mean, there are clearly some voices in the market that say, with all this AI, do we even still need software, do we even still need SAP? And I believe we do well to... There were also many voices that said, why doesn't SAP develop a large language model itself? I believe SAP's strengths lie in our incredibly deep process know-how, in the data know-how we have, in the business logic that we've developed over 50 years. And I believe we do well to, on our new AI platform, take these large language models — which are trained on Harry Potter books and other topics or software code, but not on your business data — and integrate them with our AI models, embed them, and of course also the governance. Yes, I mean, everyone has a certain amount of regulation, tax laws, labor laws that the agents also have to understand, and all of that now flows into the new platform. And what's also important is what happens with the data. For example, we don't hand any data out to the large language models. They shouldn't be trained on your business data, on your IP. That is a very, very clear rule for us on the new platform. And yes, as Jens already said, the path to it — and I know it from my own experience, when I became CEO of SAP 6 years ago — the cloud transformation, well, that took quite a bit of effort. It wasn't just an ERP change, there were many, many processes that had to run differently. A lot of change management took place in the company, sometimes painful decisions, and in the end you end up with an ERP that hopefully supports the business models of the future. And that's why we've also invested a lot in tools. We also bought companies in the area of business analytics, Signavio, LeanIX. We're now also building many agents for migration. This week we also released some agents for system configuration, simply to make this journey easier. I can only say again, in the end, the human factor is still the most crucial one. And when people often say — it always makes a good headline — "SAP project became difficult again, much more expensive." Well, I always say, to defend the company a bit, why does it work for 400,000 customers? Why does it actually work so well? Companies of the same size, same industry — and it works well there and it works badly there. Is it always the product? I claim that in 5% of cases it's a lot of customization in the past, little change management, little awareness of how to harmonize processes. It is far less often the technology. And I believe that's where SAP supports you on this path with tools, with expertise, with know-how, and of course also with change management.
Christian, the other topic I'd like to pick up on, which Jens mentioned, is: the numbers have to add up in the end. Where is AI, specifically SAP Business AI, already delivering measurable economic added value today, and where is it still music of the future?
Good. Well, we now have, over the past months since SAPire, around 900 test customers. We ourselves are a test customer, and for example at SAP, we now use Joule — 90,000 employees use Joule. We're already seeing productivity gains of 20% today in HR, human resources, finance, procurement. We're seeing even higher productivity gains in development when it comes to coding. Well, the LLM models are already trained on that, which means we're already seeing today what's actually possible with AI. I just mentioned, even the role of a CEO will change. The question is, if Joule someday also does the earnings calls, why do you still need a CEO? One might ask. So I notice it in my own daily life, and if you go through every function and what we're delivering now, it will lead to productivity gains there. Just think about every HR employee working with SAP systems, finance employees, procurement — how often have you clicked through the menus in Fiori, or maybe even still the GUI? And by the way, we also offer GUI now with Joule, so for anyone who can't part with the GUI, we still have that UI on offer. And how often do you dig through, look for the transaction code, need to release an order or run a cost center report, load data into Excel, build a PowerPoint, do an analysis on it? And today you just say to Joule, give the command: please give me an evaluation of this area. Joule checks, okay, do I even have access to the data, am I allowed to see and process the data? And then creates a completely finished analysis for management, for decision-making. It gives recommendations, gives me alerts when, for example, delivery delays occur on a project. All of this is now being brought to the market, and I believe a lot is already happening. One thing I want to say about that: of course, with these test customers we have now, a customer has many, many ERP systems, there are many, many third-party systems involved. Of course, the AI can only be as good as the data is harmonized, as the business processes are harmonized. You have to do more tweaking the more heterogeneous the landscape is. So no miracles happen. Yes, that has to be said too. That's why we're also taking the approach in consulting of getting more and more involved in these AI projects, helping customers understand the data silos, understand the semantics of the data, understand the processes. So it's not something you can just say, okay, I'm starting today. A lot works SAP to SAP, but especially when it comes to non-SAP systems, there's still some effort involved.
The verifiable, the measurable economic added value. You mentioned it, Jens, the urgent need among customers. How far are we there? What is the demand like? What is being formulated in the member companies?
