Das strategische Gold

Transcript

00:00:00Welcome to Think Different, Think AI, the podcast by Mark and Jens.

00:00:07Two technology-loving minds who not only talk about artificial intelligence but live it.

00:00:14Here, you will find clear classifications, real practical insights, and a fresh perspective on what is possible.

00:00:20Understandable, critical, and always with a wink.

00:00:24Thought-provoking, amusing, and above all, engaging.

00:00:34So, remember, we had a guest in our still young podcast not too long ago

00:00:41and we invited him back because he had already discussed very interesting

00:00:46topics with us back then, and we thought, why not revisit this,

00:00:51because not much has changed in the AI world.

00:00:55That sounds strange, right?

00:00:56Everyone is always talking about how things are moving so fast or something else, and I start this podcast with the words that not much has changed.

00:01:02But in fact, either we were back then or especially you, Rene, were already so far-sighted.

00:01:07We talked back then about the topic of agents and how they actually interact with us humans.

00:01:15So how, in principle, such an IOC model was also mentioned back then, or you brought it up.

00:01:20So the intent of humans, the operations performed by the machine, and the control ultimately provided by humans again.

00:01:26If we now look a bit at the topic we want to discuss today and deepen it,

00:01:31particularly related to agents, it’s mainly the topic of skills.

00:01:34And I believe that is indeed a crucial part in order to actually perform this operations part well

00:01:39and not just throw everything into a round, these things need good skills.

00:01:42And I'm glad you are here. I'm glad my co-host Mark is with us.

00:01:47And I think this will be another good episode. We have a lot of topics, really packed, that we can discuss.

00:01:52A lot is happening in AI again, so let’s get started.

00:01:58Yes, first of all, many, many thanks for the invitation.

00:02:02Yes, you were also our first guest on the podcast.

00:02:06August, right?

00:02:07Oh God, that hasn't been so long ago.

00:02:08And when you just started, Jens, and after the Motus, not much has happened, I thought,

00:02:14hold on a second, even if the statements made back then in the podcast definitely, I would say,

00:02:20still hold quite well today, we'll certainly get to that later.

00:02:25A lot has actually happened.

00:02:28When we started, I was singing high praises for N8N here.

00:02:35We looked at models.

00:02:37So today, I don't even know, do you still use the models from back then?

00:02:40No, I don't think so, do you?

00:02:41No, Mark, when you think about it, when we started, we were both burdened by Nautchen.

00:02:46Back then, that was suddenly brand new, how it created connections in databases and how it quickly built content for you.

00:02:55Is today more in the background, right? Today you're taking agents, they do it anyway.

00:02:59Although, Notion is a nice keyword, my cousin, I also had him here in an episode and we talked about Notion.

00:03:05They are with Anthropic, the whole topic of Managed Agents, we discussed that in the episode that went online recently.

00:03:12And it's quite funny to see how Notion is trying to stay in the woods.

00:03:17But yes, this hot topic, look what's coming out of it, look what's happening.

00:03:21I mean, we talked about Prompt Engineers and so on back then.

00:03:24And today we're talking about, and I think that's a very nice segue, skills.

00:03:30I think that's really great, you know the saying, he who asks, leads.

00:03:35Who would like to explain to us what skills are?

00:03:37I'm always so cheeky and say, no, would you?

00:03:39Yes, I'll follow along.

00:03:40So, it’s really totally unprepared for all heights, but the way I explain it, yesterday was prompts and prompting is basically a

00:03:49elaborated text that is a bit adapted to the model, it somewhat guides the model, unfold yourself like that.

00:03:56A skill is that, but it also requires access to a lot of tools, and the access is, let's say, orchestrated and guided, and there’s also code involved.

00:04:06So, then, when needed, the entire skill also requires code to execute dedicated clear instructions.

00:04:13And the combination of everything, plus the ability to take persistent storage units along with it, is a skill.

00:04:21And that's why prompt engineering was yesterday, and skill engineering is tomorrow.

00:04:26And our, our three trial signs will be that soon the world will be full of skill engineers.

00:04:33with a flavor in purchasing or flavor in controls or flavor in logistics and so on, right?

00:04:39No, I would still like a little bit of flavor-logistics on top, delicious.

00:04:43Delicious.

00:04:44So, I actually find that it was also a bit ironic, the

00:04:47motto, Promethe-engineering is dead, it really isn't, right?

00:04:49You basically find all the rules again, you just pack them into large text files today,

00:04:54skills then have these so-called markdown files, where you write your, I’d say, work instructions

00:04:59and whether it’s the agent executing them or the human, both can

00:05:05use it. And as you said, the topic, I can also store program code in there

00:05:10and bring it to execution and thus introduce determinism into the system is just

00:05:15really cool because you can actually mix that in these skill files. I saw

00:05:19a skill the other day where you write your prompt and suddenly

00:05:21a bit of Python code comes up right in the middle of the prompt and then it just continues and you

00:05:25stand there and think damn it, how cool is that, right?

00:05:29How cool is that?

00:05:31And everything can also be generated automatically.

00:05:33That is actually the crazy thing.

00:05:36Already with the prompts, you know.

00:05:38You want a prompt, and you ask the AI to create a prompt so that the AI is better.

00:05:43That's almost like the singularity that Musk likes to talk about.

00:05:47That they AI, AI makes.

00:05:49I'm not at the point to say that we are there, we're not quite there yet.

00:05:52It has a certain flavor in that direction, doesn't it?

00:05:55Yes, the range of skills, what I find is that it does, and there's a market there

00:06:02especially right now.

00:06:03Back then we talked about models and now the skills are actually manifesting

00:06:07this removal of that discussion, because of course skills bring the nice aspect that they

00:06:12are model-agnostic.

00:06:13But I can simply create the skills somewhere, in a relatively simple format

00:06:17save them, and then make them available to any model afterwards.

00:06:20And that is, I believe, one of those incredible advantages, that you no longer look at the topic so single-layered

00:06:25like, wow, what can a model do completely, but rather

00:06:30go back a bit and say, which packages, which capabilities

00:06:34can I somehow encapsulate, without it being actual program code, but rather

00:06:38a description of what needs to happen, in which behavior it should show itself

00:06:42and that I can then, depending on which model is the best choice, also

00:06:47make that model available.

00:06:49Of course, this has a few other advantages, because we are no longer so restricted.

00:06:52We are not working on the edge of the model, but rather thinking about how

00:07:00I can orchestrate all these skills, which skills do I need, are skills

00:07:05from yesterday still the skills of tomorrow that I want to have.

