Think Different. Think AI. Transcript archive

Software Ade

Published Duration 49 min

Auf Deutsch lesen

Topics Interfaces und InteraktionModelle und Anbieter

What it is about

Was ist Software noch wert, wenn sich alles nachbauen lässt?

Reverse Engineering war lange eine hohe Kunst. Mit KI wird daraus ein Alltagswerkzeug, und das stellt das Geschäftsmodell vieler Softwarehäuser infrage. Mark und Jens sprechen darüber, was Software noch wert ist, wenn fremde Programme in Stunden nachgebaut werden können. Die Folge beginnt mit Fernsehgewohnheiten aus Marks Kindheit und endet bei Spielideen und einem Ausblick auf eine künftige Folge.

Der Einstieg führt über Spiele, die sich mischen lassen. Mario in Fortnite, Bubble Bobble gekreuzt mit Populous: Wer Figuren und Mechaniken aus fremden Spielen herauslöst, erzeugt Kombinationen, auf die vorher niemand gekommen ist. Ob das legal ist, lassen die beiden offen. Es ist keine Rechtsberatung und keine Aufforderung, sondern eine Beobachtung dessen, was gerade draußen passiert.

Bei Grafiksoftware wie Figma stellt sich eine andere Frage als die, ob man sie nachbauen kann. Warum sollte man ein Werkzeug kopieren, das für eine Arbeitsweise von vor 30 Jahren gebaut wurde? Mark beschreibt, wie ein Kollege mit Opus in kurzer Zeit ein interaktives 3D-Modell im Browser gebaut hat, mit Drehreglern, die man selbst bedient. Das Werkzeug ist auf die eigene Arbeit zugeschnitten, statt eine Lizenz für ein allgemeines Programm zu bezahlen.

Ein Versuch geht weiter: Eine bestehende App wird so nachgebaut, dass sie als MCP-Server läuft. Eine KI kann dann direkt mit dem Backend sprechen, ohne Oberfläche. Damit wird aus einer Benutzeroberfläche eine Schnittstelle. Wenn Software leicht nachzubauen ist, verliert das fertige Produkt an Wert, und die Dienstleistung drumherum gewinnt. Wartung, Anpassung und die Geschäftsbeziehung bleiben. Für Anbieter heißt das, Lizenzlaufzeiten und Verträge neu zu denken, und vielleicht Beratung statt Paket zu verkaufen. Bestandskunden sind dabei ein Vorteil, den ein frischer Nachbau nicht sofort hat.

Auch alte Hardware fällt darunter. Geräte, für die der Hersteller keine Updates mehr liefert, lassen sich analysieren und neu beleben. Mark nennt eine alte Lautsprecherbox mit leuchtendem Ring, auf der er eine eigene KI-Stimme laufen lassen will.

Die Kehrseite: Wer Zugang zu guten Modellen, zu Tokens und zu Zeit hat, kann das nutzen. Wer nicht, bleibt außen vor. Die Gruppe, die damit tatsächlich Software baut, ist klein. In der Beispielrechnung der Folge kostet eine Agentur für eine kleine App schnell eine fünfstellige Summe, während die Tokens dafür im Bereich weniger Euro liegen. Die beiden sehen darin eher eine Kluft, die KI verkleinern kann, als eine, die sie vergrößert.

Zum Schluss wird es spielerisch. Maniac Mansion, Zak McKracken und Manor Lords stehen zur Auswahl. Mark würde das mittelalterliche Aufbauspiel Manor Lords mit einer globalen Simulation kreuzen, die vom Hinterhof bis zum Planeten reicht. Die Idee, Spieloberflächen als Interface für KI-Systeme zu nutzen, kommt ebenfalls auf den Tisch. Für eine künftige Folge schlagen die beiden vor, die Skills im eigenen KI-System zu besprechen.

Listen to the episode As Markdown Read the article

Transcript

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

00:00:07Two technology-loving minds who do not just talk about artificial intelligence, they live it.

00:00:14Here you get clear classifications, real practical insights and a fresh look at what is possible.

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

00:00:24AI to think about, to smile about and above all to join in on.

00:00:33A warm welcome to Think Different. Think AI.

00:00:37And before I say something wrong, Jens, is it you?

00:00:42Of course it is me, who else would it be?

00:00:44Yes, live and in colour.

00:00:45It could have been, Reverse Engineer Anything,

00:00:48you could also have conjured up a new Jens here.

00:00:51I have not tried that yet.

00:00:52Nice that you are here, Jens.

00:00:54Thanks, thanks.

00:00:55Your hint already on the broadcast.

00:00:57It just occurred to me, live and in colour, for the younger listeners among us.

00:01:01That comes from the phase when, now I may have to come out, there was still no

00:01:06colour television.

00:01:07People always said live and in black and white.

00:01:11I still know two phrases from television, when it said, we have an office, so that when

00:01:17the presenters on television called somewhere, they were in the sound booth,

00:01:21they knew when to switch over.

00:01:22And you are speaking with a living remote control.

00:01:26I remember faintly, well in case my parents, who also listen,

00:01:31send their greetings to this side, that they meant it differently, that is how I remember it.

00:01:34that there was always the case that I was sent to the television

00:01:38to switch channels, because there was no remote control, and I was then

00:01:42effectively the human remote control. My father was also a magician,

00:01:47and back then he simply, the television is off. What I did not see

00:01:51was that behind the wall he stretched his arm out so far

00:01:54that he touched the fuse box, and I was fascinated.

00:01:59That was still the phase when there were no couch potatoes, because you still had to get up

00:02:04to walk to the television.

00:02:06Yes, I do not know which vice people had back then, and why people thought back then, when

00:02:12I was young, the youth of today will come to nothing, that is somehow in every

00:02:20generation.

00:02:21Somewhere I recently heard a quote, I cannot put it together any more,

00:02:22and then it turned out it was from ancient Greece, from a time long

00:02:26before our world.

00:02:27Never mind.

00:02:28Let me put it this way, not everything was better in the old days.

00:02:30Let us record that.

