Uwe Franke — „The Rise of the Solopreneur?“

Uh I I'm not counting success or fortune on on the buck. Yeah. It's the freedom that I can do what I want to do. And I simply felt not happy anymore by only managing a company but not building something on my own.

I wanted to get back deeper in the tech stuff. And that's what I have now. For me it is a is the biggest feeling of freedom what I have now because whatever I wanted to build in my life it's now possible. [music] Welcome to another episode of our new season of Beyond Vibe Coding partnering with Impala Search the go-to tech and executive search agency in Germany.

In this podcast, we explore the transformational change in software engineering and knowledge work in general. I'm Sebastian Haidider, CTO at Northio. And I'm Andre, CTPO at Trusted Chops. Great to have you back.

Um, this time we actually talked to a solarreneur who ran a bigger digital company in the past, but is now concentrating on utilizing AI as a solreneur. Yes, he shares interesting insights into his ways of working and how it feels to be the only human in a multi- aent company. Enjoy the episode. Welcome to the show.

Um for some of you who don't know um like we talked with Ingu Inguh in May um on this podcast where he um well presented a bit on the Aishhorse principle or where we talked about the Ahon principle and after the show he recommended to talk to you. So great to have you. Um and before we kick this off you may want to introduce yourself. Yeah, thanks first of all for having me here and thanks to Inko uh for recommending me.

Um also I know very good Ralph Müller Rim Müller who has been guest on this podcast as well. I actually I will meet him in an hour about after the podcast and yeah to to introduce myself um I started my software travel in back in 1994 at direct andaga bankang this times it was the first online program in Germany yeah we were building online progress software at a time when most people didn't even have an email address or internet access just just to to get an [snorts] feeling for for what crazy time it was. And then a few years later in 1997 I founded Biteworks a software company which I run for 27 years with focus on software for banks and trading platforms. So mostly uh the the stock brokerage market and I sold it a few years before to another software company and stepped out of the company in late 2024 and after that I decided to step into the business which I did had the most passion in already before when I worked the the last years in biteworks and that was the AI I topic.

>> Yeah, that's great. Um, so you just mentioned that like today will be a lot about solarreneurship. Um, and uh, therefore, um, actually also fits perfectly to the first point we are always discussing which is about the tool st. So can you give us a bit of understanding like what kind of tools are you using like what is your workflow?

How do you manage to be so productive? Okay. So, um, if I would need to put it in a oneliner, I would say my my team is a set of colorcoded terminal sessions and I run it by talking, not typing. That's the the the one thing my focus is conversational AI, voice AI on the one hand and I set it up to myself having using the the terminal with T-Mo setups.

So I have in each terminal a set of terminals again also to to see the the servers running and and that stuff as well. The whole thing is set up on a M5 Max MacBook so I can carry it around. Uh I have another MacBook staying at home running some other agent stuff and to have a jump point here as well. And I'm a little bit addicted to Entrophic because I like CLA mostly or the the most but I have subscriptions running on all the major providers also simply to compare the competitiveness for the the models and also run some uh open source models on the machine locally.

Other than that, my my setup is simply uh hosted the the applications and the staging systems by a German internet provider and having all controlled with Koolifi and the AI uh also the the rollouts and everything. my code is on GitHub and I think that's that's basically it is >> and on how many projects you are currently working in parallel and how to not getting like I don't know confused by all the changes updates and and and so on so how do you stay on top of all the stuff >> yeah so most of the time I am running between six and eight terminal session simultaneously And I sometimes it's also three or four on one project, but sometimes it's also six different scopes or projects by its own. As I say, divided mostly by colors because that's what I can mostly easily recognize. And I simply like to to see what the AI is working on, what the agent is working on or the agents.

Uh that's why I have a 42 in screen as my main screen and two/3s of this is covered for the uh terminals and one/3 is for the browser to to test and to see the results and on and on two other screens I have some some monitoring for for other stuff and my MacBook is my main input and communication the device on it. Yeah, that's that's it how I try to to cover it. >> And how does it feel to to be responsible for everything? Because yes, you're like you're operating a a fleet of agents.