00:07:08So, I am interested in the skill from yesterday.

00:07:10From an hour ago, when you say that.

00:07:14Yes, exactly.

00:07:15You know, I have your analogy. Do you remember back in the day with digital cameras when there was the so-called pixel race at the beginning?

00:07:22My camera was not 2-pixel, megapixel, yours can do 3, the other one can do 10.

00:07:27And at some point it became clear that more pixels don't fit on the chip, then it was about a different quality.

00:07:33I mean, we are still on that track, so there will be better models, so when world models come, there will be again some, but it will be so that we will also have a saturation,

00:07:41that we will be saturated and then exactly that, what you said, the system.

00:07:47So how do I utilize that in itself? How do I create impact?

00:07:50Becoming significantly more important and when? The speed is probably already quite soon that we will be in saturation.

00:07:58So what I have with all these skill topics, I admit, this thought is actually still quite new to me.

00:08:03You can feel free to laugh at me now if I am lagging behind.

00:08:07I stood recently in front of the question, ok, well, in skills you can run program code, fair point, understood.

00:08:13Actually, something like a skill and a Kaida underneath is an operating system where software can be deployed.

00:08:21I recently built a little skill, it's available in the Git repo.

00:08:25It burrows through what is inside, throws Kotlin, PHP, no idea what.

00:08:31You tell it, create a skill from that and then it considers in a sense,

00:08:35which orchestrators I need, so how do I break down the tasks that this program,

00:08:39that I found in Git, how do I break that down? And what it cannot convert into a skill

00:08:44it really runs as software in Python in your skill.

00:08:51And then I thought to myself, wow, you know? So this topic, how do we build software

00:08:57changes? How you run things changes? And how do you integrate, I’m saying now

00:09:03sorry, such an old, mothballed Githa project that suddenly gets a new life because you say, come on, migrate this to a skill and suddenly that thing is no longer alive, whatever, here Kobol, something else, well, I probably won’t find much in Gith about Kobol, but then suddenly in such a runtime, in Codex, in Opus, in, I don’t know, what.

00:09:22I always found that kind of, huh, I know it in principle, but I found that so amazing, with suddenly so many possibilities.

00:09:30You know, I recently, I follow a lot of news channels and so on and one

00:09:35unfortunately I can’t remember the name, it was a very fresh one, I believe Z51

00:09:40or something like that.

00:09:41So the name isn’t quite correct, but somewhere close, an open source

00:09:44or also a weight model, that’s just out now, they let that thing run

00:09:49and let it run autonomously for eight hours to create a Linux clone, so just completely

00:09:55fully operational with over 50 apps. Even a WhatsApp or Telegram messenger component with

00:10:03mail and everything fully operational. And now think about it, you make this horde of

00:10:08agents that create software, running through all GitHub repos. Just think about

00:10:13what will come out of that, right? If you say, okay, take the best and build a bunch

00:10:18of apps that can do this and that. Yes, yes, that’s wild and it feels a bit like,

00:10:23I had two things from Kaperfee. Kaperfee released this Wiki topic again last week

00:10:29about how it can essentially be scientific. That’s why many of

00:10:33our posts revolve around this topic. That’s another subject. It applies once

00:10:39to the topic and in the background again, that we have some kind of scientific

00:10:42that we want to use. So also outsourced from the model, so that it can also be used agnostically.

00:10:45And aside from all these bottles somehow getting complicated

00:10:50rack structures, something like that, that one thinks about in between. But what is the other course from

00:10:54him, I found even more exciting and it fits what you both are saying right now. Because the

00:10:57capabilities are so high, he also assumes that one doesn’t actually fork a Cub base in

00:11:03Github anymore, but that it’s actually only important to fork the idea markdown file,

00:11:10you know, that we’re coming into such a world where it gets much brighter

00:11:15and much more sensible that we only exchange the ideas in a token optimized way and not swap the whole

00:11:22legacy code, if you will, and then say, that was the actual idea, that’s the actual output I want to regulate.

00:11:28So if we now take, I thought this morning overnight, from such a designer-job-to-be-done perspective

00:11:34looking at it, you used to always say, if someone wants to drill a hole in the wall, yes, then they need a drill.

00:11:39And that’s basically the job I want to do, like the hardware store, us as Brosch or other manufacturers of drills.

00:11:45Nowadays, if you look at it from that perspective again, if I say the job-to-be-done, that’s basically what has to be done.

00:11:53And I can actually do it without having a pre-made piece of software.

00:11:57I also envisioned it that way at some point, but without having any pre-made piece of software at any point in the ecosystem, whether at my place, at the partner’s, at the reseller’s, wherever,

00:12:09I can actually do without it, but I just have to state in the concept of the machine that this is the job I want to do.

00:12:15Do the rest.

00:12:17Do the rest. Or suggest something to me, right?

00:12:19So it’s, and I also believe that the key is this topic that he also chose for this podcast, the skills.

00:12:26I believe, if you look back, there used to be, well, in the Internet there were the web apps, so the big thing, that you know now

00:12:33if apps became available it was the iPhone platform and so you could also make apps available to others.

00:12:39Then of course, language models can now skip some things but this idea of language models is also going into skills in a similar way. I’m not saying apps, but rather also an epitomization.

00:12:49In the general AI world, so something happens that exactly describes what is happening.

00:12:55It’s again something that we as humans might need to make it easier for ourselves.

00:13:00Maybe not wanting to accept that there is a model that could do everything.

00:13:04So, because on the other hand, let’s say if we go into the future and the models are getting larger and better.

00:13:10You just mentioned the topic of world models, and I somehow believe that even if we were to take a normal model and compare it,

00:13:17If we were to try to fit our brain into such a model, I think we would still have 100 trillion parameters aside.

00:13:24And I believe that even the best models in the world are still miles away from that.

00:13:28Nature is still clearly superior to what we have built so far.

00:13:32But nonetheless, we are heading into a future where the

00:13:37Models always being able to do better, and then you wonder, do I need such a

00:13:40skill description?

00:13:41Because actually, these models can do everything anyway.

00:13:44And now I have a point when I let it go.

00:13:46Because you might know, I love writing here and there,

00:13:49especially with Mühlscher.

00:13:50I have a small issue.

00:13:51It's currently related to the topic of leadership and Agent DKI, because there is a paradox.

00:13:56Because I am deeply convinced and I believe I have solid evidence that jobs will only change.

00:14:02We will do a lot, because people will still work for me, so we will do a lot, but we will work differently.