00:02:32But not everything was bad either.

00:02:34So, come on.

00:02:35Now we have covered so much in general.

00:02:38Let us talk about the value of software.

00:02:46Without wanting to step on anyone's professorship.

00:02:49We had Polar Downer here.

00:02:51We also have people in the listener circle who deal with it.

00:02:55But let us really take up the question,

00:02:56what is the value of software in a time like this? I worked it out with Rea, where a framework takes off on social media that effectively promises you we can perhaps still

00:03:03talk a bit about copying software, just by having a file here,

00:03:09a link here, you do not even need access, and what kind of stylistic blossoms come out of that,

00:03:16we can also talk about right away, but what does that mean for our time? Or do we want

00:03:22to first listen to what happened. Gladly, so the topic is

00:03:26reverse engineering, so looking at something again and then in principle, by

00:03:30taking it apart, establishing how it was actually put together

00:03:35in the first place. That is in principle one of the high arts of software development,

00:03:39because it is not that simple at all. You have to know that the average

00:03:44software developer does not always follow plan A, accordingly,

00:03:46I do not want to step on the professor's toes here either, there is

00:03:50quite often code that is a bit more complicated at one point or another. You have

00:03:53intended to comment every single line you have written. That is easily forgotten.

00:03:57After the 30th page of code you have written, it is no longer so clean.

00:04:02Accordingly it is not that easy then, in partly more complicated pieces of

00:04:07software, where millions of lines of code can accumulate, to look at that,

00:04:11the code, and to work out from that code what this

00:04:17software actually does. Or in very, very complicated cases,

00:04:21because in the last few weeks, for example, complicated games were also

00:04:24reverse-engineered and made browser-capable. What then actually

00:04:28does a character do that hops around somewhere in such a game? That is

00:04:32all hidden in the code, and just by looking at the

00:04:35files, as you have just described, the source code in these

00:04:41files, turning it back into a working version is

00:04:44ZZZ

00:04:49actually a sacred work when you do it without AI.

00:04:54What you said about games, that is also what I first got into.

00:04:57Actually about two things. One is this Sony PlayStation stuff, whether that has to do with

00:05:02Rea or not, I do not know at all, but that is also a bit of a hype right now, that not only a jailbreak for the PlayStation comes out, but

00:05:05that now also basically everything that is not on the tree at 3 is becoming runnable and combinable from the PlayStation 5 on the PC, greetings go out

00:05:10to the topic, software no longer belongs to you. Perhaps the manufacturer will reconsider that,

00:05:13another topic. And again, no, again, that has already happened many

00:05:18many years ago. Sony too, from that side. One should

00:05:22not underestimate it. Meanwhile the kiddies also have more powerful

00:05:26tools at hand. But where it definitely ran more into my way with Rea was the topic of mixing games. That you

00:05:30walk around with Super Mario in Fortnite, that it goes so far that you not only have the

00:05:34character, that you not only have the sound, but that you also have such glitches, so special features

00:05:39from a game, from which you effectively pull the character out, in it as well, there they had

00:05:44something like, when Mario walks up the stairs backwards and jumps through the

00:05:49door at the top, that he then flies to the next stage, and that happened in

00:05:58Fortran, or that in Super Mario Kart you have someone sitting at the sniper and then shoots the character,

00:06:03sorry, out of the cars, and you think, damn it,

00:06:06how cool is that? Not because of the snipers and the cars, but that you basically have the

00:06:12chance, and I do not want to judge at all whether you are allowed to, and yes, but simply

00:06:18letting your imagination run free and say, I suddenly combine the heroes

00:06:24of my games, my favourite games, everything with one another, mix it into a new game,

00:06:29and then it also raises the question, how games come about,

00:06:34...

00:06:40the question, how games come about,

00:06:47...

00:06:55we had it with the Gamescom episode and other places for a long time, and partly

00:07:00also, let me say, love in it, at least love for detail, depending on which game

00:07:04you take, if it is called Wolverine, and then suddenly a thing comes round the corner

00:07:10and says, I always thought it was good to cross Populous with Fortnite and Bubble Bobble, and

00:07:15suddenly you wonder why your characters are packed in soap bubbles by little dragons.

00:07:20That is a bit wild. So we should also come back to this topic

00:07:26shortly, as you invited, what software is still worth. But of course we can

00:07:30also take a little look in there, that it of course offers an explosion of

00:07:34creativity again. Because this combination that you just mentioned,

00:07:38nobody had thought of until now.

00:07:41Well, licences are attached to that too.

00:07:44Licences are attached, all fine, all good,

00:07:46but of course it is still somehow, in a different way,

00:07:49exciting again, what is in principle possible,

00:07:52now combining the most varied game characters, game ideas

00:07:56super quickly with one another.

00:07:58Accordingly, I am often very positive about things like that,

00:08:03honestly, and I think this could be another...

00:08:05That is already, well, the internet went wild there.

00:08:09So there it is an explosion of creativity, whether it is always legal?

00:08:14No, half of it is efficient.

00:08:16No lawyers, no legal advice.

00:08:18Exactly, no legal advice.

00:08:20We do not know.

00:08:22But accordingly of course that is not everything, no matter what is done there.

00:08:24But it shows where we are heading.

00:08:26So to really say, I can by now build pretty much everything I need,

00:08:35whether it is a game or a piece of software. This moment has now been

00:08:41reached, and it is always built securely. You know all the other things. There are also enough

00:08:44at the moment on Instagram or elsewhere, where people say, yes, we love

00:08:49bad code, of course you forgot, here are the 110 things you forgot to do,

00:08:53like security or other things, therefore your app is not secure or something like that.

00:08:56Stay with us until the end, then I will tell you the most important one.

00:08:59Exactly. Accordingly, of course, it is not always immediately one hundred percent secure and clean,

00:09:07but the AI can then also solve that and help you, even if you have never done it before in your

00:09:11life. So the matter is now that I say, we are now entering

00:09:17an age where it is, well, it has never existed with this kind of software,

00:09:23but of course that is one aspect, but for me the aspect is much more exciting, that I

00:09:27say, we have to rethink everything now, because software too, just like a game,

00:09:34the respective developer thought within a certain box, in order to solve a certain problem in a

00:09:38certain situation.