Nevertheless, they need to get orchestrated, right? Someone needs to provide the context or is there some fully automated workflows where you actually don't need to care about stuff? I'm I'm I'm just wondering how does it feel to work on several different projects in parallel? Well, I think we are all used to that, right?

But like now within like a very compact workflow where you like you always get bombarded with then answer and answer and answer answer and like this happens like in parallel. I could imagine that's very exhaustive. >> Exactly. That's what it is.

Yeah. So I I really have to take care that the limit for the productiveness is not the subscription and not that what the agents are capable of doing. The limit is mostly me. I would say I try to work on a spec level with the AI.

I think it's somewhere between that what is called level three and level four I would say. So I sit down and try to specify the task or the feature what I want to do together with the AI very detailed. I have a a setup on my projects. So um with feature um bug tasks documents in GitHub where exactly is described what is all needed to be described before a feature can go into production.

And one of my um most important um lines in the prompt is uh that it needs to be on a 95% confidence before the AI is allowed to start working and that ticket template covers also tests. It also covers data sovereignity encryption and all that stuff. So sometimes I need to sit down between 1 hour and 4 hours only for the specification for one feature. But then when I kick it off then it really runs a swarm of agents and sometimes it takes 5 6 7 hours or even a day until the AI comes back to me and it's also described that it needs to do the end to end uh testing all that that stuff within as well.

And when I kicked off one task, then I can concentrate on next session and do the same and and so on. And so I try to keep up the productivity. But to be true, after three to max 4 hours, I need a break for one to two hours. Then I try to do sports and after that I come back with a fresh mind.

And I already start off like 600 6:30 in the morning to to use the time when nobody is calling me or sending an email or whatever. >> Yeah, makes sense. A short follow-up question like do you use any specific spec driven framework or is it developed on your own because there is I don't know getting done uh open spec kit uh no open spec spec kit >> I make a heavy use of the semantic anchors to valve again yeah personally I'm deeply into testdriven development yeah so I think that uh test which goes wrong first before we start implementing is is the best proof that something is is correctly speced and not yet implemented. And I I work with some semantic angers which I give to the AI and [snorts] a model which also says the which also tells the AI that if we are not confident enough then we are looking up for the state-of-the-art solution best practice and what top players in the market would do on our own and from that free perspective the AI needs to get back to me with free recommendations.

s and then I try to choose from that and going into that I also have prepared an framework on myself I call it launchpad which gives me the opportunity to directly kickstart an application and it has already in everything from P2 encryption multi-tenant architecture multi- language multicurrency multiate format all the things what I know from more than uh 25 years of doing that stuff for other clients. And that gives me an good advantage and I have a main agent which loss all my projects and another agent on every project which can also refer to the the main agent. And so I I really try to use the AI as team members as I would do it with real humans. >> Interesting.

Yeah, that that was actually one question I had before. If you have like for your different applications, if you're always using a similar stack, but that that uh obviously you you answered just now. And what you didn't answer was when Andre asked how many uh products you have uh in production. So ballpark.

Uh okay. So I'm I'm currently working on um two products or three products on my own. One is the the open-source uh product. It's the uh voice recognition for for the the Mac because I I used one paid before and then one client of me told me, "Hey, please don't use it for my project because we are doing sensitive stuff." And then I started writing it on my own and I also shared it on on GitHub.

And I'm doing one uh software as a service development as a trading journal for for day traders and swing traders. that's coming from my old area and I started it only for my own to be true because I was using an US-based trading journal and I was uh not happy with that because it did not support European progress and it did not report did not support Europe account and all that stuff and so I just started building my own and then I shared it to some friends and all that friends said okay that's Cool. And that's cool. And at the moment I'm just waiting for uh the contracts um which I also wrote with quad to be reviewed by an official lawyer and after that it will be uh started out as an own SAS service and um the other one what I'm working on is prickster.