00:14:07And what is that now? There is a paradox that says, the more you automate business processes,

00:14:14the more leadership you need, the more leadership.

00:14:17That's what you also talked about last time. You mentioned Intent, Operate, and Check.

00:14:21This intent aspect, this strategic leadership will become increasingly important, because the skill-overloaded agent corpus must somehow convey the impact and the results.

00:14:34In order to work with it, you always have to ask yourself how to lead this amount of agents so that something comes out of it.

00:14:42And this feedback loop is getting faster and shorter.

00:14:47This means, the people who are currently in the process, by entering something on the table, clicking something on the monitor at that point,

00:14:55at the end then dominate the output of the process, they will stand by and observe what comes out and then intervene very quickly,

00:15:02oh, dear agent, I need to give you a different strategy so that the outcome is what I want.

00:15:06That means more leadership and less operational work, more guidance.

00:15:10That means, and this is very good news for me, because you also addressed the people.

00:15:15They do not need to master the skills, but they need to focus much more on leadership in terms of how can I manage a corpus of software agents so that something comes out.

00:15:27Yes, that brings me back to a topic I had already mentioned back then, and for this we need, of course, even more of this level of trust.

00:15:36So then reflected in some form of user interface, that we can understand what is happening.

00:15:42because if I now let an agent talk to another, if I as a human need to intervene somewhere

00:15:48and it always becomes complex, thus the tasks become more diverse because much more tasks are

00:15:53completed in an even shorter time, I naturally need, so that's why I am still

00:15:57of the opinion that it will be an incredible challenge, both for the agents

00:16:01to build among themselves as well as for us humans involved, to build an interface that

00:16:06truly gives us the trust that the right things are happening in the delegation,

00:16:10that I have made?

00:16:12I also find the topic of delegation.

00:16:14One can say, well, as a leader one is used to delegating.

00:16:18When you now, then I wouldn't say we have everything better under control,

00:16:21than perhaps your employees themselves,

00:16:23but the ability to let go of things,

00:16:27to accept that the machine does not replace me now,

00:16:32but that a machine can do something,

00:16:34up to a certain quality level.

00:16:36And then I look at it later,

00:16:38What I might still release somehow afterwards and then it continues and I don’t hold back with the motto.

00:16:44Oh look, what the AI built there, that’s nonsense, I knew that right away.

00:16:48I recently had a case where a colleague looked at something, a skill, and built a little bit,

00:16:52he somehow accomplished weeks of work in hours and afterwards the feedback was,

00:16:56oh look, this is that and that is still wrong.

00:16:59Where I then said, yeah look, you read it, you corrected it,

00:17:03after two or three hours you were also done again.

00:17:06That means, if I take your two or three hours, the two or three hours from Skill,

00:17:10then you're somehow at six or seven hours straight versus a whole week or two

00:17:15weeks.

00:17:16That's a good average and be glad that the AI, that you're not just,

00:17:22I mean, I'm now the one who just ticks things off, like the flightless on the radar

00:17:26sitting and waiting for something to beep loudly, because some planes might otherwise

00:17:31be on a collision course or something, which he can just keep an eye on

00:17:35and this constant, you already think, in a state of alert, he also has to take regular breaks,

00:17:40otherwise they would lose their concentration. I heard somewhere that such things

00:17:45just like autonomous driving, it is safer to take the steering wheel out of the autonomous car,

00:17:51than to trust that the human will intervene when the autonomously driving car

00:17:57makes a mistake, because you simply find yourself in the absence of that

00:18:02you just drift away and the other thing you mentioned earlier, then you might later

00:18:07have no idea, 3, 4, 5 agents, that's just the other side of the coin, that you

00:18:11might be giving things over to the machine, placing it in its hands and then at some point completely

00:18:16you lose, depending on which interface we provide to people, depending on how we handle it,

00:18:21that later you say, okay, I have no idea here, counting windows open,

00:18:25whether it's a bathroom or Opus, whatever is lurking around, and then you're standing there, where

00:18:30was I with what? And I have to leave the right one for a moment, you come back after an hour, what's going on now, what is the current state of affairs?

00:18:37This also calls for a new kind of interface. How do I deal with it? How do I manage time? We really have some exciting questions ahead of us.

00:18:48And also the skillification will change the interface, the GUI will definitely change again, that's clear.

00:18:56Yes, so that's something, I'm actually curious, as I said, how that runs, and just this realization I was talking about with people, you give them something in hand, they say, come here, Kai, right? And then, well, no, did you misunderstand? I get something returned from her, I work on it again and then it falls back to us asking, okay, can we maybe get a bit more out of it because we adjust the skill again, yeah? What interests me, the skill from an hour ago, was just said, I also have something like that,

00:19:26trying to figure out through synthetic test cases how you can improve the skill overnight and hoping it works a little better the next day.

00:19:34It’s a really broad field, that, yeah, we first have to see how to position ourselves and yeah, how it all develops and the good models just come from it by themselves.

00:19:48Definitely and regarding positioning. It’s actually a bit interesting,

00:19:53now, when I think about the personal experiences we have and those that

00:19:58people outside, I believe, have on such an organizational level. So I believe,

00:20:01the ex-Twitter founder, I think his name is Jack Dorsey, he also

00:20:09published an article one or two weeks ago where he said, we actually need

00:20:14much, much flatter organizations in the future. And do not start translating your

00:20:20current org charts into agentic structures. That’s exactly one thing. So,

00:20:28I believe that’s a good impulse, you have to take that away because I think,

00:20:33there’s something in it that needs to be picked up when you now dedicate yourself to the topic and say,

00:20:38yeah well, how could my organization in the future, enhanced with skills that I

00:20:43somewhere linger, actually look, what happens, what will change

00:20:46and I believe that’s an exciting question. So this old

00:20:51pyramid is dead. You need to write a chapter in my book. So actually the old

00:20:58information brokers, i.e., leadership layers that define themselves by how

00:21:03carry a drift carrier in the message and know that A and B

00:21:06need to be together in the meeting for something to come out of this. You need them still,

00:21:11So it doesn't mean that you're doing a bad job because we're managing a complexity,

00:21:16music complexity, but that will become less and less important and it will be

00:21:21much more important to lead content-wise.

00:21:22This will naturally flatten hierarchies.

00:21:25My steak is always, hierarchies are actually always a compromise.

00:21:31Actually, I don't want to have any hierarchies at all; everything needs to be

00:21:34flat, only then do I have anarchy.

