Good morning everyone and welcome to our webinar, EOP of the Future, the AI Orchestrated Enterprise. When I have conversations with my clients, I usually ask the question, if you're building your enterprise operating model from scratch today, would you choose your EOP? And lately more often than not, the answer is no, I would not. So I'm very curious what your answer to that question is. I'm Tobias Kjern, I'm part of the digital transformation team here at Implement Consulting. I'm operating out of the Swiss office, this is where we today broadcasting from. And I believe that the era of the monolith is over. Yes, thank you for joining. I'm Marek, also from the Swiss office. And I come with a technical background and 10 years in transformation experiences. So before we kick off today, we would like to know a little bit more who is with us. So if you want to put in the chat the industry you're working in, the country or region where you're from, the role you are having and what your experience is, everything related to EOP transformations. And if you want, you can add a yes or no or whatever your question is to the answer. So while the feedback is coming in, a little bit admin for today, today's sessions will be roughly 40, 45 minutes. Give a quick introduction about the burning platform or why do we believe that the monolith is falling. We give you insights in alternative we see. We give you a bit more detail about how it really looks like or could look like in practice. So we have an example from a Swiss cap company where we have currently those conversations with. Then we will introduce it to something which we call the intelligence flywheel. And yeah, what that means for you and give you a little bit of next steps, how you could move forward. And we will have some time for question and answers. So I would like to ask you whenever you have a question, put it in the chat right away. And at the end of today's sessions, we have some time to answer it. So we have Switzerland online with Denmark, welcome Germany. Interesting, interesting. So as I said before, I do believe that the era of the monolith is over. What do I mean with that one? So we put together a couple of numbers. If we look at our P01 clients in average, they spend 48 to 66 months for an EOP transformation, costing anything between 120 to 350 million euros. And surprisingly, less than half are really happy with the result. But those numbers are not what really shocks me or really is bothering our clients. What is our clients spend four to six years for an EOP implementation. So the moment you go live, your system is already four to six years old. Your requirements have been identified six years ago. The processes you build, you build six years ago. And that might have been acceptable in the past when, you know, the EOP lifecycle was anywhere between 10 to 15 years. So spending every 15 years, five years is OK. But nowadays, everything is moved faster. It seems than every week a new AI model is, you know, outperforming the previous one. Companies cannot spend six years for an EOP transformation, despite all the costs and if they're happy or not. It's simply far too long. And some clients have to pay. And some clients, you know, notice it's at the later stage, though just recently there was an article in the German newspaper about the SAIS, a German optics company. They stopped their SAP implementation after six years and after they already spent 200 million euros on it. Yeah, I think the problem statement is quite clear that these kind of transformations are an issue. What is not the solution from my perspective, I think adding new Gen AI or GenTQs cases is probably not going to change this fundamentally. Right. And we bring an idea. How is this could change fundamentally? And this is done by a different architecture fundamentally. Right. So in general, we see three GenTQ shifts, which makes a orchestrate, an AI orchestrated enterprise possible. The one most are talking about is the GenTQ coding. It's quite famous currently at the market. Everyone is talking about. Then we see the second one, the GenTQ problem solving. And we see a third one, the GenTQ data facilitation. For the GenTQ coding, it's relevant in the ERP industry for the custom solutions, which are built either within the core, extending the core or on a platform or as a completely third party solution. These become rapidly faster and cheaper in future. So this is has significant impact on the transformation, how you run a transformation, how you execute it and also about on the total business value and business case of an ERP transformation. This agenda problem solving will tell how processes are going to be executed. So of course, we now in the past, we have the fit gap workshops or fit to standard workshops where you design a solution. You hopefully have some end to end business process owners who are incentivized in a way that the processes are running smoothly in future. This is going to be either matched by agents or completely taken over by the agents where agents can be executed. They can have a full context window about enterprises starting from emails to the data, data platforms up under the data lake. So thinking about the future where the context window is increasingly big, the agent would problem solve around how processes are going to execute and then on a daily level how to improve them. And then the third one, I just started a bit with this thinking in the future. We have a context window where the context window is quite big. And the agents could take in the full enterprise data foundation. And then we are getting to a situation where the agent could manage the data end to end or facilitate the data into end. From an incoming email where a vendor or client might change the address to invoice data which are changing from the bank. The agent would already have all this data available so that data governance, data cleansing, etc. will be managed directly by the agent. Of course, there are some limitations. So for some critical processes, this may not be feasible. But we see also today businesses don't have fully clean data and they're operating well or well enough for the for their business case. And this could be fully managed and future by the agents. So this may drastically change how we think today about data and data facilitation versus how it may be facilitated in future by agents. Let's see how these three shifts are coming together. So we think the ERP shrinks to the record and this is not a completely new trend, but the trend we already see in the market. We have this clean core ERP trend, which we see already for a couple of years together with the cloud trend. This we need to take to the next level now considering the agentic shifts as well. So we're coming from a left hand side here where we have this monolith, the ERP monolith and around that the data warehouse, the consolidation and some other best of