The Bootstrapped Founder

Arvid: 00:00

We live in a world where new technology is booming. And those booms are heard loudly, like this AI explosion on everyone's mind lately. Developers, entrepreneurs, and more, we hear this. And it's hard not to want to join the AI movement. Many of us just worry that if we don't start using AI or building AI businesses, it will leave us behind.

Arvid: 00:24

This episode of the Bootstrap founder is sponsored by acquire.com. More on that later. I want to discuss the straight line bias today. This fear of things slipping away from us and not being able to catch up. If this tech keeps developing as fast as it does right now, will I ever be able to catch up?

Arvid: 00:42

That is a question on my mind and the minds of many founders that I've been talking to recently. Over the last couple of weeks, I've realized that as tech enthusiasts, we are at the heart of every new development. Right? When something new comes out, we know about it right away. We follow the right people on Twitter and we see any new development within minutes.

Arvid: 01:01

And this often creates this sense of urgency. When we encounter something new that we don't understand, we dive in. But information about new things usually is scarce and maybe even incomplete. It's kinda hard to fully grasp what's happening out there and where it might be going. These new developments point to an unwritten future that even the experts that make it happen can't really foresee.

Arvid: 01:26

But we can dream, and we do dream. We tend to have these high expectations for technology, which occasionally, I would say maybe most of the time will be overblown. We expect this kind of tech to keep growing and to keep surprising us with new magical features forever and we forget that most tech eventually matures at a slower rate. This slower pace usually allows mainstream users to catch up and integrate it into their lives without feeling overwhelmed. I think ChatChop Tea is a good example here.

Arvid: 01:57

Right? It came out, it was only for the nerds, it was like in in tools like copy a I, and and we didn't even know that Chat GPT was behind it or the OpenAI platform, and then they released Chat GPT 3, and people really jumped onto this. And then 4 came out and things have slowed down a bit, people have started integrating it, people are starting to use it, people stop to use it, like it is becoming more of a norm for people to think about chat GPT as a potential solution, like Google is for searching, chat GPT is for answering questions. So there is a mainstream appeal and that has slowed down development. Even a recent podcast interview with Sam Altman, he said that this year, they might release something but probably not Jet GPT 5 or GPT 5 to begin with.

Arvid: 02:43

So things have slowed down a little bit. So why do entrepreneurs and founders experience such strong FOMO with AI? This fear that often leads us to over invest in choices involving this new technology that may not be the smartest choices. Honestly, I don't really know, but I'm willing to explore it with you today. I think AI is a very recent technology that has become super popular, and feels that if we don't jump on it now, we might lose out to everybody else around us.

Arvid: 03:13

And people around us communicate a lot about this. I do the same. Right? I talk about the AI systems that I use every single day and maybe that creates a kind of presence of the theme of AI in people's social circles that is much much stronger in perception than it is in reality. I'm trying to think of other recent technologies with a similar impact like probably no code might have been a while ago, or building this multimedia personal brand.

Arvid: 03:38

It's not the same, but it is also something that makes you feel like if you don't do the same as others, you're left behind, right? You're doing something that is unoptimized, where others optimize for a thing and they are better at it so they will kinda push you out of the market. But honestly over time novel developments in all these fields eventually slow down and become more research and evidence driven and stable. I think online payments is a good example here as a technology. In the beginning, there were so many experimental providers for online credit card payments or billing, invoicing, these kind of things.

Arvid: 04:09

But over time, this has consolidated into something that is much more stable and reliable. There are standards and there are expectations. Payments now rely on the bigger players and the connection between them and developments in this field are much less frequent. If you were one of Stripe's first customers when they were still called slash def slash payments back in the day, you might have had a harder time integrating certain services or the even the APIs because they changed a lot. Now that has changed over the last couple years.

Arvid: 04:40

Stripe's APIs are to in in my perception probably the best most well documented and well structured well versioned APIs you could possibly use, but things have changed. It's becoming a much more mature system. And I believe that in the world of AI, we're not there yet. We're at a point where everything is still super new, constantly changing, everything is in flux. So let's discuss some things related to AI, and how they might keep accelerating or slowing down, because I think most people would like to know, well is this going somewhere?

Arvid: 05:13

Can I invest in this now or is this gonna change a lot or is it at a stable point? One big aspect of AI that I've been working with and experiencing lately is the capacity to own the technology. We will likely continue to see big language models like gpt 5 coming up. Right? Or whatever new thing anthropic comes up with in the future being deployed on massive clusters of machines.