00:09:39And even if he did it so well that it in principle helps many, then

00:09:43many have in principle bought this software.

00:09:46Now the point is that I say, if we look at graphics software now, there was

00:09:51also a famous example in the last few weeks, there was this, I

00:09:55think it is called Handcraft, that someone in the Photoshop address...

00:09:59Figma, it is called, is it not?

00:10:00Figma too, I had not caught that yet.

00:10:02Now Figma as well, by now.

00:10:03For everyone who now thinks we are a decent podcast, yes, we are, those who

00:10:09are supposed to be called that.

00:10:10Is not that Hessian.

00:10:11Yes, not Hessian Figma and always Azure with Azure, yes, but the software is called Figma.

00:10:16Good heavens.

00:10:17Okay, yes.

00:10:18Spelled F-I-G-M-A, they also had an even funnier name, I find,

00:10:23in German.

00:10:24They also introduced another whiteboard, collaboration board, which was then called Jam, that was then

00:10:28FigJam, the name of this whiteboard tool from Figma, somehow strange as well, all

00:10:33right, it cannot always work with localising names across the whole

00:10:37world, that is a difficult business too.

00:10:39Back to the topic, all of that can now effectively be rebuilt, does it make sense to rebuild it?

00:10:45No, that makes sense if I want to save the money, that is one thing

00:10:49one could do.

00:10:50What I think is much more exciting about this topic is that you can now say, do I need

00:10:55this software at all, which exists anyway, then I could also buy it?

00:11:00Always like that?

00:11:01Or do I actually need something else, because that is a bit of my survival,

00:11:04slowly, that I say, yes, it is of course cool if I have such a graphics editing tool,

00:11:09but honestly, I can by now produce almost all the graphics I need in some way

00:11:13with AI as well.

00:11:15What do I need?

00:11:16actually a tool that helps me very specifically, perhaps as an editor, to be able to influence this graphic in my

00:11:23personal way, and not, as has been usual in desktop publishing for the last 20

00:11:29or 30 years, that I somehow move a mouse pointer over

00:11:34a frame around several objects, mark these objects somehow, give them a colour, give them

00:11:39a line width, a background colour and a foreground colour. That is basically what

00:11:44this graphics program has been doing for the last 30 years, with a bit more fancy stuff.

00:11:49I think that is what is most exciting that I see right now, that we are in a phase

00:11:54where everyone out there can build their own tools for their own situation.

00:12:00And those do not have to be the same tools, because that is actually silly.

00:12:02Replicating the same tools that already exist, with AI, is nice to look at,

00:12:07but apart from a certain kind of perhaps modern piracy, nothing

00:12:14new that moves the world at that moment. It only shows where the AI capability is,

00:12:19in my opinion it is zero creativity for the people who are doing it right now.

00:12:23The creativity comes more from what we have seen with the games, that people

00:12:27combine things, or that other people build their own way

00:12:32of working with devices, with appliances, because reverse engineering

00:12:36works just as well for, I do not know, the stereo system at home that is somehow

00:12:40connected, or the light switching that I have in my house

00:12:43at that moment. I found that totally great with this reverse engineering, that is,

00:12:47Rea GitHub. What are the top projects? The top projects do not start at all

00:12:51with rebuilding Word, Excel, PowerPoint. Greetings go out, as said, to Adobe, Word, Excel,

00:12:55PowerPoint, blah blah blah, all of that has been tinkered with by now. Special

00:12:59topic for old things, reactivating drivers, for

00:13:05old things, reactivating connections, bringing back support for old file formats,

00:13:10that I actually found astonishing, that it was very far at the top in the

00:13:15statistics. I noticed that I was the old white man and the uncreative one while you were talking, because I

00:13:20first thought, what does that mean for the software market, what

00:13:24first came to mind, what does that mean for the software market, what

00:13:28software licences are concerned, what does that mean for the software market, what it

00:13:33means that companies also, let me say in quotation marks, whether with or

00:13:36without AI, make an effort for software, when, yes, that is stolen again anyway or whatever,

00:13:42yes.

00:13:43So as I said, no matter whether it is legal, illegal or whether it is from somewhere

00:13:47in the world, that is completely irrelevant, just assessing the result, what that means, and

00:13:53my creativity is not such that I have not even tried the thing out.

00:13:57My creativity was actually a completely different topic, namely, I took one

00:13:58and said, you watch, do not rebuild the app. So a bit of Xcode and so on

00:14:02I can manage well with Claude, I do not need anyone there who tells me,

00:14:08build an app. My question was, look at the app, rebuild the app for me, but as an MCP server,

00:14:12so that you effectively do not have the app frontend, but something that AI can consume.

00:14:19And then I thought, okay, I can anticipate it, the result was

00:14:30that it works, and then you had something that you can integrate as a tool in Codex

00:14:35or wherever in the cloud, and then, after you have done a login,

00:14:43talk to the backend that would actually have spoken to you yourself.

00:14:48I find the question fascinating, what the backend thinks, thinking is also

00:14:54relevant, but what the backend thinks, what it speaks with, whether it was that Android version

00:14:58that was the basis there, or whether it effectively noticed, haha, that is something different. And that

00:15:02is not meant to encourage imitation at all, but the topic is that suddenly you can also manage

00:15:09conversions, to say, I have a result here, damn it, make it

00:15:13run again, make it run on another system, just as PlayStation players suddenly run in the

00:15:19browser, suddenly an app runs there as an MCP server. It is dissolving boundaries. So

00:15:24runs in the browser, suddenly an app runs there as an MCP server. The boundaries are dissolving. So

00:15:31things are changing extremely fast right now. And you recently showed me several videos

00:15:36from someone who tries to present knowledge as HTML pages, perhaps

00:15:42run it again. You can explain that better than I can. If you now go and say,

00:15:46okay, I like the idea, I would like to apply it to another topic. Dear Rea, or

00:15:51Claude can by now probably do a great deal of that alone too,

00:15:54let us leave that aside. But dear AI, with whatever tools, orient yourself on that,

00:15:58bring it to this and that level, bring something new into it. I think

00:16:03we will experience more and more that this exponential growth of floods with

00:16:09faster, higher, further, that the system really picks up speed, because software is no longer

00:16:16copied, it is imitated, because software is merged, because software, I would call

00:16:22it a different state of matter now, so a frontend becomes an interface,

00:16:27an interface becomes... I do not know what. So suddenly entirely new things will arise.