00:21:35So I have to somehow try to lead a large number of people efficiently in an organization so that the maximum productive output is generated.

00:21:45So I have to create such branching teams. And as soon as you make two teams, you have two teams that build a wall and bridge.

00:21:55That happens automatically, and at some point, when you really live it out, we in our company are fortunately not doing that,

00:22:01Companies, I've seen them, they have really stretched that. You need a visa for that,

00:22:05when you want to go from HR to the marketing department. So that's actually quite far apart.

00:22:12And this pyramidal setup is indeed in a world where we work

00:22:19can be divided into agentic and, let's say, strategic work, less important.

00:22:25Much less important, and you can work much more hierarchically free here. And we know that.

00:22:30Hierarchy fly is indeed the essential core of agile methods. I could remember earlier in

00:22:35other companies, when I first tried to teach management agile concepts. People said,

00:22:39agile, that's quick, then you do it faster. Yeah, no, that's not the

00:22:46idea, rather the idea is actually to work hierarchically free. No matter who has

00:22:52what rank on your shoulder, the people who know stand around a problem

00:22:57around and solve a problem to generate the best

00:23:04solution for the user groups, the customers, who have the problem. Hierarchy-free. Yes, hierarchy-free and knowledge power free. I knew

00:23:10the term knowledge power as well. It was sometimes expressed that

00:23:13certain, well, certain things should be known. Well, of course, there are still always some

00:23:17topics that have relevance, that not everyone always knows about,

00:23:20but I believe in such an organization of the future, it is actually unacceptable

00:23:26if not all knowledge is readily available and especially if knowledge

00:23:31is not just the stored data format or the file, but also the skills

00:23:38needed to simply implement certain topics. And I believe we also need to

00:23:41address that. I believe that the organization of the future needs, I think, both alongside

00:23:45the accessible knowledge in a, for example, in Markdown format or in whatever format

00:23:50so that the machine can read it well. MCP servers that can access certain units

00:23:54simply, in my opinion, also need a kind of skill database and workflow

00:24:00that is available to the entire company. Because what you are just describing. So that,

00:24:05what the energy system needs is that I basically have all the skills

00:24:08to implement things. Because as soon as I don’t have all the skills to implement a certain

00:24:12topic, the energy system fails. Because then I have to wait for someone else

00:24:16to implement it for me. And I believe we can also go there. That we

00:24:19say that in the future we can query the necessary skill to already take

00:24:25the next step. That doesn’t mean that we have to take the step entirely alone,

00:24:28but yes, we still need the other person who will maintain this skill group

00:24:31we can rely on. But I will become more flexible, simply,

00:24:34I will become significantly more flexible. We need this as organizations or

00:24:37Organizations will need some sort of skill in the future, skill, we just talked about that.

00:24:41I still remember an episode with a,

00:24:46with two people from a kitchen appliance manufacturer,

00:24:50with whom we had the discussion, Jens, you were just,

00:24:54I think at Carnival, whatever, we had the topic,

00:24:58that organizations will certainly change in the future also because

00:25:02boundaries will disappear that were created,

00:25:05because in the past people were simply given a topic as a task.

00:25:10And this task, there were, what I say, you had 10 people doing web,

00:25:1310 people doing apps, 10 people doing backend, 10 people doing

00:25:17data management, 10 people doing service management.

00:25:19And now you find yourself with an agent or with tools that make these boundaries

00:25:24disappear because you are now working on a solution and not just on a

00:25:29very small part.

00:25:30You do a crash course alone, you do a feedback form alone,

00:25:36you click on deploy and the database learns your field, web and app get

00:25:41the stuff and if you say tomorrow, I’d like a new layout, then it has a new

00:25:44layout. Certainly, there are still people who say, okay, just a moment, this is the

00:25:49font that should be used and this is the color that should be used and we have

00:25:52perhaps a few rules we want to set, guardrails, how development occurs and

00:25:56what needs to be considered, please also think of testing, nobody should forget the tests and

00:25:59stuff like that, but still team structures will change because I also believe,

00:26:05when I see it alone, what people are capable of and I experience

00:26:10the sweet taste of honey, who previously would jump into my face with a

00:26:15naked ass if I had told them, your expertise is important and correct, but you leave your

00:26:21beaten path, you open yourself to new technologies,

00:26:26you are now front and back and you have seen it? They considered me crazy.

00:26:31And now that’s normal for you. Now they come to the office with joy,

00:26:34because all of a sudden boundaries disappear. You talk about solutions and he says please also,

00:26:39I know you’re just getting into a talking spree. We have guests, yes.

00:26:41I mean, hello, I also had to give the guest the floor again.

00:26:43But I always remember a story my teacher once

00:26:49told me in secondary school.

00:26:51She said that a person who contributes to a solution from start to finish,

00:26:57identifies better with the solution, so if you stand at the assembly line and just screw in a screw.

00:27:02Well, we have robots for screwing in screws today, fair point.

00:27:05But you identify with it more.

00:27:07But now people get the tools to do something that a customer,

00:27:14another user, can make something out of, and with that you identify and that

00:27:20brings joy.

00:27:21At least that’s my experience.

00:27:22And I believe that is also meaningful.

00:27:25So yesterday we trimmed organizations according to good skills and abilities, but

00:27:31based on a terrorist process world.

00:27:33Yes.

00:27:34A specialist for this, then comes a specialist for that. We do handovers and have programmed an IT interface.

00:27:41SAP did something integrative back then, but it happens one after the other, and that's how this specialization was created.

00:27:48Now, of course, domain knowledge is becoming, how shall I say it, a little less important.

00:27:55It is still important, but its significance will diminish somewhat because, somehow, when you go up to level 5, where there are 100 levels,

00:28:03anyone can tackle any topic now. Anyone who can’t godengar is godengar.

00:28:08Of course, not on a huge scale, oh, but who knows, we have a large Erdmann, but

00:28:13everyone can do piece-controlling, everyone can basically buy in, everyone can basically

00:28:17make the first competence filters for HR developments, basically. So, domain knowledge will

00:28:22be less important. This means the way we structure our teams needs to

00:28:25be adjusted. But they have to be. That is exactly what you said,

00:28:29you hit the nail on the head, because in the future we will cut much more based on

00:28:34giving people the feeling that it’s their thing, their baby, that

00:28:38is their motivation, that is their output. And that has always been my credo. All of this,

00:28:44what we collect with this wonderful technology, is still just a screwdriver. That is

00:28:48a tool. I use it in the end to either get the screw into the wall

00:28:52or to borrow the hole in the wall. That’s a tool. And the one who

00:28:56drilled or screwed is then the person who is proud of the output.