Arvid: 05:39

It is really massive, like the both the technology behind it and the price is quite bizarre. Nvidia recently announced their b200 graphics card, which costs between 30,000 and what $40,000 per GPU. That's one single GPU there, that's such a high price that is kind of comparable with a vehicle. You can buy a car, a truck, or a GPU, quite mind blowing to me. And those extremely powerful graphics cards when put together in clusters, well, obviously they allow to those companies to run these very advanced language models in the cloud and then give us a little tiny slice through an API.

Arvid: 06:18

But at the same time, and I guess in in wonderful ways, more and more people are using these large language models and a different scale on affordable consumer hardware in local settings. That's something that I've been doing a lot in my recent business ventures. I try not to use API systems at all and try to run all my AI stuff locally. It still runs on a server with a GPU in a data center. So that local is not local in my house, but it's local on a machine that I have control over.

Arvid: 06:51

Local on a machine that I can take a complete disk snapshot of and then reinstall on another machine. Like, there is more control over this. That's what I mean with local AI. And if you visit huggingface.co, a website where a lot of these models are hosted, a lot of people train their models and exchange information about them, and you see which models are being released on a daily basis. You will find that hundreds of language models are updated or tuned every single day.

Arvid: 07:17

They become more specific and they improve and you can download them and run them on your own consumer hardware. That is new. That is new in terms of technology. The availability and accessibility of this tech is becoming distributed. And while those things may not be as capable as GPT-four is right now or GPT-five will be or models like Claude from Anthropic they don't need to be.

Arvid: 07:39

Local AI is often good enough for certain tasks. You just have to know how to get it to that point, how to measure it, how to evaluate it. And big models, like the GPTs, they rely on complex infrastructure as well, something that we might not even want. If you go to chat gpt, or you use an API for large language model like this, the platform actually had internally runs multiple calls to its own back end to provide the best possible answer, and to scan the Internet, and to run code, or whatever. For Local AI, that's quite different.

Arvid: 08:12

It's more like a client server situation. You send a request to get a response, and then you do with that what you want. You can build this functionality, but it's not part of the initial platform, and it becomes something you build on top of it. But I think that's okay because we can achieve whatever we want at a reduced capacity, but it's in our control and on our own hardware. And I believe we'll see AI moving closer and closer to edge computing devices, like our phones, our laptops, and the personal servers that we run.

Arvid: 08:41

By default, these devices will eventually be able to run AI systems. They will always be able to do this, but you see this in Macs already, Apple Computers. They have neural engines and they have the the core ml system that allows you to run like in inference based, machine learning things on those things already. Like they built in chips to be able to do this. They they it's it's an infrastructure thing that Apple Computers already have.

Arvid: 09:06

And, I think we will keep seeing this on all kinds of devices. And I know that many businesses, many AI tools out there currently are rappers. They rely on these APIs provided by bigger companies, but I think we will see this shift towards local AI, owned AI rather than rented AI. And it's there's always this risk. I I talk a lot about platform risk in many different ways, both in terms of audience building and in terms of business building, and here is one.

Arvid: 09:34

Like, building a wrapper business around an AI is obviously super risky because if somebody else runs the AI, they can also turn off both the AI or your access to it. So local AI has that benefit that even though it's still on a platform like AWS or Google Cloud or whatever thing you use, you at least have the capacity to shift these things around. You're not locked into this one API system, a that cloud out there, that one thing. Right? And owned AI, you locally run stuff may not be as up to date as rented AI is.

Arvid: 10:07

But it's a personal instance. It's private and it's much more secure because you can run it inside of your own systems without sharing data with others. Or having your input used as training data for future versions of the AI, which is exactly what happens when you use a public API like chat gpt. Your data becomes training material for new and improved versions of the software, Which immediately makes it unusable for many corporations, for many businesses, right? And a business like a big company will never use their internal documents and send them over to Microsoft's servers or whatever, that's not gonna happen.

Arvid: 10:41

They need a waterproof system that will keep their data internal. So local AI systems facilitate this. And that's an important thing for you to think as an entrepreneur, that certain expectations cannot be dealt with if you use these kind of rented systems. Alright. So what I see as an interesting development in this space is this move towards the edge.

Arvid: 11:07

It's that AI will move closer to us, just like devices have in the past. Right? We've seen this trend in technology over the last like 30, 40 years, where technology moves closer to us as physical beings. First, we had these room sized computers. I mean that's a bit longer ago than 30 years, but you know what I mean, like these gigantic machines.