That's a a Lego pricking system where my son had the ID for and I have one major client um who pays me for for the work. Uh that's that's the the part where I spend the biggest time on. It's a German IT company, a bigger one like with 35,000 employees um helping to enable them to jump start into AIdriven software development and handing them over more and more and beside that I have three to four smaller clients which partly are already ready developed and I maintain and implement new features or where I help to enable them to build the software with the help of AI on their own. >> So basically I don't know a handful of actual products, right?

And then additional like capacities. Okay. Yeah, that's a lot. Thank you.

>> Absolutely. >> How does it feel com like compared to the old days where you ran a software company? So now being in charge for everything I could imagine that is like that is nice but also has some I don't know downsides. So like how does it feel compared to to the past?

>> Oh that's that's uh very interesting because I was reviewing myself actually only a few days before to to get back the the old feeling. So there are as always there are pros and cons. The the biggest pro for me is I was all the time I was a technical guy and when I founded the company I wanted to build real cool software especially for banks and brokerage and after the years passing by uh I recognized that I wasn't the tech guy anymore. I was more on sales, on regulations, on reading contracts and managing the the the team.

Yeah. So that the company always have been like too small for having an extra management line but too big to handle it on your own. And the the other thing is I'm I'm interested on money. Yes, sure.

Everybody is interested on money, but I'm not interested on on uh I I'm not counting success or fortune on on the buck. Yeah. It's the freedom that I can do what I want to do. And I simply felt not happy anymore by only managing a company but not building something on my own.

And I wanted to get back deeper in the tech stuff and that's what I have now. For me it is a it's the biggest feeling of freedom what I have now because whatever I wanted to build in my life it's now possible and I really I when when my son came up with that idea on can we can we build threeescribed 3D models of of Lego bricks with the plans how I need to stick it together from a picture or photograph or from a given set what I have and I had I have personally I have no idea I have no intelligence in that how a Lego model is built and so on and we I simply started discussing with the AI and I I saw that it's a complete different domain but I could build it because I could imagine it and I think that's the that's also what I tell everybody who wants to get in touch with AI. I say don't ask can I do it ask the AI please do it and the AI will tell you what it can do and what it cannot do and you will be surprised that almost everything that in software is possible even if you cannot believe it right now it's possible you can do it and that's the biggest thing of freedom to the downside it's very easy I need to manage my time um because the next feature is only 10 minutes away and that's something what I really feel that I get sucked into it like it it's been when I was a kid and I was doing text adventures. You look on the watch and say ah the next level I can check on to it.

And the next time when you look on your watch, it's 4:00 a.m. and you are still completely driven into it and you cannot get away from the machine. That's that's a very very big thing. My son was really coming to to me and say, "Yeah, Dad, you sold the company and one of your reasons was you want to have more time for sports, family, more free time, and so on.

Now you're sitting in front of the computer like 16 hours a day and what the hell are you doing sir? Why? And I say and then I tell him I have so much fun. I have I really have the biggest fun in my my life because I'm free to do everything.

I'm I'm not an CI/CD specialist to deploy things but I can do it. I can manage it. that I can manage with the AI to set up things like Koolifi to deploy all the stuff automatically to have the monitoring and all the stuff where I needed tons of people before which told me oh it will take a week and now I see it can take two hours to to that that's the big difference and the other very big downside is aside managing time is talking to humans that's what I'm really missing I try to make two to three appointments a week to get out of my self-made cubicle um and not to end up as a whatever strange guy coming out from the cellar uh and have not talked to human people for three weeks. Yeah, that's the the biggest downside.

what I really see um that direct exchange going to the office talking to the people face to face it's a complete different thing you cannot you cannot get this with AI >> thanks >> yeah would be bad if you could get this from AI because then you really wouldn't need other people right >> I think a lot of people think they can get this from AI and they think that they can even get it better from AI than from humans >> that's also too. >> Yeah. In regard to impact, so how would you how would that you mentioned earlier it was not really a fulfilling life so to say just managing a company. Um nevertheless at the end of the day right you need to sell something to pay the bills.