00:29:01And that’s how we need to approach it.

00:29:04Exactly, that’s how it should be, and if we directly should understand that the person actually

00:29:08wanted to hang a picture on the wall.

00:29:09That is the actual output.

00:29:10That’s also the story I started earlier, to see

00:29:13and I believe that is really important.

00:29:14I think this is exactly the topic, we need to focus more and confess

00:29:18that more on what we actually do out there as a business.

00:29:21So we as a company offer something that enables people out there at this moment,

00:29:27to perhaps hang the picture up for someone.

00:29:28And I believe that is exactly the topic, to focus more on the real impact.

00:29:32and less to work through layers to find someone who can catch the ball somewhere.

00:29:37.

00:29:38And I believe AI can really help there.

00:29:41I think, Mark, because there’s something positive to ask.

00:29:44I actually have two critical topics there.

00:29:47Yes, I actually wanted to pick something up.

00:29:49Before drifting too far forward, I wanted to briefly pick up the ball.

00:29:52We already talked about it; when teams are formed, bridges need to be built and such things.

00:29:57And while we were talking, a sentence came to my mind that I might not be able to

00:30:00articulate very smoothly, but in the past, you basically shared your expertise

00:30:05by specializing more and more in topics.

00:30:10You have acquired knowledge, you've applied it, you're basically, in quotes,

00:30:15you've spent a long time with it and then people called you,

00:30:18people reached out to you. Your opinion was important, not just because perhaps people

00:30:22like you as a person, but simply because you have, as someone once said, yes, you've

00:30:28seen horses throw up before, you've experienced all sorts of things

00:30:32and have an opinion and expertise based on all this experience that helps.

00:30:37now continue in the project. And if a project is in trouble, it doesn't really matter

00:30:41in which company, there was always someone you could say, oh yeah, has he then,

00:30:45Yes, has someone taken a look at it, because if someone knows, then they'll just call them.

00:30:50And that is becoming more democratic now.

00:30:53Now, it's not really the one who has the deepest knowledge

00:30:58in ABC who is automatically the most sought after, but rather the one who might be capable

00:31:04of creatively bringing together A and B, applying things, opening doors, connecting the right

00:31:10people with each other.

00:31:11So suddenly, completely different skills are emerging and adapting. I mean, who likes it when their tool, if you speak of a screwdriver, yes, today you have a screwdriver in your hand, that's a tool. Tomorrow it's a hammer, the day after it's a drill.

00:31:24Next week it's a jackhammer or something else, and then suddenly it’s a spoon. Yes, because the thing is striking and significant, new products are coming out and you have to keep your way of dealing with it always a little up to date.

00:31:37much more than before. And I believe these are some traits that we will

00:31:42increasingly encounter. I wanted to mention that. Thanks Jens for

00:31:46noticing this, so thank you for allowing me to say it again.

00:31:49Of course. You are a team. What do you see?

00:31:53Yes, definitely. Like a well-oiled machine. But only because I am always

00:31:58getting updated skills... No, no, I always get my updated skills

00:32:02in D, I throw that into my LL-M, then I know exactly how to

00:32:05And fortunately, I have to say, I made a thing on Github in my

00:32:11press information, so that all my Github pages always have that in the press information.

00:32:14And I took your Identity-MD on my Impressum’s Gill, so that I

00:32:20can always tell my AI how I think, what I expect, how I speak, so that I

00:32:26can address that and I have now also made it public, which means anyone

00:32:30could basically try out how I react, in the brand as well.

00:32:34Yes, I have my own identity in Debate, I don't even know how much I share it, but I've adjusted it a bit.

00:32:39But yes, another topic, we can also sometimes talk about that, because the issue of passing identities is also a totally important topic when I interact with an LMM.

00:32:48And especially, if we're not always working with the same one, then such identity files together with the context that I pass to the AI, whether I do it through a small database,

00:32:58or also through cat files or through prompts.

00:33:02I believe that will somewhat be the handshake of the future,

00:33:05the information handshake of the future.

00:33:08Because that’s already a different output.

00:33:10Whether it's Mark, René, you or I somehow talking to an AI

00:33:14or speaking with our trusted AIs,

00:33:16there will always be different results,

00:33:18because our identity plays a role in that.

00:33:20And I believe that's also an important core

00:33:22for the discussion for organizations that we have.

00:33:24The identity of us, that shapes an organization too.

00:33:28So, if we only start copying skills, we are now starting to describe our skills, what we do in our activities,

00:33:36then that is of course also broken down into the individual parts, as that small assembly line work,

00:33:42as described, this one small step that I take on the assembly line at that moment.

00:33:47This would probably also be reflected in our work maybe one or another time, when we describe our skills.

00:33:51But of course, additionally, it is the case with all humans that this identity comes into play.

00:33:56comes into play. And this identity changes a bit, also within organizations, because one

00:34:00has to still show certain behaviors all the time to figure out the best pathway out

00:34:06because nothing is perfect. It will never be the streamlined super AI organization.

00:34:11give in this sense. There will always be an interplay between an AI that fails, whose

00:34:15tokens have run out, where the server is shut down because

00:34:20someone just pulled the plug or something else happened. So this will also

00:34:23continue to happen in the future and accordingly, there is a need for identities in between,

00:34:29which we humans are also quite capable of, who can still react to such a system

00:34:33.

00:34:34We can also, if people are unsure, make a nice analogy.

00:34:38We have laws in the world, every legal state has laws, and yet we still have

00:34:42judges.

00:34:43Because what is between the laws must be interpreted and somehow weighed

00:34:47.

00:34:48And that can also be changed in our world just like that.

00:34:51The world is not logical at all; it's actually the opposite of logical, which unfortunately we see every day. But we need a middle ground in the sequences that happen in the processes to give a bit of an assessment, like, yes, oh, I think I'll just let this run, or no, I'd better stop this now because, for example, AI has no intention, AI wants nothing, so we first have to bring in the intention so that I want something.

00:35:18And then AI also has no risk assessment, so it can't say whether this now,

00:35:23I can maybe say a little bit like, okay, this could trigger a nuclear war, so I'm not going to do that. On the

00:35:28level, it might probably work by now, but on a, let's say, more normal level that concerns us every day, then someone says, ]} ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ``` ```

00:35:35hm, maybe it's a bit too risky to let people stand ten meters closer to this high wool facility now and so,

00:35:42something like that, so this overall assessment can only be done, it can actually only be done by humans,

00:35:48and I believe always only by humans. That will be it, so the goal is also to automate this,

00:35:53but the operational side, the one that generates costs and eats up time, that's what we want to automate.