Well, I think the need to actually make this added value tangible somewhere — that is, I believe, exactly the great challenge. And we know that from many, and it's not even SAP-specific, I believe, but also from many other AI solutions that you perhaps build yourself outside of SAP, where you do smaller things. There too, I always see the necessity in the end to look at where the added value is, and not everything works. And we simply have the situation — and also to come back to Christian's previous point — when I look at where SAP actually comes from: SAP's over 50-year history was shaped for many, many years by providing a system where it was explicitly said, here you have the system, please adapt it to your business processes, to your business models. You have the option via a development workbench to really intervene in the processes. It's not just the customization, and we customers gratefully accepted that — I did too. I don't exclude myself at all. In our company, we introduced SAP in 1997. And I still remember which user exits I fiddled around with in order entry, displaying new fields or building process shortcuts. That is exactly it, and that is the topic of transformation. SAP, with Christian, transformed in 2019, 2020 towards cloud, now towards AI, and I've often mentioned this in Walldorf. Now the customers have to be brought along. So if SAP's transformation, which isn't finished yet, is on its way, then the customers now have to be brought along and this added value made tangible. That is exactly the topic of standardization, and I see it similarly to Christian. It is a shared responsibility. Yes, on Tuesday I spoke about complexity. SAP landscapes today are complex at customers, and they grew complex, and there's a reason for that, but we have to reduce this complexity. The topic of Clean Core and standardization — from an IT manager's perspective, these are valid points, but I also have to get the business on board with it. It's not like I just go and say, well, I'll snap my fingers three times and then I have a standardized process, but I have to — that's real work. And that work has to be done, and for that we need time, and that's the only way it works. And in the AI environment, it's becoming even more apparent, because there we had the topic of data, we had the topic of complexity — yes, we have to see how it all works together. And I once talked with a colleague from a user company, and we talked about excellence in AI. And he then said to me, he said, Jens, look at where the input in the large language models comes from. That is the average knowledge of the internet. Where do you expect excellence from there? And that is exactly the challenge, and that's why... [cough] [applause] ...and that's why we have to go and say, it's encapsulated, it's separated, but we also have to make the data understandable. Large language models can't handle structured corporate data in the first place. That's why I had it in the keynote, the topic of data, structured corporate data, also the understanding of the data — that is simply a huge challenge. And what we as a company find is that SAP is moving forward with enormous innovative speed and trying to drag us along behind it. Yes, I'll put it a bit bluntly: the topic of cloud transformation, now also towards AI. And as a company — I come from a mid-sized company — you somehow feel like you're always stumbling along behind it, but you actually have no chance of getting ahead of the wave. And I believe that is a topic that occupies many of our members, because the majority of our DSAG members also come from the mid-sized sector. And there we simply have to try to find something together. And what you can also bring into this perspective: on the one hand, SAP with the products and innovations, with the maturity level of the things that are brought to market accordingly. On the other hand, we have the companies, but of course also the partners, and the partners are also called upon here. And we had it on Tuesday too — there was also a partner on stage who clearly stated that the partners also have to see that they adapt to these new circumstances, that they try to support the companies as best as possible, so that we really have the chance to move forward and do that. That is our claim as DSAG, but that is also our claim, I believe, of every member company we have here — that we want to be understood and that we want to be brought along.
So as not to lose the connection. We want to go into that a bit more in terms of transformation shortly. Jens, you survey extensively, you conduct surveys every year in the companies, in the member companies. How many companies, how many organizations are actually ready for AI?
So, I think fundamentally, every company is ready for AI in principle. The question is always for what exactly? And I also said it on Tuesday, I believe that the strategy of really embedding AI into business processes and going down that path, simplifying the processes, automating them, and perhaps in the end also autonomizing them where that is possible. The topic of human in the loop is of course also a topic that belongs with it. I believe every company can already do that. And I also believe that every company should really actively engage with it, to see where exactly the topics are for me. "Claim your ground, lead your business" — our motto for these three days — that's exactly it, and it pays into exactly that. You have to engage with it. You have to see where you stand as a company, what technologies are available out there, what SAP can offer me so that I can be better positioned in these processes, and then you have to develop your own strategy. And the strategy doesn't automatically mean that I go and say I'm going into a public cloud system. It doesn't have to. It also doesn't have to go in the direction of cloud ERP. It doesn't have to, but as a company I really have to think about what my strategy is, what I want to do, so that I'm not driven in some way by technological developments. And in the end, by the software manufacturers — it's not just SAP that's out there with various ideas on how to bring different products to market. And that, I believe, is the responsibility. Every CEO, every company really has to define that clearly for themselves, because otherwise it won't work.