00:16:32The question that remains, of course, for me, from a societal point of view, is,

00:16:35what does that mean for the people who build software today? What does it mean

00:16:39for people who invest in software today? What does it mean for startups, when nobody

00:16:46knows what happens. Suddenly anyone can now rebuild any character whatsoever,

00:16:54play it. What does that mean for copyright? I may not copy Lara Croft, but I could,

00:17:00in theory, theoretically, no call to action, rebuild the whole game and just have the

00:17:05textures changed. Is it still Lara Croft, or is it like it was back then with Mario and

00:17:10Yes, well, suddenly it gets really wild.

00:17:15Definitely, and let me just say briefly, what you had just written, this

00:17:20example, we can put the link in as well, that is a colleague who is actually

00:17:24at the moment, with Opus, quickly producing three-dimensional, clearly understandable things

00:17:35that you can fully rotate in the browser, these things he builds, that means he has such

00:17:38built like a cardiovascular simulator, like erosion happening in a valley

00:17:47when a river winds through it, and how a turbine works, how a rocket turbine works.

00:17:52All of these things, you can still play around with sliders, you can turn it from all sides...

00:17:56So interactive information graphics, three-dimensional, in the browser.

00:18:01Yes, and he has, in principle, a bit of that, that was for me,

00:18:05and Zap we had both talked about that too, that was a bit of the occasion when I saw that,

00:18:09and all the rotary controls he built in, and that it was built so quickly every time,

00:18:12we thought, that is actually a good idea, to say, okay, why does one not build that

00:18:19for one's own workflows in principle, which I have just described?

00:18:22Do I actually have to learn Figma or Adobe programs somehow to edit graphics?

00:18:29Or do I not simply say to myself, look, if I can now build an interactive module

00:18:34to work out a cardiovascular system, then I will also be able to build an interactive module

00:18:38to change the colour of an image from black to colourful, or to introduce some style,

00:18:45or to say, make a three-dimensional object out of this two-dimensional one, and I can

00:18:50do that with my own sliders, because I may still want to fine-tune it a bit

00:18:54and every time by text again, perhaps with the AI, the prompt image,

00:18:59I need my own tool landscape. That is why it fascinated me, and that is, I think,

00:19:03just a little bit what I notice, what I see, what is happening

00:19:07out there right now. And of course that has an impact on the software industry, as

00:19:13it currently still works. It will still be able to work that way for a while, no question, but let us

00:19:18maybe slice that a bit further. When I introduce software, for example in a large

00:19:21company, then there is often a big cost driver hidden in it, namely that this software only

00:19:25fits the respective company perhaps 80% or 90%. Be it, it depends on the skills of the people in the

00:19:30company, who first have to be trained with it, or that in principle

00:19:37some topics still have to be transported differently, other data

00:19:41has to be connected, then there are perhaps structures that are a bit different, the workflows were different. That does not fit exactly

00:19:43with how the software designer perhaps imagined it, or the large company that came up with it. And there a lot of

00:19:48money is invested in success managers, software, people who come into the company

00:19:51to help the company learn and adapt this software.

00:19:54I would say that is a first big lever for the software companies, to think

00:19:57about how they can actually take a different approach in future,

00:19:59by presenting this part of their software in a much more personalised way for the respective

00:20:05company, because I simply say, what people outside can do when

00:20:10my software can be completely reverse-engineered, why not offer it as a service,

00:20:15for example, by saying, dear company X and dear company A, you get here on

00:20:19this proven product, which has run stably for years, which has been developed according to software-development principles,

00:20:25which is secure, which you already use, or

00:20:29if you have heard that it is always used well at other companies,

00:20:34we can personalise it perfectly for you, so that the phase in which you actually

00:20:40have a long induction period becomes much shorter. And your processes are then also much shorter,

00:20:47because this software will keep adapting to you, because we can always

00:20:50involve the topic further with AI help. I would do it that way. That is a change,

00:20:52the software. You will perhaps not need so many people any more to carry out

00:20:58trainings, because the training will become less. But you can simply use

00:21:03the people for other things. So that means, one has to adapt. That

00:21:07AI help involve the topic further. I would do it that way. That is a change,

00:21:11the software. Perhaps you will not need so many people any more to carry out

00:21:15trainings, because the training will decline. But you can simply

00:21:19use the people for other things. So that means one has to adapt. That

00:21:23has always been so when progress came. We all know that with AI it has become a bit crazier

00:21:27still, that things go quickly. But there is really a lot of music in it too, also for existing software providers out there, who should not simply throw in the towel

00:21:32and say, hey, sorry, everything can be reverse-engineered anyway now.

00:21:37We close our shop, because anyone from the street can now rebuild our software.

00:21:42No, I think there is also a chance in continuing to say, yes, business is not just

00:21:46about putting a piece of tool out there, it also includes a business relationship

00:21:51that I have with someone. I have that, and that is an asset that many, who are now

00:21:54first rebuilding a new piece of software, may not have yet. That can be used very well for some time

00:21:59and placed in this market, to also learn from this market

00:22:04how the software market can function in future. Because I do not believe,

00:22:07and correct me if I am wrong, that in every situation all software will now be generated automatically by an AI, even if back then our beloved NVIDIA boss said

00:22:15that in future NVIDIA pixels will be generated. Yes, that may be in some distant future.