Arvid: 11:24

Then they turned into desktop machines that we could sit in front of, Then laptops that would, like, be right in front of us. And now mobile phones that are in our pockets or tablets that are right in our hands and wearable devices. Now AI is on our wrist or sitting on our head in these weird Apple Vision Pro things. So it's moving closer to our physical bodies. And I think AI is gonna follow the same path.

Arvid: 11:47

Currently, it's in somebody else's data center. Right? It's an OpenAI's API, but soon it could be in our own data center or maybe on our personal devices. Apple is already working on local AI systems for the iPhone. The chip is already in there and the LLMs are in production.

Arvid: 12:03

So as AI continues to advance, this is gonna be a move towards the person. And since we're talking about people, I also see another interesting consolidation here. That's human AI interaction. And don't forget that we are talking about the idea of not missing out and what to do and what not to do in AI. I think this is one of the most important parts of this conversation because it has to a lot to do with the the connection, the relationship that we build as entrepreneurs, as creators with the people that consume our products, our customers, our clients, our audiences, everybody we interact with.

Arvid: 12:38

And right now, AI is often used as a tool. Right? We might ask an AI to help write an article and suggesting a title, drafting paragraphs, that kind of stuff. But there is a movement towards AI being a mediator or even a companion. Many people are using AI as a part of their creative process and not just for input output tasks.

Arvid: 13:01

Eventually, there will be there will have to be I guess a balance between back end AI systems and and those that interact with humans, right? That that will that will happen. I'm gonna explain what I mean in a second. Simple tools or some tools will simply process data and produce results. If you look at at OpenAI Soarer, their video engine that can generate videos, you're gonna feed it a full movie script, it's gonna create a movie.

Arvid: 13:25

That's how it's gonna happen, right? This is the kind of tool approach to AI, But I believe that many people will have an AI co writer or a research assistant to collaborate with in real time. The magic here is not just what Sora is doing with the video creation. Even writing the script in the first place will be heavily AI assisted. A lot of writers out there and writers of any kind, right?

Arvid: 13:47

Authors of fiction books, nonfiction books, journalists, people who do a lot of copywriting, people who do a lot of social media interactions. Almost every single person is using or leveraging AI at some point in their journey. Not everybody uses it to create results, but a lot of people use it to help them in their process. And anyone working on complex things will by default want to have an AI companion over time. I think that's where the balance is gonna be.

Arvid: 14:16

Researchers, developers, they will benefit from having an AI partner to help them with decision making and data analysis problems. I already do this. As a software developer myself, I work with an AI codeveloper every single day, like hours every single day. The AI systems that are currently in place for PodScan probably know more about the software than I do, and are highly capable of building features, integrations, and tests. I can do so much more with my AI companion in less time than if I did it alone.

Arvid: 14:46

I think if I spend 8 hours on something all by myself, you know, I could get it done, but I can also do it in 2 hours with the AI, and that is massive in terms of time savings. And to many of us, this is a big threat. Right? To a lot of developers, they're quite afraid. We there's been a lot of talk about AI taking over our jobs, especially with AI companions like Devon that was recently released that is a fully autonomous agent.

Arvid: 15:11

Now maybe I from what I read it was like successful in 9% of cases, just not too much. Like, 1 in 10 ish, I don't know. But still, the the fact that there is an AI system already today, which is always gonna be the worst it will ever be from now on. Right? It's only gonna improve over time.

Arvid: 15:29

That shows us that coding, the process of building code, may not be the thing that we as developers are gonna have to do more of in the future. We might have to, or want to, or be able to do less of it and do other things like architecture or system design. But honestly, I don't think AI will replace us because of this. It will become more like a companion that helps us in doing the actual work that needs to be done. AI will be connected to us and our devices, and will be involved in every professional process that we facilitate.

Arvid: 16:00

And it'll be more than just a tool, it will be a constant always on 247 companion that has these tool capacities, but also has reasoning abilities. I use AI a lot in my own business, but here's the thing though. Right? I don't sell AI services. I don't sell an AI.

Arvid: 16:16

PodScan monitors podcasts for brand mentions. That's the basic function of this tool. And while AI makes this possible, I still sell notifications about people discussing brands. AI is not what I sell. It facilitates human connection, and it always should.