00:36:01But I would also say on some topics, I believe René, those topics could also be automated.

00:36:07So now, I'm going to take a critical look at China.

00:36:11What we are of course seeing in the factories on site is that we have a lot of workers who are basically performing repetitive tasks, and they are now being equipped with video cameras that record their manual work.

00:36:28Entire factories are completely fitted with video cameras, and the same small spot is being recorded, which you could also imagine on a larger scale.

00:36:36So not just sitting at a conveyor belt but also rolling through the offices,

00:36:41We can also document that I would do the entire roof if I went through the office,

00:36:44It was something to deduce what is needed, what does one want,

00:36:50But what does such a robot actually need, and another topic that I will just throw in as criticism,

00:36:56That was something I also saw, where I was also surprised, where fortunately quickly,

00:37:00a folding was also loaded.

00:37:02There are certain Chinese companies that are already forcing their employees to basically,

00:37:06really document their skills through their activities, in order to then say, yes, this

00:37:11I can automate later with an AI what this person does.

00:37:16Interestingly, a service was created directly, which was called, let me check,

00:37:19Anti-Distillation Skill, that's great, there's a Chinese woman,

00:37:24She wrote it directly.

00:37:27I was just about to say, one in three can probably pronounce it.

00:37:34But the skill helps you; it's like a cleaning layer.

00:37:39It pulls out all the really interesting things you wrote in your skill,

00:37:45pulls them out and leaves you just the bare essentials to tweak a bit here,

00:37:49do a bit there.

00:37:51And you can hand that over as management. You can be sure that if

00:37:54management uses the MD skill, your job will still be needed.

00:37:59Those are of course also. So the care side of the coin. We have to somehow

00:38:03talk openly about that these things happen.

00:38:05Yes, well, I lived in China for a few years and I definitely have, let's say,

00:38:10my own perspective. I actually find the order of freedom super,

00:38:15and I don’t think total surveillance is great, that's clear. But within that system,

00:38:19the railway itself. So it's not that bad at all. It's actually very efficient when you

00:38:27mention the railway, and it’s no coincidence that currently it's the

00:38:31global winner. First of all, they don’t make idiomatic mistakes; they function,

00:38:35they go in very slowly and now, if you look at the last five-year plan,

00:38:41which he has looked at with twelve points, there are six points where they want to become world leaders

00:38:45and six others where they at least want to achieve global breakthroughs, and in this, there's quantum

00:38:52involved, breakthroughs in that, there's also Neuralink in the brain and so on. And that,

00:38:57their concentrated and focused use of technology is also due to the fact

00:39:05that this nation will lead globally in this technology and that’s why it’s also

00:39:10more a business model, to try something out so that the others around the world,

00:39:14who then have to buy the technology to reach productivity, thus become

00:39:20customers. It’s quite simple. Yes, so my personal freedom is very

00:39:27important to me. I would never accept that. It’s probably hardly possible in a western

00:39:31society anyway. But we must not be so parochial and

00:39:37disregard that China has had a culture for thousands of years. That

00:39:42doesn’t mean that people there feel differently or that freedom is just valued differently.

00:39:49It is simply a different good, as a different good is.

00:39:53One is not better than the other; it’s a different valuation, a different system.

00:39:58We don’t have to like it, you don’t have to like us.

00:40:01I always have something that relates to a completely different example.

00:40:04When I lived there, it was still very common to spit on the street in China.

00:40:10So we find that not only spitting, but I won't demonstrate it, but you can imagine, really from the depths, because behind it was the idea of cleansing, letting the unclean out of the body.

00:40:24At the same time, all the Chinese people I spoke to looked at me with disgust when I sneezed, when I went 'Hatshi'.

00:40:33They looked at me with disgust and said, oh, what is that? Just as disgusted as I was.

00:40:37One is not better than the other, but there are simply different perspectives in a cultural context, and you have to learn that there are differences and deal with the differences without judging.

00:40:50But now back to technology.

00:40:52There is this high affinity for technology.

00:40:55I can only recommend to everyone to find some x-random video that is not older than three or four months on YouTube about Shen-Sensee.

00:41:03Man, and I have experienced parts of what is being shown live, so the technology, to already introduce skill-based work in companies now.

00:41:15Training sessions with a camera on your head so I can run alongside and that later a robotics will even, not only on the robotics and the world leaders, world market leaders, they will look at it as it is today already.

00:41:26This is part of the overall plan of this industry or this nation, let's say.

00:41:33And I dare say they will be very successful with their plan.

00:41:37So, when I think about when we made our episode from Open Claw,

00:41:42we saw in preparation how there, let's say,

00:41:45installation parties or installation services, I don’t know what else, take place.

00:41:49And when you walk through there, no matter where, and talk about Open Claw,

00:41:55you first hear, yes, help. So, no, I don’t know. More the rejection.

00:42:03Or difficult to convince, you get hit back, while there are people sitting

00:42:09with whatever drinks and notebooks and installing themselves

00:42:14the stuff, while we, so to speak, not us as a company or us as

00:42:19individuals, but more as a society, are thinking, oh, damn it and

00:42:24couldn't we regulate something or anything? Yes. And it somehow goes

00:42:29fire at will. Maybe somewhere in between lies the truth, where one

00:42:32might meet. Yes, I wanted to say, there is no truth.

00:42:35Both have their validity. We have a different culture and we want

00:42:38to understand things. It's said that the Asian way is

00:42:43more about the journey than the destination. For us, the destination is the goal. And

00:42:48both have their validity. One is not better than the other.

00:42:51Right now, it seems one is faster, okay? So we can also actually see how much in particular Europe is currently not functioning. That is also all true.

00:43:01But over the entire, let's say, the next ten years, it will also readjust. So every culture has its advantages at the moment.

00:43:09But behind it, in China, there is definitely a large business model. And Open Cloud Installation is now a service.

00:43:16Service. You go like to a kiosk, bring your device with you, and an hour later

00:43:19everything is installed by someone. That brings us back to skills. That would have

00:43:26I could have used it in the last few weeks because I've spent some nights here

00:43:30with one, I do it very, very carefully and very safely and I want to as well

00:43:33learn for myself again and what you do there. And I must also say, honestly

00:43:37also this process that I have now gone through again, just with this Orpenclaw installation,

00:43:42also with the topics about the Mark and which I then also exchange with Spalinkt in us,

00:43:46how to build a memory like that, how to set up a Bolt, how to essentially

00:43:50establish the handshake between agents and such. I believe it was necessary that

00:43:54I went this distance to feel it for myself again. And

00:43:59I think that's also kind of a thing, I have to say.