00:22:23Is it economical that we say everyone in the world now builds their own

00:22:28word processor or their own adapted one? No, probably not,

00:22:33honestly. It also makes sense that someone, from my point of view, builds a vibe-coded

00:22:37writing document or writing editor software that is better than perhaps established ones,

00:22:42but not everyone has to build after it.

00:22:47I think there is also room and music there, to say, I have a piece of software for you, whether AI-generated or not,

00:22:53that does not matter.

00:23:07But that will help you in a certain situation. Please feel free to pay a few euros for being allowed to use it. That will go on.

00:23:15I think this will change the way we build software once more.

00:23:20Because the time when you say, I now have a project.

00:23:25This project builds a piece of code.

00:23:28This code is used in CancelNAPE, on the website, in one piece, I do not know, a kitchen appliance, whatever.

00:23:37And then you sit there with large teams in rituals and build something like that for a year

00:23:44and look forward to a release like an Apple operating system release every year,

00:23:49and that it gets modernised a bit so that you see

00:23:57adaptations faster, that you see topics taken over faster, along the lines of

00:24:03yes, who knows. People, a fictional example. You have Mac OS and Windows, and there are people

00:24:10who have Mac OS but still miss Windows when they see the Explorer of Windows,

00:24:15that is, the file manager, because the Mac Finder cannot do that. Now you are basically

00:24:20able, as a user, to say, let us build such a thing that looks like the

00:24:25Explorer, I would like that for Mac. But you could also, as Apple, which you could do before too,

00:24:30you perhaps had limited resources, limited time. You thought, I will not rebuild

00:24:34that, the Finder is one of the special features of my operating system, blah blah blah.

00:24:38I will not do it, I do not want to, I cannot, I have no time or whatever.

00:24:41And I think at that point we will experience a change, which again ends in an

00:24:46acceleration, because suddenly the possibility, how long does it take between,

00:24:51I have the idea, a vision for my product, to, it is tested in 40,000

00:24:58different variants. You do a test or whatever, along the lines of Monday

00:25:02a rough idea, Friday in the product. That is the possibility you now have, and you

00:25:07can test things much more easily. You can simply create facts

00:25:12and if it does not fit, I change it again. You have a much more

00:25:16dynamic setup, with which you can, in quotation marks, operate with fewer people in

00:25:20the shortest time. And that, I think, we will notice.

00:25:25Whatever the size, whether you are a consultant or a software vendor, that will

00:25:34change, and thereby also make the basic question harder for the people who say,

00:25:40but this is hand-made here, this is crafted, I need, the

00:25:46question earlier, when you said, you are building software, I need four weeks.

00:25:50People thought, why do you need four weeks, the screens are finished.

00:25:54Because I still have to do this and that. Today someone says, I need another four

00:25:59weeks, then he is asked, why? You do not even have the screens yet, but that does not matter. Today by

00:26:02evening everything is done, through the AI. What more do you want? Yes, well, I still find that

00:26:08difficult. That will of course change, and the cycles are shorter,

00:26:13definitely, and I think I would now also say that with the

00:26:17classic licence sale, often in the corporate environment software licences

00:26:23are also often sold for one, two, three years or even longer. I think that will

00:26:30also become an interesting question for one or other decision-maker. For how long do I buy

00:26:37software now? So something will simply change there. The cycles will,

00:26:40I think, partly become shorter. What could also happen, in my opinion, and perhaps also an opportunity for one or another,

00:26:44is to say, yes, you no longer buy any finished piece of software at all.

00:26:48I simply deliver you a solution. I am simply the one

00:26:52with my software who helps you, in your workflow, to actually always have the right

00:27:00software solution.

00:27:01Because we both know that in this AI world everything could perhaps look different again in five months,

00:27:07I simply sell you my consulting contract only, and no longer a fixed piece of software at that moment for this part

00:27:12of the assignment.

00:27:15And that scales, because I am a software house, I have the people, I have in principle

00:27:16the software engineers, I have the success managers, who can be on site with you and observe

00:27:20how software is in principle used, what needs your internal people have in

00:27:24can, how software is in principle used, what needs your internal people in

00:27:29principle have in order to access things.

00:27:31Let me into your house, or into your company, and I build, in principle, with

00:27:37my two-year contract that I have, you will have 400 versions

00:27:44of this software in the next two years.

00:27:45And you will probably look completely different at the end of those two years,

00:27:48these software solutions that you bought on day one. I have also been through something like that.

00:27:52I think such business models have to be thought through now, how that can work.

00:27:56These are also new contracts that have to be drawn up, because you no longer

00:27:59buy the package the way you perhaps used to. But I think there is definitely,

00:28:04there is always this topic again, as I said now, it is also a question of trust, of course. Do I trust any software first, that has somewhere been quickly

00:28:09reverse-engineered. Or do I rather trust the software partner

00:28:14with whom I have worked for a little longer.

00:28:19I think pressure will come onto the market. The expectation will rise at the moment. The

00:28:21expectation also regarding pricing. But I have just given a few

00:28:25hints of how one might be creative there. Nevertheless, still cut off a slice of the

00:28:29cake. But of course there will be a change there. I think,

00:28:33that is actually now, that was already happening since the AI phase,

00:28:38I mean, the software industry was the first, where the production of pure code,

00:28:44which was then sold in the best case in a shoebox, as it used to be,

00:28:51and installed on Mercedes. That phase is over.

00:28:56You cannot complain it back now, however you like.

00:29:00It will of course still take a while, we are not always immediately everything different,

00:29:03but it is now slowly ending. Because there you can really say, okay, that can

00:29:09Because there you can really say, okay, that can

00:29:15be different. The phase of inflated prices for software may also be over. That

00:29:21means the actual piece of product is no longer the decisive thing. The product today,

00:29:27when it is digital, can be copied very, very quickly. We also see physical products,

00:29:33it comes gradually, will also at some point be very, very quickly copyable. But above all digital

00:29:37products, at that moment the product is no longer worth so much. And the product,

00:29:43if it is software, is no longer worth much. The service around it is what

00:29:46is worth something. The maintenance, the adaptation, that costs us money and time at that moment. And

00:29:52for me that actually belongs to a complete product development. That is what

00:29:56people always forget. And that is also where, but of course, that means,

00:30:01of course, it is also where people, first of all, look more at the one-off price

00:30:08than at the long-term service, along the lines of, look,

00:30:13it costs 80 euros and not 300. For the 80 euros I can buy it three times again

00:30:19before I reach the 300. Why on earth should I buy again, take the more expensive product, there is,

00:30:23that is very simply calculated now, I admit that. But as you say, that will create movement.

