Okay. So this is going to be about from moving from rules to go more into autopilot mode. So it's really about how can we just be a little bit more autonomous to get a lot of this manual work basically out of the door. Okay. How could that work? Oh, yeah. Well, that's me, Unshultad of consulting, 12 years of Magnolia. So for the ones that download me yet. Yep. Okay. So we just saw it perfectly already in the slides before. If we speak about personalization, how it's really done traditionally in the market since ever, ever, we are going to hit absolute scaling problems. That is something that we see all over the place again. So it just starts to multiply out 10 segments and then we have to create all those content pieces. We have to monitor them. So in the end, it is just a lot, a lot, a lot of work that is actually on the table. And then sooner or later, you're going to hit those really nasty discussions about, hey, you put in so many man hours and where exactly is this return of invest? And, yeah. How could it be done better? We thought about how about we just going to ditch basically the typical approach from identity? What is a persona? How about we just get rid of this and we're going to switch to real-time intent? So we really just care about what is really the interest of that person. So the main idea behind it is, for example, I bought a MacBook somewhere and then suddenly all the data points click, hey, we have here this Apple lover, perfect. And I'm going to come back to this page and I'm going to get peppered with ads for the next iPhone, for example. There's a high likelihood that my intent is completely different. So that means at this very moment that persona starts to fail me. I mean, in the worst case, for example, I want to deal with a return or whatever it might be. Bad example. But you get the idea. Just the profile that I got right now is not necessarily really a good match for the intent right now. Okay. So a new way could be basically that we start to really look at the user, what he does on this very page. Because in the end, he gives a lot of information away just by interacting with this content. Quick opening a page, that could be something that is already interesting. But it really becomes much more interesting if I really see how he interacts with that content, really. Because if he starts to really spend time to read and so on, that gives me a pretty good data point that this very topic might be of high interest for this person. And from this one, it's becoming much easier to basically come up with what is the intent and how can I really make sure that this person has really a good experience. Okay. So, I mean, I say the old way. I mean, this is how everybody does this. We just focus on a persona, on a segment. Another way would be really that we exclusively – or not exclusively. I mean, you could easily mix and match. But another way could be that we really start to focus more on the actual intent of the user. Because especially nowadays, we all know it. The usual attention span becomes smaller and smaller and smaller. And somebody said, I don't think it's right. We're just coming close to a goldfish. Well, I don't think that's true. But what is definitely true is that we just become much, much quicker in our decisioning and our typical journey has a certain tendency to be much more wishy-washy and not as engineered as we basically plan those things typically. Okay. Let's have a look how it could be done. So, the idea, first of all, is that we are going to tap into order that we organize our content universe. So, pre-AI, it was typically extremely hard to really classify the data the right way. So, it's a lot of work that you have to put in. And especially as the content changes over time, you have to rinse and repeat. I mean, this is obviously where AI is absolutely excellent at. So, you can just really, at scale, take your site or all your sites. If you have a good framework for the different topics, intents, and so on, you can just basically one-shot it. And you have all the metadata available. That makes it so much easier to basically spot the intent without all the other data. The next thing is that we are going to start to map the actual intent to the next best action that really brings value for someone that has a certain need. And we think that you can even hybrid because sometimes you have already quite a bit of content available. You don't even have to create something new. How about we just take those different pages and we create just a teaser out of that. Just a small nudge that makes sure that this person will go in the very right direction to meet this intent. So, he has a pleasant experience and he comes back because stuff actually works for that person. And, okay, now we have just basically a lot of metadata. Cool. We can now basically start to measure how that person interacts with the page. It's not just, okay, I just went to this and that URL. I just now understand pretty much, okay, high scroll depth, a lot of clicks and so on. Though his intent is definitely to, for example, invest into precious metals or whatever it is. And, yeah, with that we can start to calculate. And this is not even AI, it's just boring, good old math scoring models since day one, basically. So, we can really start to score that client pretty, pretty well on the intent. And the good thing also for regulated environments, it's not some kind of black box magic that happens. You can pretty much sort of prove why you presented certain bits of content that you have. All right. So, the core idea is really that the editors declare the purpose implicitly by that content. This is completely detected by the AI. And the engine by itself handles completely the delivery. So, there's a detection of the intent. We have a clear mapping of existing content or even dedicated teasers and banners. And those are just then automatically delivered by the recommendation engine. Okay. How could that look in reality? So, let's say, I mean, we have here this virtual bank. And I just simply start to browse that page. The slot here, keep it in mind, is on purpose, kind of visible. So, I start to basically explore the page. I see what's around it. So, we have quite a bit of, a little bit of scroll depth. Then I'm just going to look for investments on the Insights, what's actually available on this web page. Again, I'm not going to immediately bounce. I start to scroll. Obviously, this is sped up. If somebody really reads it slower, it's just to show you basically how some kind of unplanned journey could look like. And then, again, I start to look more into investment, into private banking and so on. And that gives me already some kind of idea of what might be really relevant for this user. And on and on it goes. Now, I'm going to hit the investor opportunities. Yeah. You name it. Okay. Finally, I just look into the, I don't know, what do I do? Sustainable investments. Same thing. It's not something that I immediately ditch. I start to really consume this content, which in the background really starts to accumulate, though, the intent. If I go back to the beginning, suddenly I just get content that is matched to this very intent. And the overall setup and the management behind this is suddenly becoming extremely easy. Because also here with AI, you could even just find what kind of content could be for this intent really relevant. I'm just going to create a simple teaser here with that thing. Okay. So, how