00:44:00Yes, it's simply valuable when you deal with technology that, to be honest,

00:44:07is also so easily accessible, is preoccupying. So, to put it bluntly, I have

00:44:11done. I basically just did this installation and keep asking

00:44:14all the time, wouldn’t another AI be different from what I’m actually doing? And let

00:44:17me explain that too. Forget it sometimes, I really don't have to. So as always

00:44:20ask again when I no longer remember the next terminal command.

00:44:22But it's possible, you know? So I’m doing things, I’ve also

00:44:26activated my GitHub, which had been dormant for five years,

00:44:30again. Careful, I put everything from back then

00:44:32on private, so it’s not so embarrassing, because they are rather old

00:44:34honestly. But despite next, so the call again,

00:44:39so get involved with these things. It’s something different when you experience it,

00:44:43because you come up with new ideas, you understand what it means. And that's why I come across topics,

00:44:49which are partly not really addressed yet. Who keeps the

00:44:55context alive in principle? So this contextual information that is necessary, which for example in

00:45:00a normal organization is maintained through meetings, through information exchange, that I am

00:45:05trained in my skills through a university or through a career path,

00:45:11that I know what norms, what security checks or any

00:45:15other things need to be done when I want to provide something. Who keeps all of that

00:45:19running in this agentic world in the future? That's going to be a totally exciting question. It is

00:45:23never really answerable, honestly. I see that right now. When I take on these things,

00:45:27these little errors keep popping up, whether they are just small,

00:45:31temporary errors. If one AI thinks that the other has already done something,

00:45:35but hasn’t finished yet. If a model on my local machine is actually supposed to

00:45:39do something, but times out because it hasn’t warmed up yet and the other

00:45:43then just going back and doing something, even if there is already confidence

00:45:48Gores or something different in the light side. So that is complex, I'm really glad,

00:45:53that I managed to push through a few night shifts over the last months and weekend shifts

00:45:58in order to get through it, because it helped me a lot to understand these things better

00:46:02. And I believe that is still valuable in this world we are in

00:46:05and engaging with these topics.

00:46:07And this is, if I support a call to action, for everyone who is currently listening,

00:46:12so not everyone has to tear themselves apart with open-cloid. That’s perhaps, only people do that

00:46:15who really have a bit of knowledge, because there’s always a security risk involved

00:46:20and you simply have to know what you are doing. But for example, to engage for

00:46:24a weekend with a topic, with cloud code, and to deal with it

00:46:29to ask or also on the i-codex or or or google i studio and maybe write a

00:46:34piece of software myself I am convinced that this is today a

00:46:38live skill just like we all know how to change a tire it can no longer

00:46:43be done but we understand it in principle this is a live skill that one

00:46:47understands the mechanics of so that it doesn’t look like one of those

00:46:51black box and I have absolutely no idea and one can

00:46:55program a small game, it doesn't have to be big, something

00:46:58Some kind of hangman game, some word guessing game, it's probably a prompt with five words and then it happens automatically anyway, but at least you can see how it works.

00:47:08So, dealing with it is really important.

00:47:11Totally, totally.

00:47:12I actually, because you mentioned the game, in my youth I played a game called Bubble Bobble.

00:47:18Back then, you controlled these little dragons that shot balloons, not soap bubbles.

00:47:24I loved that back then, but somehow you can't really find it anywhere anymore, and I think it was Gemini that built it for me again.

00:47:31To which I said, look at this and that, and here are a few websites and a few videos, check it out and do it.

00:47:37It was quite funny, and while you were talking, and you brought up the tire change example, I thought of another example.

00:47:43I mean, now we've all been betrayed, we won't reveal any details, but there may have been the odd hiccup in the IT world or seen one or the other thing.

00:47:51When I started with IT, so in my internship, there were punch cards, and later I worked with DOS Terminal, and for me, word processing didn't even have

00:48:01What you see is what you get, yeah, but you've worked a lot with text, like.

00:48:05Then eventually Windows came and the graphical user interface is like blah blah blah and that's quite nice.

00:48:09And people looked at me as if I came from another planet when I was still using Windows back then; those who know me know it must have been a long time ago.

00:48:18I then said here CMD, I think that was it back then with Windows, so you get to the terminal, i.e., DOS, and they were like, what's that black window? Where's the color? What are you doing there? And you're thinking, hey, what do you want from me? This is DOS, so you can enter something here, create, move, and delete directories, and I don't know what.

00:48:39and it was independent of the fact that I haven't actively used Windows since Windows 19

00:48:44I'm using. Is it still knowledge when you know how it is

00:48:48terminal and what is it based on? Definitely something you learn, oh I see'

00:48:53that already, shaking my head for everyone who is just listening. Is it definitely

00:48:58something that has accompanied me on my learning journey and today

00:49:01is still valuable knowledge for me to understand how the

00:49:05whole thing is connected and I believe that takes away this whole

00:49:08AI journey, when models, methodologies change every two, three, four, eight, twelve weeks.

00:49:13Whatever the case may be. It doesn't take away from us because we are on this journey, yes, because we

00:49:18walk with an open heart and accompany this journey and do not shy away from it.

00:49:22close our eyes and say it will go away and no idea, what and us

00:49:26let ourselves be surprised, but we are going on this journey. Jens, you shook your head.

00:49:31Yes, I just reacted, all good. I am, yes, I shook my head, everything,

00:49:34I am, let's say, up to a certain degree.

00:49:37Well, I actually think it's also a graphical,

00:49:40where you can use such an area well, because not everyone has to think their way

00:49:43into everything all the time, and I want to do something similar to what René just said.

00:49:46Not everyone needs to set up an Open Cloud installation outside,

00:49:48to engage with the topic of AI.

00:49:50That simply doesn't need to be the case, and similarly, not everyone has to do two Open Cloud installations.

00:49:54Exactly, it doesn't have to be any of that.

00:49:56We don't have to move around in a terminal environment anymore.

00:49:59Yes, the reason I do it now is simply because

00:50:02it is just easier to work with the Cloud Code Agent.