breed solutions. This could change a bit taking into consideration the three shifts just discussed, right? On the very right you see this ERP core record much smaller. There you may have the financial processes where you need a hundred percent accuracy on everything you're doing and you need to be auditable. You need to generate like very reliable records. But then majority of business operates in a slightly different approach, right? The goods movements, the deliveries, et cetera. Would be facilitated through the agentic coding, the hands. So this is more or less a UI. So you can have best of breed solutions or fully custom coded solutions with agentic coding. And this will become much faster to code in future and also cheaper. And businesses need to think about how to internalize the profits. This is what we see right now is some concern in that area that different stakeholders in the markets, of course, try to internalize these profits coming from the cheaper and faster solutions. Then the agentic problem solving, which sits in the middle. This will dictate in future how the processes are being optimized and designed and run. So this agentic problem solving agents will have the full context window and knowledge about the company end to end and can then really optimize end to end with all the knowledge. And at the bottom, we have the agentic data facilitation, the memory. And as I just said, there we will see similar trends, which we see with the agentic coding, moving also over to the data topic. So for agentic coding, we already see quite a drastic change in the market. So less and less developers are coding, but reviewing. So moving from execution to the revenue to the revenue. And for data, we may see something similar that it's much cheaper and quicker to let data be facilitated through agents instead of having data governance processes in place where you, for example, submit a master data change through a form, et cetera. But facilitated by the agent. Next, I want to change. I want to give an example how we are. We have seen this kind of question. Actually. So this is how it can work in practice. We had discussions with one Swiss mid cap company, which was approaching us with an already business case and transformation in place. So there we had this very typical on a list one transformation, which we see here on the left hand side. Right. So it was a for our transformation of our upgrade and transformation approach. We had the data warehouse, the CRM, the web shop, of course. We had some integration layer and which was planned. We are sub BTP. And then we had some product information management system and warehouse management system. We were talking around this was mid cap, right? So not exactly the numbers we have seen at the beginning, but but mid cap example where we were talking about a transformation or costs total cost of ownership of the transformation of roughly 16 million. Right. Over time period of 37 months and 3,000, 3,600 days to execute this transformation. And then almost 700,000 Swiss francs external run costs. So I would say this is a rather typical business case for for running one of these traditional ERP transformations. Now we wanted to think this through with a new agentified shifts. And then we see quite a similar picture, which I had on the earlier page, a bit more generic, but but here more and more precise. We see exactly the three shifts I described. Right. So we have this agentic problem solving at the top. This can be this will be an agent which is having the full context window like a data lake, so to say, in the in the context window and will then on a continuous level optimize the processes end to end. Below that we have some best of breed solutions. Rapidly built with agentic coding for P2P pricing and order processing. And then on the very right hand side, we have this ERP core, which is mostly used for the finance and controlling. Below that we see the agentic data facilitation. So we will not have the very traditional data governance processes in place just for some of the data objects. We still have it. But for the majority data will be facilitated through agents which directly take data from the emails. From some other incoming data channels, et cetera. And then at the bottom, the PIM and warehouse management system will be mostly untouched. They will proceed as planned on the left hand side. Taking this new approach. We would have a. In total, much lower total cost of ownership. Roughly six million. We will build solutions much more rapidly. We're talking about a five to nine months. Go live dates for the individual solutions. So we are breaking the solutions down and can realize value for the business much earlier. We don't need to wait for the full S4 HANA or ERP go live at the end of the transformation. But we can set individual solutions live after these five to nine months. For these individual solutions, we are talking about roughly 700 person days. And then in the end for the one costs, we are ending up with an almost similar or slightly higher costs. This is already taking into account increased credit costs from the from the agents. So overall, this would be a much faster solution to set individual solutions live for business and can realize value earlier. value earlier. Exactly. A couple of things I would like to mention here as well. So for that setup, it is still ongoing. It's not going to be a big bang. So it's not like, you know, we do everything at once. So you might ask where is the best place to start. Our typical answer, it depends and it does depends. It depends a little bit where your situations where you're coming from. Well, for that particular client, we choose the value based approach. So they identified the biggest gain or the biggest challenges they're currently facing in their pricing and into purchasing processes. So that is what we picked out first. First, excuse me, as a front end solution still with the data in in the EOP and the current data lake. And at a later stage that will be then also transferred to a genetic data facilitation. There might be other cases where you just started with your data cleansing activities and implemented data lake or whatever. Then obviously that will be not the place to start with. And there might be others where you're looking for a best of breed solution. Then the authentic coding might be the place to start. So it depends a lot where you currently are on your journey. But important, we would not recommend like we do with the monolith to do a big bang, really focus on what gives you the biggest value or what solves the biggest problems and so on. What is obviously important if you are in a situation like a lot of our clients are, it's end of life cycle. So you might need to find a workaround for that. Do you need to replace your current EOP right away or keep a part of it? And then, as we just explained, yeah, solve others first and then at