Um I think it's the way you described it. I I like the idea of of freedom. I like uh the idea of of power. The question to me is is how do you see this from an impact point of view?

Right? Like so you can work on a lot of stuff. Does it but does it pay off? I I could imagine or like I remember when I think it was Sam Orman said like there will be a $1 billion company run by a single person.

Um so do you think that is possible that is something like you should strive for or like yeah what's your view here? So the first one, the first question, can a one person company really become a billiondoll product? Yeah, I I think that was the question what some old man answered and at the point when he said that I was shaking my head and said I don't think so. But in the meantime, I generally believe it can.

[snorts] the on the state of today one person to a billion dollar product is possible now but I think there's a condition that the people skip right over it will not be possible for everyone only if you were already a professional in that field and if you have the the passion to go deeper than everybody else then it's possible. So when we talking of software products because that's the the area where I can talk about I cannot talk about medicine products or cars or whatever because that's not not where I have some expertise in but looking on the on the software field SAS services or whatever I would say someone with no real software background simply can't operate AI as put as effectively as a pro can and at today at today's state of art I think in two to three years that's a little bit of the dystopia what I had from the club code conference I'm not I'm not sure if I would repeat that sentence yeah I can I can imagine that it then will turn around um but what really helps me is when I talk to AI and when I specify the things that I spent three decades on going the classic route. I have been a junior developer. I have been a senior.

I then turned into a team lead. Then I had the architecture ownership. After that the the software architect and finally the solution architect. And as I have been also the the owner of the business, I had a an additional qualification I would call it.

It was always paired with the understanding what the customer actually needs, how they make the money and how to manage a room full of stakeholders. And that's that's an a part I not sure if I can do this nowadays. And um I think that's exactly the judgment the model still doesn't have. And uh the stakeholders must not be humans must not be physical persons but it's also something like GDPR like in in voice AI we have in in Germany especially that voice is biometric data and has the highest responsibility for someone acting with with voice and all that things and so yes it's possible not from Germany I would say nowadays and Not if you are not an absolutely pro in the area where you want to test it.

>> Not sure not sure if you want to uh share that or if it goes beyond uh but I'm I'm wondering if if you're a soloreneur and like you basically do consulting like so I think this is one of your uh jobs is like how do you sell that? So, how do you find the right price point? Because like I don't know I I I would be I curious to hear like how companies think about the price if they realize it's just you, right? So, because at the end you're operating a fleet.

So, like what is an agent worth? Is it the same as a consultant back in the days or how do you how do you do this in take this in consideration? >> Yes. So, so I have two different uh models what I'm selling.

The one is really workshops on introducing for example voice AI or conversational AI helping on the setups explaining what needs to be uh done uh from legal to software stack and and also with sample prompts and so on. This is simply where I put on a on a price tag which I know also from from other people in the area. Uh [snorts] I would say that's that's like a like a good lawyer also Texas's work that's this this level and I have longerlasting projects which is not really consulting only. This is more about de developing an MVP start for a software product and enabling the company and the people of working with the AI together on on themselves with that later on.

And this is a mixture of value and time. So it's and it to be true it's it's not on everybody the same. It's also a little bit how interesting the stuff is. That's that's it.

But I I don't I don't have a a a price tag like it must be exactly that um sum per hour or per month or or whatever because uh also my clients know that I have several projects running and they give me the freedom that I can share between this and um yeah >> thanks for being so open >> and to be true I try to find a way that in 3 years I will have the opportunity to be free from uh outside contracts I would call so that really I can uh earn the money what I need for a living out from the company itself. So example with the trading channel or with the the legal service um um as a stock trader for 30 years also some kind of trading has some influence on that because I'm I I really make my thoughts on on the point what I saw on on that code conference in in London. If the speed keeps on going up that the same speed as it is right now, what will happen in 3 years with the knowledge what I have? I'm aware of that that this is coming but still I I make my mind up.