00:30:28Yes, definitely. Now let us briefly take the microphone away. Definitely.

00:30:33By the way, the AI will cut that out again. That means nobody noticed the pause, and you

00:30:39say the microphone is off. So Jens had a really long pause at that moment,

00:30:45and I was briefly afraid he had gone.

00:30:50No, no, all good.

00:30:52But you see, that shows that we are still real.

00:30:53Real handmade work is still being done here.

00:30:56When we talk here, mistakes may also happen.

00:30:58I will slip away briefly with my microphone button.

00:31:01I have found the following for you on the internet.

00:31:04I have found the following on the internet for you.

00:31:07Oh, I see.

00:31:08I have found the following for...

00:31:09That is Siri, when she does not understand you.

00:31:11Damn it, I wanted to make a little joke there,

00:31:14and you stab me in the back like that.

00:31:16So, carry on.

00:31:17Which series?

00:31:19Siri, not series, Siri, when we...

00:31:22Oh, Siri, Siri.

00:31:25You will have to work on our mumble, Mark.

00:31:27Mumble?

00:31:28Hold on, ash with ash.

00:31:30Really?

00:31:31Yes, well.

00:31:33In any case, let us come back to reverse engineering,

00:31:35which also simply has other possibilities.

00:31:37Well, we have now, I think, trampled to death enough

00:31:38that it will change the software industry.

00:31:40That is part of this AI revolution.

00:31:43And that should now also become clear to the last one,

00:31:46since the last two, three weeks, since this topic has become so hot.

00:31:48What is exciting, in fact, is what we may have left a bit out of view, that we do not only, when you think of reverse engineering,

00:31:50think about it in that way, and that we,

00:31:54did not leave out, that we do not only, when you think of reverse engineering,

00:31:59then you actually always think of the digital products that you see.

00:32:03Because we had just this topic of hardware too, where software is sometimes

00:32:08inside. Chips run there, small things that sit on it, which

00:32:12sometimes also connect to the internet and other things.

00:32:14All of that can of course in principle also be reverse-engineered.

00:32:18So that means old products that I perhaps have at home, where a software provider

00:32:23or the product manufacturer from back then says, you know what, the device, your old fridge,

00:32:28or something else, I am no longer maintaining that, there is no update any more, something else

00:32:32for.

00:32:33All these topics can now be, whether a connection is somehow there,

00:32:37reverse-engineered, and I can write my own software again

00:32:41for other things.

00:32:42For example, I have somewhere lying around a small box from a larger provider.

00:32:47Back then a rather bad pseudo-AI voice ran on it, which wanted to answer me.

00:32:55But they are good, because they have such a nice ring that lights up when you address them.

00:33:01For their size they actually have no bad quality in the speaker either.

00:33:07I will simply flash this thing now, reverse-engineer it, and put my own AI variant

00:33:13on it, and connect it with my second brain and co.,

00:33:15in order to build it, in principle, here in the household.

00:33:17I find that a little cool, you know, I paid for this product years ago.

00:33:21I find that a bit nerdy too, but that is...

00:33:22Yes, but it is cool, you know, I had briefly considered, because I wanted

00:33:23to finally build a physical product here with AI for once, and had

00:33:28already looked for quiet again, and such, which microphone do you buy, and which

00:33:32microphone do you buy, and which

00:33:37then I buy a 3D printer too, so that I can then put this microphone into the speaker in a nice 3D case.

00:33:44Yes, that was again one of those things that again feels like a huge project.

00:33:50So then I came across something else through this reverse-engineering story and thought, well.

00:33:55Up there I do not go, I simply look at this device from the manufacturer, which I will not name, that I have here at home, and

00:34:02see whether I cannot also work it out with the help of Claude.

00:34:06The device is already decades old and can then be made fit again.

00:34:09Otherwise we would no longer use it.

00:34:10Accordingly I actually find that quite nice then,

00:34:14that you can breathe new life into old hardware like that.

00:34:19That is also one of those things.

00:34:20Just now we talked about

00:34:21talked about how I can personalise software,

00:34:25that I can adapt it to my needs.

00:34:27And it goes on.

00:34:28I can indeed also do that with hardware,

00:34:30the way it perhaps was never intended, or maybe with that one device. I mean,

00:34:34everyone probably knows that, whether you have a smart vacuum cleaner, a fan

00:34:39or whatever in your household. The software solutions, the applications,

00:34:44that the manufacturer sometimes provided for them were sometimes only semi-sexy,

00:34:48honestly. For some people that is perhaps again old news,

00:34:53because we can simply go and say, for this hardware solution

00:34:59that I have, I will now simply build my own ventilation software

00:35:04in my own design, with my own cloth. And then I find what I see there cool.

00:35:08And maybe I can look at my own house in 3D and see where the

00:35:13various devices are currently using which energy, which air flows they produce,

00:35:17whatever. That is simply nice. I was just thinking, I am sitting here in front of

00:35:22a Dyson, hot-cold air stuff, what I could do with you while I have

00:35:29my little child-AI bot running on my watch. I wanted to show a little prototype like that to a colleague,

00:35:37along the lines of, oh yes, look, it can answer questions here in a child-friendly way and such

00:35:42stuff. Not because I wanted to release it, but he had the idea to build something and then

00:35:47I thought, I will tinker a prototype for him so he can see how it feels.

00:35:51A different topic, I do not want to digress, but that is also the source of possibilities. But

00:35:56but, how do I put this properly, so as not to get into trouble, so that I like it hot again, the village put it.