How do you see it? Christian, are companies basically all ready for AI, or not quite yet?
That depends — it depends very much on it, and as far as I know, it also depends on the corporate culture, on the willingness. The topic, of course, of opportunity and risk. Let's take SAP, but I think SAP is a good example for actually all companies, all industries. When we introduced AI for our software development, and now Joule for all our employees, I do experience employees saying, where is my role actually in the future, if everything is automated like this, if it all works so well now? There are of course fears involved. And that also means you have to communicate well. Where do you actually want to go? And also to say, hey, we have 110,000 employees, and we're going to use AI to make this company successful. Will jobs change? Certainly. Will we perhaps also lose some jobs? Yes. Yes. But will we make this company damn successful for the next 10 years? Certainly yes. And we need AI for the innovative speed, for the way we develop products, what we develop, and how we work. Otherwise we're no longer competitive. And in Germany we always have this debate about where are the startups and the companies that became successful in the last 50 years? And I believe that with AI, entirely new possibilities are opening up in all industries. When I see — we just talked about the many problems with an SAP implementation today. Now that we have Joule and the agents in use, the employees have added over 500,000, 1000 skills themselves, developed them themselves, no longer have to wait for IT, and suddenly notice — instead of doing customization — these agents understand how I keep my daily work running, what activities I actually used to do with the system, and suddenly someone does it for me, and with a speed that would never have been possible before. And I think it's simply very important that you bring everyone along with it. The works council — the works council also came to me. We wanted to understand, okay, who is actually coding with AI and who isn't? Christian, why do you want to understand that so precisely? I said, I don't want to do anyone any harm. I just want to understand who is using AI, who needs training, who needs further education. And I think it's simply very, very important that you also bring along this human side, the people side, and bring everyone along with it. And I think then every manager must realize, it won't work without AI — in some industries faster, in some industries perhaps a bit slower, for example the public sector — but AI will certainly make its way into all industries.
Jens also mentioned a point earlier about how important it is to bring all the companies along with the high pace of innovation at the moment, with these transformation requirements — S/4HANA, cloud, artificial intelligence, everything is going faster and faster. What are you doing at SAP so that no one is left behind?
Yes, for one, I think over the last 5 years we've increased our research and development spending by around 25%. We now spend over 6.5 billion annually on research and development. And that was always very important to me, even when savings are made here and there, because we, as a public company, also have to be more efficient, of course also have to make sure we work efficiently. That's important to me because I see that this is a very decisive time also for the customers, to bring them along on this journey. Cloud, AI, what we've also seen. Yes, and I take that now — I mean, my side role is still research and development at SAP. It's a lot of fun, and what I've also seen is that many customers use all the features we've developed in recent years — some are used well, but some, only 3% of customers use them at all. At some point the question arises, how much do we actually still need to invest in the products, if they already cover relatively broad business processes today? And how much can we now actually invest in AI agents that might lift entire business processes to a whole new level in terms of automation and intelligence? And these are always the topics where we, of course, have to bring the customers along, but also have to explain to the customers where we're investing and why. And let's take the topic of cloud — we're now investing a lot of time every week, a lot of resources, to make the ontology even better, so that we can put even more intelligence into the agents. We're investing a lot of time to improve the governance of the agents. And in the cloud, I can deliver this development to customers weekly, and even daily now.
provide. On-premise, I'm waiting for my annual release, and you can really see that innovation speed is very, very important, because a lot is happening in technology, and we have to explain that. We have to help customers go along on this journey. I mentioned at the beginning, it's of course good to explain where SAP is heading, where the market is heading, why our customers become more competitive through that, and then of course also providing the tools and resources so that we can walk this path well together. And I can remember well, Jens, we had one or another emotional discussion about why some innovations are cloud-only. But it also always led to something, especially with DSAG, that we came to rethink and said, okay, we have to invest more in connectors, in on-premise development, in on-premise landscapes, hybrid landscapes. How do we support these better on the way to the cloud? How do we really manage to cover the entire landscapes? And there is simply this dialogue that is very, very important, and that SAP is ready to listen.
Of course, to keep going down the path. I believe the path is clear. But of course, never forget that we also have to bring all the customers along on the journey.
Yes, bringing customers along. Jens, what do customers expect from SAP so that bringing them along in the transformation works?