Will my knowledge be needed in still in 3 years or not? because the AI is then probably so much ahead of that what we have today that I'm not sure if somebody can simply put in a prompt and say hey make me this up and 4 hours later the AR is back yeah do you have an domain which I should point to the solution >> yeah I think we would cover that topic at the end uh of that episode um no no all good all good but this is already a nice cliffhanger Yeah. So maybe uh let's go already a bit in the direction before we talk about uh hottest um and um and and that stuff. I was wondering is there anything where you like improve your organization?

So you mentioned earlier you try to well like use agents in a way as a classical organization. So you do some kind of organizational development. Is there some stuff where you improve where you say, "Okay, Andre, if we meet next time, I have another 42 in um um um monitor because um like I I like I expanded my company right now. Now I'm running [clears throat] uh 20 um agents in parallel because I learned like I I introduced the middle layer also by agents.

So is there you have any plans in that direction?" Uh so actually every Friday afternoon is the time when I make some let's say sprint review with my team and we see what has changed the the last week and this is mostly about what new skills or what changes in the skills we had in the last week what changes in the claw and p files we had last week and all that stuff And I I tried then to go from bottom up to check for duplicates. What should be in the in the master cl and def file. I have a structure of of cloud and D files from the the main level to project level and sometimes even in a in a sub [snorts] part of it. Um so so my my cloud hierarchy is the the really the main setup is in the main cloud MD and in the in the project files itself it's domain specific and I try to clean it up on every Friday.

I also use that review function of claude a lot which give me hindsights on where I should get more productive and so I try to refine my work but I also have running for example uh for marketing stuff a domain where I have no idea on yeah I'm absolutely not a creative person um [snorts] so in in things of making advertising and so on I have outsourced it to a paperclip instance which takes care of that. This is looking also for my competitors mostly in that trading channel uh thing and Claude is managing that paperclip as well and so I try to extend my my team because the what I found good on the on the paperclip system is that it runs as a company and I hired this as a marketing company to fulfill my needs. So my target is really to run a fully automated marketing system with paperclipip. I also made up a new credit card um with spending small spending of €1,000 per month for now which can be used by it.

Test the water before you go in deep. And uh that's mainly the way I try to um double or triple my team size by finding special agents and things what I can automate even more. I also have a company dashboard which is driven by uh cloud co-work which checks my bank account for pay bills, open bills um which is coming from the trading account which amounts and so on and how we move on to the target to get 100% independent by end of 2028. H >> and it also checks my calendar ask me how we want to structure the day to be most productive and uh for this is important to know that really every task has a ticket for me in GitHub even organizational tasks and we do a triage every day in the morning on which task needs to be done today.

So, so I I'm I'm >> impressive. >> I >> I I have a lot of control on my my workflow and I also need this this kind of control this structure in a in a Yeah. And I think also makes sense to not reinvent the wheel but depend on or like like make use of something which is well known right like so organizational structures like like roads um and these kind of things maybe to not sure if everyone uh knows what cap paperclipip is about maybe can you summarize it in one or two sentences. Uh, Paperclip is an AI agent framework.

Maybe a little bit also similar to OpenClaw. Uh, but not on the the same level of being independent, but um, with with Paperclip, you create um a company s and you only talk to the CEO of the company and the rest the agent starts managing on its own. So when I told him I I want to research uh link it in on my own posts which worked best and uh to give me new topics where I can write on and to prepare this as well. He uh he made a contract with an LinkedIn in professional assistant or something like that and he also get an researcher on other trading journal companies to see what the market uh does there and it works also ticket based which is very friendly for me and yeah the organizations that grows on it own only where you have to to watch on all the time is how your tokens are spent and with the also yeah your money is spent for the AI but also different models and also open source models.

Yeah, I got to know uh paperclipip some months ago after I like tried to build up my own organization and then while I researching I found out hey there is a huge uh huge um project I think it's a very well known in the uh in the open-source space and this is maybe how um maybe maybe final question how much do you make use of these kind of opensource uh products I think they like you use it on purpose to leverage um yourself your solarreneurship >> or you you tend to build stuff on your own very precisely for your needs. Sorry. >> Um mainly I am really the second part. Yeah.