00:36:01Access to education, access to possibilities, access to tokens, access to good models

00:36:04increasingly decides what you are enabled to do, and where you perhaps

00:36:12learn about it, but cannot do it, or where you perhaps do not even learn

00:36:19that this works. So I mean, we must not believe here that just because we

00:36:23have a podcast and talk about it, that Vertex is being copied for PowerPoint here and

00:36:28pontificate loftily about what that means for software development and how processes

00:36:33change, outside there are still people who are, let me first say, receptive to it

00:36:37along the lines of, I can save a subscription price that would be very hard for me

00:36:42in my wallet, or those who only see the results and

00:36:48can perhaps assess again what that means socially and politically, simply

00:36:53use it, simply apply it, who knows what tools and solutions there will be in future,

00:36:56or who do not get anything at all, because they are cut off

00:37:00from empowerment and access and goodness knows what, regardless of age, money or region

00:37:04in the world. I believe the scissors of knowledge and of possibility

00:37:11will cut in properly again, more than just, you cannot see my air quotes,

00:37:17which is why I mention it, more than just the topic of which family background, with which

00:37:23you were equipped as a young person or whatever, in order to get access to

00:37:28you were equipped with as a young person, or whatever, to have access to

00:37:32something like that. This is now all becoming much, much more extreme.

00:37:36Yes, definitely. Yes, totally. I mean, I have also looked up, at the same time, André Capati,

00:37:42whom we have also already quoted in the show once or

00:37:46twice, who is also a bit of the

00:37:49namesake of this second-brain topic, I looked for a tweet,

00:37:52I have unfortunately not found it that quickly, I

00:37:54had to look it up again, that went, and so I do not want to say the

00:37:56numbers exactly now either,

00:37:59because I am doing it wrong anyway, but it was a bit about

00:38:01him saying, we are of course still early, that

00:38:04means, and we are a small group, we, we have

00:38:07by now, I think, indeed reached the point where a billion

00:38:10people have access to AI, but that billion is then also more

00:38:16chatting around.

00:38:17And then it is another small group that is concerned with writing code,

00:38:22and another small group perhaps concerned with the topic of agentic,

00:38:26and then another small group that actually says, well, we

00:38:29need to rebuild all software again at some point.

00:38:31What we are describing here has, of course, something to do with it, first, whether

00:38:38I am privileged enough to have the time to deal with these topics,

00:38:42how I am set up, either through work or in private life, so that one

00:38:46can in principle afford all these things, there is absolutely no question, that is

00:38:50of course completely fine, but in spite of all that the needle is of course shifting in

00:38:56the direction that more is possible nonetheless.

00:38:59We have just described it, that we can say, software too, which

00:39:03I had to buy until now, also costs money.

00:39:06And when these software solutions that I no longer have to pay a lot for, because

00:39:12whether I reverse-engineer them and get the same product more cheaply,

00:39:15or with AI I can build something in between, which may save time and money,

00:39:19because I no longer have to load a Word document, or any Drive document, back and forth,

00:39:26convert it into a PowerPoint, a presentation, and afterwards it has become a requirements

00:39:30document, which a software developer then wrote into code in another

00:39:34program. Those are not weeks and days of work that have flowed into it, after all.

00:39:39Such things are then, perhaps, possible for someone who did not have the means until then. That person must of course actually

00:39:45spend money on tokens, but that is by far less than he would have spent if he had sat down and said, now I want

00:39:50my own app for my flea market, or for my flower shop. That would have cost 10,000 euros

00:39:55yesterday. Today and tomorrow, that is of course perhaps a few hundred, if at all,

00:39:59perhaps even only 10 euros of token costs that he has to incur

00:40:05to build his solution. So I think, yes, you are entirely right, that there is still

00:40:11a gap. But in my view this gap is closing

00:40:16more through AI than through anything else. And now perhaps people will stone me for that,

00:40:21but I think AI can really help there. Before you stone me, I would like to put a picture

00:40:27into your head. You have just given me a shower moment. No, we did not

00:40:33shower during this podcast, but while you were speaking, I experienced mental

00:40:37relaxation. And an idea came to me about how I would test Rea. So could,

00:40:44or one puts it like this, I know someone who would try that out,

00:40:50if it were so. Imagine, you say, there are these code management systems,

00:40:56let us call Gitter as an example. And there are systems like OneDrive, Dropbox, BoxNet.

00:41:02if it were so. Imagine, you say, there are these code management systems,

00:41:08let us call Gitter as an example. And there are systems like OneDrive, Dropbox, BoxNet.

00:41:14How about asking the system, let us build something like

00:41:20file management with Git functionality, so versioning is simply accessible.

00:41:26I think nerd hearts could be very happy with that. Let us see what comes out of that. I think we should always do an episode, not along the lines of, what have we

00:41:32all rebuilt, but what helpers have we actually built for ourselves? Which

00:41:35skills do we actually use in our AI system? We should definitely also do an episode

00:41:41on that. I have said that very often just now, I have to apologise again. That is

00:41:44probably because of the late hour. In truth it is 22:00, I am an old man, I have to go to bed.

00:41:48We should definitely do it again. But before we do that. Jens. I started with the sentence,

00:41:52Jens, is it you? You confirmed it. We also suspect, because of the microphone action,

00:41:58that we picked that up. If you allow this assumed sphere,

00:42:05which games would you cross? I have an idea. You have an idea of what I would cross,

00:42:11or do you have an idea of what you would cross? I have an idea of what I would cross. Okay,

00:42:22good, then let me think briefly. And then I return the ball first,

00:42:27because the question is so good that I do not want to answer off the cuff. For me

00:42:32the answer also came to mind while I was asking, yes? I would... Warning, totally boring.

00:42:36Maniac Mansion and Sack McCracken. Why?

00:42:42Maniac Mansion and Sack McCracken. Okay, interesting.

00:42:47Why? In Maniac Mansion you had a chainsaw. In Sack McCracken the petrol for the chainsaw was on the camper van on Mars.

00:42:50But there was no chainsaw.

00:43:00I would really love to see what you could do with the chainsaw in the game.

00:43:02So I would like to cross those games.