does this work behind the scenes? Well, obviously what we see here is just the metadata. I mean, this one here is just the data collection bit. We just, for the different sites, we have the different trackers. We have the different topics that I predefined in the system. As we can see here, we can do this very specific for each and every single site and brand that I have. And each and every single tracking is just different. What I also can do is, since all the data is basically collected, we could just take this and get at least some kind of insights about what this user does. At scale, not so interesting at getting a very good scan about certain persons to form a better mental model. Perfection. Okay. So, we have all the data available in the intent marketing engine. So, I get all the stats that I can basically wish for. Okay. Okay. So, what I'm going to do now basically with this intent, how I'm going to place the different, not just placeholders. Pretty simple. Oh, no. Sorry. First of all, how do I basically get the actual metadata for the journey in? So, for that set, AI can just change everything. First of all, this is the data that was detected based on the different topics and interests and so on. Everything is just auto-classified. If I want to, I can obviously start to fine-tune this. And if I have to do this otherwise extremely labor-intensive work, I can just select the page or the full site and run a full-on scan on the full site to get this metadata that I actually need to detect the intent. All right. Good. So, as you can see, with one click, you can really get a lot of work done, which is otherwise not possible without AI. Okay. Good. Let's look at the Philly. So, how can I basically configure those customer journeys? I mean, what we see here, this is a metadata that you can then really set up for your different businesses. Okay. So, I can create as many journeys as I like. And what I can do is I can really set up in this journey for each different stage where I have a certain intent or a certain interest in a certain topic. This is where I can then set up those experiences that I will actually deliver in those slots fully autonomous without setting any rules. Okay. So, if I'm going to use here, again, AI helps a lot. I can also even start to run a content gap analyzer. I could just basically see in all the metadata that is available how many experiences do I really run, what do I have basically available. And in this case, I could then just also select one and get already a proposal by AI what it could be. Just some kind of helper, some kind of glimpse what to do. Let's pretend this proposal is exactly what I need. Cool. Let's create this experience. Basically, this connection between the content that I'm going to deliver to the actual intent. Okay. So, one click. I get that thing generated just for the sake of time. If I have a quick look at it. If we hit edit, then intentionally simplistic, I can just take an existing page, which will just create a teaser element for me. So, I can, again, just bring those costs down for extra material. Because maybe sometimes I have already something that will already satisfy this intent. And it helps, again, to make sure that the user reaches its goal as fast as possible. Or what Tobias also showed, we can obviously create modular content that we can then just set up super fast. So, you can create those content modules super quick and attach those to the actual intent. Again, the full idea behind it is just to make it as easy and fast as possible to create those contents without setting all those rules. Okay. Good. Last but not least, since we have all the data under control and we have the experience management also under control, we can also much easier start to measure success. Because I can very easily find out... Okay. Sorry. It's a different talk track. What I also can do is... I mean, obviously, this year is something that cannot be completely neglected. I can also just do more than intent. I can also create stage journeys. But, again, where I'm just going to... I'm just going to log in those pre-created experiences. So, you can just really do it free-for-all just to make sure that everything works. Or if you have really then a call that you really have to come up with a pre-configured user journey, you can do this as well. Another thing that I definitely also want to show you here exactly is this bit here. You can start to learn on this data and you can basically make sure that certain experiences that are clearly over-delivered and they are not getting any hits, that you automatically reduce the number of experiences that you deliver in that thing because they clearly don't work. Okay. Really good. But, exactly the last thing, how to set it up. Super simple. You just strategically decide on your website where do you want basically to present those not just... Again, to make it as easy as possible just to get rid of setting up all those rules. Just have those strategic hotspots. The rest will be completely done by the system. Good. What is the strategic value behind it? For a marketeer, obviously the maintenance effort is much, much lower. You could probably reduce a lot of those logic trees that you have with all the different data points that you have to set up manually. And the audience matching would be much more automated. For visitors, obviously it's finally content that is really relevant that is not showing me this dumb MacBook, but it shows me how to get basically to the returns page or whatever it is. It's first-party data that for sure helps as well. So it just basically lands in your own repository, which makes also from the regulatory standpoint things a lot easier. For business, I think in general measuring the impact becomes much, much easier. You see the value of your content. You really see how it contributes to the journey in general. And I think if we speak about hyperpersonalization, that this finally could be really the unlock to solve a lot of the problems that we have since personalization was basically invented. And almost, just to reiterate a tiny little bit, how the self-optimization loop is also going to help to automate the system over time. Since we really have everything completely combined, we have the content, we have the metadata, and we have the clicks. We can definitely run constantly jobs to make sure that stuff that is simply not working is tamed down, so it's not doing any harm. And also that makes it then, obviously for editors, much, much easier to see what works, what's not working. And this should also make the optimization loop much, much more efficient compared to a lot of those distributed systems that we typically see. So in the end, measuring attribution beyond the last click is also something that should be much, much simpler compared to all these scattered solutions that we typically see. And with that, I would already come to the questions bit. Thank you, Jan. Are there any questions for Jan? Oh, there they are. There they are, yeah. Awesome. Yes, so absolutely. As I said, really, you have some personalization needs. You want to delve into this intent way of doing things. Drop a note. We're going to make it happen, and we're just going to pull it off together. Because we can really, if we have special needs, we can completely adjust to the stuff that really matters, also in complex scenarios. Thank you, Jan. Thank you. Thank you. Thank you. Thank you. Thank you. Bye. Bye. Bye.