It's it's hard to me to pass by to build something on my own because I simply have fun on that. Yeah. So I also a lot of sports and um I also have my my own sports coach which optimizees my my training stuff within my week plan to get the optimum out of it and I found it fascinating that I really can build it on my own and that it more or less really in an area like 90% exactly matches that what I get from professional training splans or from human trainers. I had an a human person as as trainer before but that person um really wanted to push on its own training plan and simply not really recognize that for me work and family is first and I do not earn any buck from sports.

So it's not my priority on that. I I want to be competitive but it's like my third priority, not my first priority. But I also t all the time I tend to build things on my own. And I really need to control myself to not to do it.

And the paper clip thing was intentionally I heard of it and I played around a little bit with it and then I wanted to rebuild it by myself but recommended it to a friend of mine and he then came back after two or three weeks to me and said hey that's that's great or I can do all with that and he exactly uses it for marketing and then I had a session with him and he showed me what he has done and he has abs he He is not a developer and not not nothing. He simply used it as the tool as it is and I was really fascinated by that and then I said okay for that marketing stuff I really need to rely on on paperclipip and I need to stop wanting to doing it by myself. Thanks. >> All right.

Shall we uh slowly but surely get to the last lap basically? >> Yeah. Reality check. Right.

So um as as we said back in the days high so the hottest things uh and for sure also the what the moments like can you share some do you want the the really actual uh one sure fable the disappear of fable and it it's really funny you can look on my my link it in timeline [snorts] on Uh I think on on Thursday or Wednesday I posted something about how important is it that you own the the stack and that nobody can put a kill switch on you and exactly what the later fable has been turned off by the US government. Yeah. And the crazy thing about this it is for now it really hit all because entropic say so um like the US government in intended it that it's it should not be used by non US people it's simply not possible for them because they cannot control uh when somebody is employed in US company if this is an US resident or not. So all the people have been hit by that decision.

But if my business would really rely on fable as this model, it it would have been a kill switch. Yeah. >> So this was the the absolutely not wanted proof on what I wrote in that one post. And that was really the the biggest what the moment right now in this these days.

In this case, it's not not a technical one that I was surprised by the result, but I was really hard hit that this can happen that this what I never expected that this can happen that this can happen and really from one day to another. I woke up, looked at my terminals because I had some jobs running in the night and they all stopped at the same time because I was using Fable for doing it. Um, and I was what happened? Did I had an internet outage or whatever?

And then I I opened up and the first message was Androphic stopped the model. >> Yeah, I think everyone is doing everyone is doing risk assess lot of larger companies are doing risk assessment. I usually say, right? Um, so what happens if like the access is is is denied and then usually I think the the comment there is likelihood low um or close to zero, right?

Or very very low and I think after Friday or after Saturday we all learned we may need to recalibrate. Um not sure if ZAS and I would do a podcast on that but I think that is really a tough thing. Um actually I at least I do not have a simple answer how to react on that but uh you absolutely have a proof you have a proof on the post before uh so congratulations [laughter] >> nevertheless congratulations >> congratulations >> yeah I'm I'm just thinking like everybody is saying okay this is a wakeup call now right I mean how many do we need pressing the snooze button again I'm pretty So uh let's see and uh in regard to the uh the hottest uh so what are you uh giving a try at the moment what is like really fascinating you can you >> yeah the the what's really fascinating for me is right now that's what I named right before that usage of paperclip and the the one thing where I have really no competence in is is marketing and I had some dry runs right now and uh when I see what that agent also produces on creatives and also summarizes me where I have to go in for for marketing to to sell my subscription service uh for the trading journal. It even brings me up that there is some guy sitting in Switzerland using an US-based journal and he would be a good influencer if I and if he should make up an email to him to get him using my journal and to having as a a leading person there and brings me up to complete different business models what I had before.

So my my main model was driven by a SAS and then um he came back with a research telling me um maybe it's not not the the what you should go for. You should can you should bundle up with uh brokers and they pay you a success fee when somebody opens an account. And this can be like uh on a on a one fee on a on a one performance fee could be like three to four years subscription fee. So it's an model which I did not had on my mind and also what I have seen on the the picture images creatives uh what it reproduced for me it's really cool and it really touches the the fine stereo touch group and so on.