00:43:06Are you sure there is no chainsaw in the bug?

00:43:08Okay.

00:43:09Okay.

00:43:10Yes, yes, okay.

00:43:11Yes, yes.

00:43:12So.

00:43:13You are back again now.

00:43:14But could you not, in Zecken McQuacken, on such a games console, play Maniac

00:43:19Mansion?

00:43:20Yes, but you cannot bring the petrol in there.

00:43:23That is right.

00:43:24That is right.

00:43:25Yes, yes.

00:43:26Okay.

00:43:27Good.

00:43:28I would rather have thought of something like this.

00:43:29And that was in Edgar's room.

00:43:30Mhm.

00:43:31Never mind.

00:43:32Yes, I would rather have thought of something like...

00:43:35No, right, wrong, it was Day of the Tentacle where that worked, that was not...

00:43:38Ah, okay, okay, yes, very good, you have saved us from Puzzle Shack.

00:43:42I would have crossed something like this, I am a big

00:43:48fan of Mannerlords, it is still in the, in the early, a kind of simulation game,

00:43:54where you build medieval worlds, yes, it is still in early access, accordingly

00:43:59not quite finished and therefore always a bit limited. I would cross that with some

00:44:05kind of, interestingly, global simulation, like Sim City or something like that.

00:44:18You know that you can say, because in Mannerlords you can

00:44:21go down to a very small area, very, very small, to the individual farm, and decide what happens on the individual farm,

00:44:26in the back courtyard, so to speak, whether chickens or pigs are farmed there, kept and tended,

00:44:32or whether carrots are grown there, with a larger world simulation, where you

00:44:36can then gradually conquer the whole planet.

00:44:41I would combine that, I would find that very, very nice, because there is always only

00:44:44one of the two.

00:44:45You know, where you either move the big troops on the very large level, on some

00:44:48hexagon level, and can only roughly simulate the cities?

00:44:52the two properly. That is what I would do, for example. But only because you asked me like that.

00:44:53If you had asked me what I would want to learn from the reverse-engineering approach

00:45:03in order to make a really new, great game, then it would more probably be

00:45:10something like a really great space simulator with combat content, which does not exist yet.

00:45:14Actually I am currently considering whether I should build my own game now,

00:45:20because interestingly, we also talked about that in the Gamescom episode, through the topic

00:45:25reverse engineering, the momentum in the games industry is turning. In this episode we

00:45:31talked about how the games industry is still proceeding very, very cautiously with the topic,

00:45:36how much software may be developed with AI and how to handle that even in

00:45:40public, because there has been super backlash from the gaming scene so far,

00:45:45because the players did not want so many AI-generated contents in it.

00:45:48Now the players suddenly think, wow, what is possible? I can basically

00:45:52walk around as Super Mario in some shooters, that is fun after all, and play it freely in the

00:45:58browser world. Microsoft has, as one of the first, ruled, the Microsoft lab is fully

00:46:03Browser world. Microsoft has, as one of the first,

00:46:09been ruling; Microsoft's lab is fully committed

00:46:12to this topic and wants to look at it very quickly, so that one can say,

00:46:15how can we now make game combinations from our games,

00:46:18that we have released, available much, much faster in principle,

00:46:24in order to create new things with players. So a lot is happening there. You have creators,

00:46:29who generate their own game worlds,

00:46:36so to speak. And I think the games industry will now also adapt to that very quickly. That is usually how it goes when the

00:46:40old software houses with normal software for normal people,

00:46:43and games were always a bit something else, are concerned. I

00:46:49think we will only see the cool ideas in the games industry, in the media industry,

00:46:54which can then be carried over half a year later, a year later,

00:47:00also to the normal software industry. So much for Sunday's word. First, what was the

00:47:05game with AI and people that you showed me recently? AI with people,

00:47:12I showed you many of them, I do not know. AI-with-people working and AI-with-people

00:47:16working was it. Perhaps it will come back to me at some point.

00:47:18That was simply because I wanted to check that again, sorry, the private question. At this point, perhaps, we just had it, when

00:47:23Microsoft is involved with it, it is still a question of licence, so it is

00:47:27not a call along the lines of legalising it, build it and mix whatever you want.

00:47:32Yes, so from that side, as I said, no legal advice, no call

00:47:36to commit crimes or anything like that. It is simply that it

00:47:40happens out there. I also think that what you said, that this is well rooted in the

00:47:45games industry, is certain. If you had asked me what I would build as a

00:47:50game, I would have picked up what you mentioned once,

00:47:54whether game interfaces do not become interfaces for AI systems, whether

00:47:59one could not say, take Maniac Mansion, SimCity and make

00:48:05an interface out of it for yourself, for Claude or something like that, I would have found that quite

00:48:11exciting. But before we talk ourselves into trouble, Jens, time cannot be reverse-engineered,

00:48:16as of today. We now have 49 minutes on the clock. I would say,

00:48:23we put a small button on it, perhaps a bigger one, perhaps a colourful one,

00:48:28a pretty one, whatever. That we can then perhaps also reverse-engineer.

00:48:33No, there is no such thing. Today we have done enough of that. We build that with the 3D printer.

00:48:36Never mind, you wanted something physical. I look forward to your game. I look forward just as much

00:48:40to the next episode with you. And Jens, at this point I would say, let us say goodbye

00:48:45to our viewers, listeners, whoever. If you enjoyed the episode,

00:48:50leave a like. Recommend us, tell others about it. The more people hear us, the more

00:48:55can share in our intellectual wishes. If you enjoyed it, tell

00:49:00your friends. If you did not like it, tell us. And at this point we wish

00:49:05you a nice day. Until next time. Ciao. Ciao, ciao.

00:49:09Welcome to Think Different. Think AI., the podcast by Mark and Jens. Two technology-loving

00:49:18minds who do not just talk about artificial intelligence, but they live it. Here

00:49:23you get clear classifications, real practical insights and a fresh look at what is

00:49:28possible. Understandable, critical and always with a wink.

00:49:33AI to think about, to smile about and above all to join in on.