Well, we've already talked a lot about the challenges, and I believe many of our members and many of the companies, they simply need time to make this transformation. But they don't want to give up on innovations during that time. And that's a topic that has moved both of us in recent months and years. I think last year we had a small emotional clash over it. That we as DSAG go in and say, we have a lot of customers who are on the one hand working on-premise, who can maybe only work on-premise, for whatever reasons, I can't judge that. I have to take up the voice of the users, of our members accordingly. And they also want to use AI, they want to use AI, yes, and they also have an interest in using AI functionalities from SAP, they also want to use innovation. And in the end, for S/4HANA we have a maintenance commitment until 2040. That was very innovative, I would say, at the time this maintenance commitment was made. That surprised me and also pleased me that there is this maintenance commitment. But it applies in the end and it still stands. And there are many customers who are today on S/4 on-premise, and the expectation is already there that we simply have the chance to participate in the innovations, in order to then also give feedback to SAP again at the end. And to look at where we really have opportunities to develop things together. And it simply takes time in many places. Yes, on the other hand we had the topic of Fiori and the end-to-end business processes with Fiori applications. I also had a topic in my circle about what drives the community, that you go and say, how can this actually work? At the end you have to look at how it works with the new UI. And how you can put it together. And the development speed was also mentioned. Yes, you can also look at how you can use AI to complete things in one place or another. And what we as DSAG have always strongly initiated on the one hand and supported on the other hand is the whole topic of influencing. A lot has happened in this whole area, and as DSAG we have already initiated a lot of projects and made a lot of influencing requests, which were then taken up in product development on the other side. There were always some highs and lows in the collaboration on the influencing topic, but I believe there is also a chance there to go in with the feedback of the DSAG members and say, okay, where do we really still have pain points? Where do we really still have challenges in the on-premise or private cloud landscapes, where we say, man, we have to round something off there, we have to finish something there, so that we really have the chance to go exactly this path forward according to the new development guidelines, clean core, standard processes.
Yes, so I can assure everyone here, Jens has fought very strongly for your concerns. It also got emotional at times. I don't really like it with my friends from the press that we argue it out through the press, but I think you've also noticed that I listen very, very well, or that we listen very, very well. And I think we'll come back to it later, but in this moment I simply have to say thank you to the many DSAG working groups, because a lot of input comes in, and some of it comes back, I think. And where it hasn't come back, we've helped along a bit. So we really appreciate that. And I'd like to mention one more thing. Then [applause] as I mentioned at the beginning, it's a big change, I believe, for all companies. And if we now, for example, follow the discussion on-premise, cloud. We're currently having an internal discussion where I'm being shown that some features we've developed are only used by 1% of customers. Then I naturally ask the product managers, why are you investing in that? I mean, 1% of customers, couldn't that be handled through change management? There should be a standard if the other 99% can live without these features. And can we now completely rethink processes with agents? That means we're currently looking at how we can actually get the best developers onto agent development. That means we have to go this way. Why? You have to imagine it. I have a little radar on my mobile or desktop, and it shows me how the landscape, the competitive landscape of SAP, is actually changing. And suddenly many new points appear, and they're all AI startups in the areas of finance, procurement, HR, and they all want a piece of the cake. And that's where we have to keep SAP successful. That means I need the best developers on these points, because they have the know-how. I have to pair them with data scientists who train the models, who train the harnesses, etc. And I believe that's something I always want to appeal to. I believe that's simply a path we have to go together. Yes, we have to make sure we keep this path as simple as possible. But we also have to keep promoting the fact that the world, especially technology, is developing so fast that SAP naturally wants to develop too. I could make it easy for myself. I could sit here and say, come on, let's just let this ship keep going. I'm convinced of that. If we didn't transform SAP now, in five years we'd have an SAP in enormous trouble. And that's exactly what I don't want. I believe none of us want that. And that's why we have to go this path together, and we will continue to pay attention, especially with DSAG, with the working groups, to make this path as easy as possible for everyone.
Go the path together. You've also essentially been going this path together for 6 years, you two here on stage. Jens, I didn't introduce you today as chairman of DSAG. For good reason. You already said goodbye here on Tuesday morning during your keynote. Last night your successor was elected, and I'd say it's time to announce him and bring him on stage, because he will now also go this path together. You experienced him here on stage yesterday as board member for sales, production, logistics. Now he is the new chairman of DSAG, Thomas Henzler.
Congratulations again. Congratulations Thomas, and you can take a seat right away.
Ah, I thought I'd sit up here in front.
