So, I'm going to talk a bit about how it led from the learning system to AI-driven development. So, just five minutes, new situation for me, not much time, no time for jokes and anecdotes, but for two questions. Who of you uses Cloud Code? Good. Who uses CodeX? So, you will know what I talk about, so, but... Let's proceed. So, yeah, the trainer, some of you have been in my trainings. So, where do we come from? Yeah, it started 15 years ago. A side job, been giving trainings, material created in Confluence. I guess most of you know Confluence. And it's, like, not really fun to work with, at least for complex training material. It's still there, holding us back. It's a maintenance nightmare, you can imagine. Like, 20 exercises have the same light module name, path, or whatever. You need to update it. You get crazy. Yeah, and it's not a learning system, so, no enrollments, all these kind of things you want from an educational system. So, we knew, yeah, we need to change that in the future. And, especially, it's not really AI ready, even it has an MCP server, but I think it's really not nice to say it in a nice word. Yeah, and it's not how it should be. So, yeah, we started building new concept, evaluating different learning systems, what are possibilities, having all in Magnolia. So, we decided, let's do it in Magnolia. We have Magnolia, let's do it in Magnolia, have it all in there. More than just courses, also certificate management, all these kind of things you want to have with it. Enrollment, different levels of users can see different kind of content and so on. So, that was the decision. And, let's use ASCII doc in Git as being the course content. Much better than in JCR or in Confluence. It's for the content itself, much better for AI, also for changing things and all this. So, yeah, clear concept, but it's just on paper or paper in Confluence. So, I thought, getting all these resources internally to do that, that's a lot of time, a lot of resources. Let's do it the other way. I just start now with Cloud Code and see how far I get. So, yeah, it started as a POC, first with the rendering of the ASCII doc, not over Antora, but by the JS library. So, and to build the course how it should like. So, it was just really a side job besides giving training sometimes in exercises of participants. I went a bit on. And I had a clear vision of how it should look like. So, just plugging in feature after feature, trying out. So, I got, you use Cloud Code very fast, very far as a one-man show, as a side job. And, yeah, what I really love about it is this rapid feature delivery. You start with an ID, you get into this brainstorming, you let it create things, you guide it to the right place, and, poof, couple minutes later, you have a kind of feature. So, yeah, you can look at it, you get more ideas, you improve it, and so on. So, yeah, just a few things to show. Yeah, the course rendering implemented with progress bar and also the progress ticks of all these exercises and topics you can do. So, yeah, I think there's also internal, like, learning management of the learners' enrollment of courses or general statistics. How many complete courses, how many enrollments and so on. And then specific statistics of courses and of each course and the user interactions there. So, how many pages have been clicked, which topics have been finished and so on. So, yeah, it was like, how far do I get? How hard is it to add something like this, these statistics? And that's, yeah, super fast done. So, yeah. So, what makes it good at Magnolia with AI in this spec of, like, a project? I think one of the main things is context. Providing as much context as possible. I think you know it well when you do code with low code. Giving the architecture specs, giving as much as you can. I gave you the whole source code of Magnolia. All documentation, everything, I just threw it in, here you have it. Then, guardrails. I mean, you know it often, it's not doing what you want, so rules were growing. So, you can see here the skills and, I mean, rules and hooks. So, like, really hard guard, oh, that's 39 seconds, huh? So, hooks, rules, and from that came skills. So, but you steer it, you know it, you need to tell it, hey, that was the wrong way. So, now, the real workers are the skills. So, a lot of skills evolved from Jan Schulte's demos, from this working on this learning system, they evolved and evolved into a state we said, we want to publish them. So, yesterday, first information of publishing them. They are available for all of you. And, basically, that's, you can take a picture, that's the end, that's what I build upon. These skills are there for all of you. I think they're a big help with developing with Magnolia. Nice. Nice. Very cool. Christian, do you have any, like, suggestions about, like, where would be the right use case to start using the skills? Would it be a new project, or in the middle of a project, or any kind of context you can give there? I would say in any part of the project, when you use AI, because they teach your AI really, like, specific knowledge, how to build UI, how to use the exact, the React design system is a skill, it's 39 skills, it's really a big set of skills. So, I would say also for beginners and advanced people, same, yeah, for beginners it's nice, they can proceed fast, for people which know it well, it helps AI to guide. So, yeah, the next topic there is like contribution. So, if you want to contribute, write me, add skills. It's not for normal contribution open yet, but that's what we plan. If we see there's really the need of people who want to contribute, that would be cool. Yeah. Nice. Thanks so much, Christian.