a later stage transfer it to whatever the final EOP solution might look like. What we have not discussed yet. And there's a conversation I like to have in particular because I think that is the biggest lever with AI and AI is new to all of us. And to be honest, we do not really know what is happening, what is not happening. But what we see and that is the biggest differentiator, in my opinion, to the current IT world we are living in, that AI support, AI orchestrated processes can improve themselves. And that is what we call the intelligence or authentic flywheel. So let me walk you through that one. So in a standard EOP approach where you start defining your requirements, then you start defining your processes. Then you do your customization, your ABAP developments, if it's SAP or see if it's Microsoft or whatever. Then you start rolling out, you do your testing and then you go live. And then with the moment you go live, as I said before, all your processes, all your requirements are already a couple of months or even a couple of years old. And then everyone who was involved in larger transactions also knows that once you're alive or it is a little bit set in stone. So you're not going to reinvent or optimize your processes right away. So whatever you go live with, stay like that. I don't know for half a year minimum or for a year or even longer until you do a process optimization. So with an authentic process orchestration, every single time suppose this runs through, it can be set up that is proving itself. So with every time you do a month and close, you optimize with every time you run through a purchasing process, you optimize. You could imagine it's like you have a team of employees which only does one thing whenever a person executes a step or ever system executes a step asking, is there anything we can do better? Is there anything we can improve? And if the answer is yes. OK, let me implement it right away. So in our opinion, it's not about only, you know, be the first one to implement a new technology or save money doing it or save time doing it. So I think the differentiating factor will be the ones doing that in a proper way very early. We'll have an advantage to the competition who is not doing it because with every single transaction, with every single interaction, with every single process execution, it will optimize itself. Let's say, you know, what does it mean for you? What is a proper way forward? And as I said before, there is not one way forward. It depends a lot where you are on your journey. But basically we say, you know, there are four pillars you should, in our opinion, consider. One is keep your legacy core. That is also what Mark said before. And that is a trend we are seeing lean core, however you call it. But keep that in place or don't touch it, at least not for now. Make it smaller and smaller over time. But keep that. That is the record that is for your audit. That is very important for your organization to run forward. Then what I would also say build or differentiate, not do a big bang, focus on what really gives you value. Solving a problem you're having or optimizing processes or whatever it is. So try to build on that one. Leverage that compounding advantage or whatever you do with AI, with agents, do it in a way that they can and will improve themselves. Don't be shy of it. It will help you in the long run. And then the last one start with business cases. Or that is something we have not seen in the past with EOP transactions or transformations. That is something why so many people say they are not happy with the new EOP system. There was never a real business case. So it was we have to do it because it's end of life or I don't know Microsoft SAP called us up. There's a new tool and we need to implement it. But there was a really a business case. So don't do that mistake. Really focus on what is important for you. Where is the business case behind it and start with that one. With that said, if you want to read a little bit more about that, feel free to download our white paper. You see the code here. Make a picture, download it or even better, reach out to either myself or Marek for 30 minute conversations. We are happy to start discussions with you about that one. We are happy to link in our colleagues from other locations. If you don't sit in Switzerland, talk to our colleagues from Denmark, Germany, US or wherever. Feel free to connect to us with LinkedIn as well. Reach out. We are more than happy to walk you through our thoughts. Answer your questions. Yeah. And talking about questions. Let me see if there is something in the chat or if you have questions, put it in the chat and then we try to answer a couple of them for the remaining minutes. So there is already one here. How do you ensure 100% compliance in the EOP record when using more, and I need to get closer, using more approximate methods for the authentic business process? What checks human AI need to be in place? That's a very good question. And there is not one answer fits all. As before, it depends. It depends a lot. So the challenge is going to be really separating the authentic bits or if it's data or if it's the orchestration or if it's decoded from really the EOP core. So making sure that the record, how we call it, is, I'm not going to say safe, but is stable. And that is where your books are and that is where you audit on. The other thing is it needs to be a little bit shift in mindset. So I do know tokens are not for free and I do know a lot of companies spend a lot with tokens, but in general it is cheaper than labor. So we work in risk lines from Swiss banks, which are in a very regulated environment. And what a lot of them do, they basically run the processes 100 times in parallel. That sounds weird, but with AI, you know, it runs two minutes. So who cares if it runs 1000 time or 100 times two minutes in parallel? And then you can really, you know, if 98 come to the same conclusion, you're probably very safe that that is exactly the right answer. And there are companies in Switzerland will already do that with approval by FINMA. So FINMA is the regulator here in Switzerland. So you are very, very safe that it gives you the right result. So it needs a bit long story short. It needs a bit a change of how you approach things and don't be shy for redundant work. I know in the past we always say we need to keep it streamlined and, you know, avoid redundancies. I think that will change a bit and don't be afraid of redundant work to assure that the process and execution of the AI change is correct. Let me see if there are other questions are coming in. Perfect. Thank you very much. I hope that gave you a little bit of insights. What we think, what we believe the market is heading towards. If you want to discuss again, reach out to Marek and myself, download the white paper, reach out to our colleagues and we're happy to continue the conversation.