Uh this is really interesting because it really um um yeah gives a contrast to that what I said before that you need to have the core competence in that what you are doing with AI because it shows that I absolutely do not have competence on marketing that's something what I it's it's not my thing yeah I also do not like to post on social media but I have good link in posts because I I learned it and I tried to step outside uh a little bit. But this is really the the for me the hottest to get into it. How I can make more use of that helping me also on other things where I have deficits or which I do not like to do and to scale my one person army. [laughter] my Swiss knife.

>> Good. I mentioned earlier like um so there will be we will also you will talk about the prediction. Uh so maybe now is the time. Uh so where does this all lead to from your perspective?

>> Okay. Um the the prediction it's al so prediction for for for which time frame or overall prediction? I let's say in like short so I got I worked with some consultants back in the days when I was at DHL so they always said like Andre explain it short mid and long term uh so [snorts] >> okay uh okay so so so shortterm let's see if you think on shortterm like like two to three years I think that the within the three years the software sector is the first industry hit hit by a full force. So I would say if you are a mid-level developer who is not passionate about software development and who stepped into that industry because there was a good salary to earn, I would say run as fast as you can.

uh learn on something else maybe how to cut bonsai trees in the favorite way because this will be not be replaced by AI not not in 10 years not in 20 years why because it's a passion of of of people and not a business thing but all the the all the not passionate people in the industry will simply disappear uh you don't need it anymore And that's that's a predterm prediction. I think on on the long term of view, we are probably will also need to find a maybe a new social contract for our whole society because work and and life will will change. I I'm not sure if we will have enough work for for all the people at least not on a industrial intended way. But if we manage it to to make a change there then probably our social life um will get much better than it is today.

But it's it's a challenge and I think that wave is coming much much faster than everybody expecting right now especially when robotics and AI get on the next level >> that that's a topic we have debated on this podcast before and there's two sides to the or two perspectives right so one perspective is the I always forget is it jeans paradox or anyhow so like any >> any new technology ology in the history, right? be it the mechanization of um of sorry agriculture or be it the steam engine whatever >> did not lead to less work but more work because of the rebound effect right [snorts] and it's like to me it's really still open how this will end up because ultimately what what you have and and I think you described it nicely um at some point as long as you're doing or creating products and services for humans. And the models are not like perfect, how do I say it, digital twins of humans that that have taste, that have the the emotions, that have like a world model of of humans, if you wish, right? Then it will be hard to do this fully without humans.

Um, but I mean ultimately, yeah, the the goal I think back in the I don't know 30s or 50s already was like at some point we only need that what was it 15 hour work week or something or maybe even less because uh everything is automated. Let's see. >> Now, now our chancellor talks of like 60 hours work weeks to get Germany back again. But anyhow also yeah I heard on a yeah event I have been some weeks before I heard interesting information.

I I'm not sure if it's right or not right, but it was said that we already like skipped the highest point of population on Earth [snorts] and that it uh it's only Africa which still has a growing population and all the other continents have already an steady red in the shrinking in the population and for example if China does not makes a turnaround in in the the birth rate it's expected that already in 2070 so like 50 years from now uh they will be back like to 550 millions. So then anyhow it it solves the problem of maybe having less jobs. Yeah. Because [laughter] also will be less people and that's also an interesting approach.

But uh yeah, >> it's true. Just for completeness, >> yeah, just just for completeness, um population is still growing by 0.84% at least from last year to this year. But the growth rate is declining and it's expected to stop at some point and then pretty much turn around and then growth will be negative. >> Mhm.

>> Yeah. Thank you for sharing. Um let's see if it turns out. So we also need to meet this uh this is the good thing.

Um so looking forward uh to meet you again. Maybe not in two or 3 years, maybe earlier over um and until then much success with your solarreneurship. Thanks for having you. >> Thanks for having me and that was really interesting.