00:50:05also working as a terminal window, to then copy things over from one terminal window as another

00:50:08and so on. Yes, but that shouldn't be the end of the line,

00:50:12because we need to keep moving forward, ensuring that it's accessible for the

00:50:16rest as well. And I think the only point I would emphasize is,

00:50:21that fundamentally, it's understood what you're saying, how can such a thing work? What is

00:50:27perhaps a folder structure, even if I don’t need to know anymore on the iPhone,

00:50:31where the pictures are stored in which folder.

00:50:33It's still okay for me to know that, just as it was okay when I started

00:50:38back then, when it was said, yes, one didn’t work with a WYSIWYG editor, but

00:50:41rather occasionally with BabyEdit to describe the code oneself or

00:50:45something else.

00:50:46In the end, maybe with a Visibik editor, or back then already against or something like that.

00:50:49To fundamentally understand what is happening, I'm with you, because I believe René

00:50:53has already reacted, as Alte, as Thila, yes, and because we also still

00:50:57say, no, you all have to work in the terminal.

00:51:00Yes, exactly. But I want to make a point. Why? So, it's not so nerdy at all.

00:51:06Mark and I, we are of course the mega nerds, sure. I’m also hitting myself the whole time.

00:51:10But I believe, for the world out there, for the people, it's important to remain sovereign.

00:51:17So that it doesn’t feel awkward, that you don’t feel uncomfortable, because when you press the button, what happens then?

00:51:23Everyone knows you don’t feel awkward when you get in the car and drive, because somehow you know up front,

00:51:29usually in the motor, there are also people that come to mind, yes, but in principle, let’s say,

00:51:37or conversely, the people who drive with great insecurity are also the ones who have absolutely no clue about the whole thing.

00:51:43There’s a slight correlation, just a little bit, that means, since the world is increasingly being

00:51:52equipped with IT artifacts, hopefully changing our lives, simplifying it,

00:51:59but also making it very complicated. It is advisable for everyone to navigate confidently

00:52:06in this world by at least having a true sense of what is actually happening?

00:52:10That makes me, yes, sovereign, simply, I am not so dependent, I don’t feel so

00:52:18alone. I feel okay, I can handle it, I can assess what happens. I can take the

00:52:25next step, I don't stand still. When I look at it like this, I mean, what have

00:52:29we talked about today? I believe what comes across to the listeners is that skills are more

00:52:36than just the skill that the agent learns. Skills is, yes, how the agent already approaches the topic

00:52:43of work skills, but also how does the person handle it? How do we as humans

00:52:48deal with each other and with skills? And the third was also the topic of how does one

00:52:52handle it as an organization? We had the topic of boxes, flat hierarchies. And I believe, and the question

00:52:58I don't want to dwell on that too much, I just want to leave it with you for some thought,

00:53:01to think about it again. Last time on the podcast you said, not with the motto,

00:53:06What do I need soon, or what do I need in five years, but how will it be in 50 years,

00:53:13you had brought it up similarly, I believe, for the first time.

00:53:16From that perspective, I found what we discussed today totally fascinating, I also found it totally exciting,

00:53:21what has all happened since our last time, we hinted at it briefly at the beginning,

00:53:25because if we had said back then what technology is today capable of,

00:53:32I think we all would never have thought, and I personally do not exempt myself from that,

00:53:37that these technological possibilities, which emerged at the time of recording,

00:53:43and also the recording is on a different day than the broadcast. From that perspective,

00:53:48let's see what might come by then as well. This is a fascinatingly intense time.

00:53:52Yes, so from that perspective, I don't know how you all feel, but I found it very successful,

00:53:58While we were talking, I definitely want to get in touch again, so that we can

00:54:02have another episode someday, because I mean, this is a totally exciting

00:54:05topic.

00:54:06And from that perspective, unless someone feels they would like to give a concluding word

00:54:10to us.

00:54:11Yes, I would just give you a great tip, so try to always discuss people

00:54:15from a year ago and then discuss the delta basically what was discussed back then

00:54:20for about 10 minutes.

00:54:21Yes, gladly.

00:54:22The 10 from back then, a year later, riff again with the same person.

00:54:25That would certainly be exciting.

00:54:26I would be happy to offer that as well.

00:54:27And for all listeners, think briefly about yourself now and a year ago, what you believed

00:54:35a year ago was possible today, what is actually possible today. And this delta

00:54:40can also be projected forward. Then you have a little idea that the environment

00:54:44is inevitably approaching us, but in such a positive sense that we can draw a lot of

00:54:49impact from it. So engage with the topic. That’s the most important thing,

00:54:53that you could tell yourself and yes I think it's really great to chat with you, it's really great because we

00:54:59can all tell that everyone has content, everyone is talking about something and with content and Jens your open talk,

00:55:06I’d love to take a look at that,

00:55:10But with that background, man, man, then I would joyfully

00:55:17initiate the closing, no thanks for being here,

00:55:20Yes, as always a pleasure. Jens, it's nice that you're also part of an episode with a guest.

00:55:27I always enjoy that. By the way, we have the next one lined up.

00:55:30That’s carnival.

00:55:31That’s carnival.

00:55:32That’s a repeat offender.

00:55:33You still need to correct yourself so you don’t miss the next one.

00:55:35You have to say it’s carnival and not faschingen.

00:55:38You say that to a Hessian, you know?

00:55:40I can handle that, Würmscher on the Türmsche, with the Schirmscher and at the Ämge comes the Stürmscher,

00:55:43Würmscher with the Schirmscher and at the Ämge from the Türmsche.

00:55:45You know, I can do that, right?

00:55:47And I can't separate fascism and carnival there.

00:55:49Yes, I used to be from Hesse, went to school in Rhineland-Palatinate.

00:55:53Someone once said to me, so you're going over to the Lacher in the Anne Bundesland.

00:55:56Well, let's leave that aside for now.

00:55:58It was nice with you all.

00:56:00Thank you for being here.

00:56:01I'm looking forward to the next one.

00:56:02If you liked it, then leave a like.

00:56:04A comment, tell your friends and acquaintances that this podcast exists.

00:56:08Should listen to an Openclaw or a myth.

00:56:11We come in peace and are happy to have you as subscribers.

00:56:14And with that in mind, I thank you, see you soon.

00:56:17Bye.

00:56:19Welcome to ThinkDifferent, ThinkAI, the podcast by Mark and Jens.

00:56:25Two technology-loving minds who not only talk about artificial intelligence but live it.

00:56:31Here you will find clear classifications, real practical insights, and a fresh look at what is possible.

00:56:38Understandable, critical, and always with a wink.

00:56:42H.I. for thought, for a chuckle, and above all, for conversation.