Arvid: 16:34

AI is becoming the norm in many industries like customer service, and often the first interaction between customers and businesses is this AI system that filters or sorts complaints. It's a moderating task, and it determines like whether communication with a human is necessary or if an automated reply will suffice. And because it's easy and practical, I think companies will keep using this. AI will continue to be integrated into our daily lives in these capacities and processes at work will be facilitated by it being a helpful companion. But it's not gonna replace human workers because as AI has become the new baseline acting as this first point of contact and a judge of relevance for human communication, there is a little problem.

Arvid: 17:15

I think it's essential that AI is understood to work best as an addition to its service, like customer service, and not a full replacement for people. Because the goal of customer service is not just solving their customer's problems. That is a main goal of it, but there is an underlying goal here. It's about building relationships with customers and your brand. It's it's strengthening trust between people, and relying solely on AI based communication actually hinders that potential.

Arvid: 17:43

But if you wanna focus on human interaction, you have to be a human. You have to actually communicate like a human and to be a human also in the perception of the person you're talking to. And that will make you stand out in the world that is dominated by AI systems because they're cheap. Think about robotics delivery services. Right?

Arvid: 18:03

Wouldn't it be exciting to get a thing delivered by drone? They could have a drone fly in and drop it off and fly away. That'd be great, but it kinda isolates me. Right? I don't wanna I don't wanna never to have a anybody show up at my door and hand me something valuable or even ask me for a signature.

Arvid: 18:21

I think there's still value in having a person hand you a package, or me going to the post office picking it up, having some kind of social interaction. The human element is part of this, and it might not be forever, but at least right now, I kinda still want this. It shouldn't be lost in the pursuit of automation. And AI is great for automation. That's what it's really good at it, but it should never completely replace human connection.

Arvid: 18:45

So you don't have to keep up with every AI trend to succeed. In fact, I think you can stand out by not doing that. You can focus on being present and giving your time to the people you interact with because that makes AI a side dish. Like I said, better to have something facilitated by AI, than selling the AI itself. Because if AI is the main course, you're all of a sudden competing with massively capitalized companies.

Arvid: 19:10

For independent businesses, and entrepreneurs who are not selling AI tools, I think it's crucial not to make AI the primary focus of your business and even your entrepreneurial perspective right now. I know everybody's talking about it, and I'm talking about it right now, but I'm trying to caution you against making choices to use AI just for the sake of using AI. It's a tool. It's an additional tool. And you might just as well emphasize your real presence and your commitment to serving the people you want to empower.

Arvid: 19:40

There's no need to fear falling behind in technology development by being present, being real with the people you wanna serve. So as AI becomes part of the mainstream, its volatility and development will slow down a bit. And I think it becomes an expectation that AI is gonna be there. And that allows you to stand out meaningfully by actually prioritizing genuine human connection over the ease and money saving capacity of AI. You're gonna stand out by being a person.

Arvid: 20:10

Who would have thought? And that's it for today. I wanna briefly thank my sponsor, acquire.com. Imagine you're a founder, you built this great business that has AI as a side dish, not the main dish. You get your customers, and you're generating good m r.

Arvid: 20:25

Problem is you don't really know what to do. You kinda you've you've hit the ceiling, you don't have you can't focus anymore, you don't have the skills, or you just don't have the interest to work on this business anymore. Well, what should you do? Ideally, you still keep growing the business, but maybe you can't. Maybe that doesn't happen.

Arvid: 20:43

And all the stories in this sphere, in this space that you are in are slightly different, but they they kinda show the same results. People stop caring, people stop working, and then the business becomes less and less valuable, or at the end, completely worthless. And that is a problem because you've built this great asset. You should be at least compensated for having built it. And the thing you can do at this point is to consider selling it on acquire.com.

Arvid: 21:09

Acquire.com has helped hundreds of founders already trying to sell their businesses. They've matched them with the right buyers. They've shown them how to set up an offer so that people are interested in it. It's free to list and honestly capitalizing on the value of your time today is a smart move. Right?

Arvid: 21:27

You have other things to take care of, other things to do. So go to try. Acquire.com/arvid and see for yourself if this is the right option for you and your business at this time. Or maybe just check it out so that you know what to do when that time comes for you. Thank you so much for listening to the Bootstrap founder today.

Arvid: 21:46

You can find me on Twitter at avidkahl, arvidkahl. You'll find my books and my Twitter crush there too. If you wanna support me in this show, please subscribe to my YouTube channel, get the podcast new player of choice, and leave a rating and a review by going to ratethispodcast.com/founder. That makes a massive difference if you show up there because then the podcast will show up in other people's feeds and this will really help the show. Thank you so much for listening.

Arvid: 22:09

Have a wonderful day, and bye bye.