>> Thank you. All right. Now for the recap. What stuck with us?

Well, yeah, from my point of view again, right, uh, three things. First thing for me is, uh, was interesting to hear about his workflow. Um, and I think it's mainly not 100% related to solarreneurship, but the way he's working to really um, accomplish a lot. Um I think that was interesting to hear because at the end I think we all know this but it's about compacting things right so um and um yes you can compact a lot of things in with short cycles but then I think you get productive maybe 100 150% on top but if you really want to leverage AI I think you need to go for long loops and that was actually for me a nice reminder um and I think he is mastering that quite Well, at least that's what it sounds like, right?

And it reminds me of how Boris Churnney and others describe their ways of working where there's an extensive planning cycle, I would say, or session in the beginning, right? And then the agent has a lot of contextual information to then go ahead and implement the the feature whatever. Yeah, it makes a lot of sense. Yeah, I can also like I can confirm that from my side whenever I'm uh doing agendic engineering it like I I spent the majority of the time um on the first phase.

Yeah. Um so but the second topic is actually like that is a it's it's actually in the same area. So what I also notice and what he what Uber confirmed is um how exhaustive uh it is u after 3 4 hours he needs a break for one or two hours I think um is maybe also something we can explore in the future with another guest. Yeah.

Yeah. And it really reminded me of Steve Yaggi's article that I didn't I don't know if we discussed it in the podcast but shared it on LinkedIn once where he writes about the AI vampires sucking our blood or something but he also writes that it's very exhaustive and that he also needs to take a break after 3 to four hours and even thinks about should our working days not h only have 3 hours in general and the remainder of the time should for socializing and for more yeah collaborative work instead of this really high intensity brain work I would say right and one thing I also took away from the podcast or when when he said that he's utilizing the agent 16 hours a day I felt I'm not sure if he actually nails the balance right now I can understand the the FOMO actually right because now's the time if you wish. But I personally have dialed back a little bit my AI hours to yeah just improve sleep and and relax a little bit more because in the beginning it was quite intense. Yeah, I that's something I took away.

>> Yeah, absolutely agree. I think especially the point on 60 hours I think maybe we will also see some kind of the same what happened to social media, right? this kind of addiction then also like the way you handled it I think it doesn't sound really healthy. It's like it's I I can also confirm from my side it's like sometimes it's like it's pure fun right like really pushing through things.

Um but overall I think in the long run like always having mind this like life is a marathon not a sprint. Not sure if 60 hours is is is healthy. All goods are three right. So my third point maybe also to a recap on solarreneurship.

I think after the talk now with Uber I think overall we got a glimpse what is possible right how you can leverage agents like more or less I think from a setup point of view I think it's possible right I think it's possible that a single person can run a large business honestly I'm unsure if that is if if that is enough to run a billion dollar business I think there There is still I would also say a long way to go but I would say it's also not impossible. What's your take here? Yeah, I think that Uber probably he has a background in tech, right? So he knows how to build software and he can build these solutions and he can teach others to to like engage this way and learn and adopt these techniques, right?

I am less sure about the validity sorry validity of his marketing workflows right because yes he might be impressed but probably I would also impre be impressed while a person who really understands marketing might be less impressed by what his marketing team uh creates right though I haven't looked into I I cannot say maybe it is great what the the the the agent team that he's running is coming up with right. It's just I've seen this over and over again that I personally created something I was super impressed and then I like showed it to someone who knew their in this >> in this uh domain and yeah they were less much less impressed. Yeah, >> but nevertheless, maybe also a reason to invite him. Maybe half a year, a year to see how things are going, how far his marketing marketing campaigns took him.

Fair, fair point. Fair enough. Yeah. All right, cool.

Well, then see you later or see you next time. >> The Beyond Vip Coding podcast is a project by Sebastian Haidam and Andre Noau partnering with Impala Search. The content is created by us and our guests. Join the discussion on LinkedIn or visit our website where we publish all episodes.

For questions and inquiries, feel free to reach out via LinkedIn. Thank you for your time and see you in the next