[00:00:00] Speaker A: If I could go back a long time is really just to listen to everyone. But don't follow anyone, like literally don't blindly follow anyone. There's a lot of services that just need a human touch. There just is. I think it just comes back to just being original, being yourself and being a person behind the brand. What I think is important is just be careful you're not building a social connection with these AI agents because I see that so often.
[00:00:31] Speaker B: Welcome to The Entrepreneur's Logbook Podcast. I'm your host, Zachary Nerd. You can find me on Social Zack B. In each episodes I bring on experts from various industries for you to learn about their strategies and insights driving extra business growth. Today we're joined by someone pretty interesting. Jeremy Shoemaker, a software and AI infrastructure engineer who first made a name for himself in the 2000 as one of the biggest voice in early Internet marketing. He co founded Auction Ads which was acquired for 17 million and later founded PAR program which was then acquired acquired by go social for 12 million. After that run he kind of went a little bit quiet and under the radar for about a decade to build machine learning trading systems and he's now a full time hands on AI engineer. His code has actually been merged into open source projects at companies like OpenAI, Google, Microsoft, Anthropic, Apple, Hugging Face and Mozilla. And he's also the author of various books that have sold for more than a million copies including Nothing's Changed But My Change to Shoe Money Story and and a Feel Good A Field Guide to Living and Working With AI Agents. Jeremy, it's great to have you on the show. Welcome aboard.
[00:01:37] Speaker A: Thanks for having me Zach. It's a pleasure.
[00:01:40] Speaker B: I was really looking forward to this conversation because you and I had a pretty interesting like get to know each other where our, our company. We re. We reached out to Jeremy I think it was like a couple weeks ago if I'm not like totally mistaken because obviously we get people booked on podcasts and it was interesting. Jeremy's like hey you know what? Like cool, I love to get on podcasts. And one of the first things which I thought was pretty interesting because I've never had anyone do that for us and I thought it was hilarious was he just replied to us, he was like yeah, your website on mobile is ass. I built you a new one. And then you just sent a link with a download link to the exact like website rebuilt which it's pretty good, kind of better than our website to be honest with you. And I was like, you know that's kind of interesting. I got to talk with Jeremy because I've never had anyone do that directly. And I'm sure you've probably seen, like, people where they're like, oh, I rebuilt you a new website. They haven't built it yet. They just wait for you to say yes or they build it. But you already built it. It was like, what? Like, how do you do that? It's been 20 minutes. How did you do that? And we can obviously talk about, like, how you do it in a little bit, but I thought that was really interesting. It was a unique way of introducing each other.
[00:02:49] Speaker A: I mean, one, I thought your, your service was very, a very. I mean, I thought the, the value proposition was really enticing, especially me and especially the, the state I'm in. And also a lot of people I know are looking for a service like this. And it just, everything about it was right. And then I went to the site and I could only view like half a video. And I was like, man, I'm. This site is just not good. And so I, I was working on something at the time to, to quickly rapidly redesign sites. And so I just had it do it. And I just said, hey, you know, your mobile site kind of looks like ass. Or I said, I was, I mean, I should own it. I said, what? Your mobile site looks like ass.
And I made you a new one. And here's the GitHub repo for the code and it's hosted here, so if you want to use it, go ahead and then you email me back, like, super cool.
Which there's one of two ways you can handle that.
[00:03:46] Speaker B: I thought it was cool because, I mean, so many people are going to spend so much time and money building a website when you just need a core component that look good. And I was looking, I was like, it's been 20 minutes since you sent that last email. Like, how do you do that? And then I started doing a little bit more due diligence, like AI engineer in the 2000, pretty big Internet marketing guy. I was like, it's kind of cool. He probably has a couple things that he could talk about in terms like AI and everything like that. And we're obviously going to talk about a lot of those things in today's conversation. And I was kind of excited because you've been able to see both sides of the coin in a way where the 2000, like, early boom technology, everything like that, like, you were there at that time, but now, like, when AI is coming up, like, you're able to use those same skills and apply in, like, a different fashion. So we're going to dive into, like, all those things. But seeing that you've been, like, quite around the block, to put it very simply, one of the things I'd. I'd really be curious to hear from you is if you had to, like, restart maybe, like, your career or your companies from scratch, because I know you founded a couple companies. What's. Maybe the one thing that you feel you would do, like, different or your approach differently?
[00:04:59] Speaker A: Yeah. And I feel. I feel like I've had this exercise throughout my life of, like, you know, it's like the definition of insanity is doing the same thing and not learning from it. Right. Or committing having the same thing over and over and expecting a different result. And I think I've had to, like, look at that throughout my businesses and say, okay, like, what did I really like about this one? And what. What could I do differently, you know, and in approaching new things? And I think one of the big things is if I could go back a long time is really just to listen to everyone, but don't follow anyone. Like, literally don't blindly follow anyone. Like.
Yeah. And as we've talked, I mean, I've. I talk about people, I call them kids, you know, who are now really big people that I knew back in the day, you know, before they did anything, you know, And I would say, like, Gary Vaynerchuk is a really good example of somebody who, you know, literally just. His parents owned a wine business, you know, and. But he had talent and, you know, but. But like, he's one that everybody, you know, looks to and looks to follow, and it's like his journey is. Is just so unique and how he got there. And if you just blindly follow everything he says, you're not going to get there. But if you just listen to. I mean, I could listen to him, you know, all day and I'll come away with a page of notes that has nothing of what he said, but it's just, like, the point behind it and how I can apply it to me. And I think that's. That's a really big thing. I think a lot of people blindly follow these influencers and stuff like that, but they don't look at their life and say, like, okay, how can I implement this into my life and what I do? Because you're not. I mean, just to flip over and start, you know, you know, becoming just a total flip, you know, is just not going to happen. It's not. At least it's not A high probability of success.
[00:06:54] Speaker B: Yeah, I think that's interesting because yeah, like so many people are going to look at the big influencers, everything like that, like, oh, they are doing this. I should probably be doing the exact same thing. But when you look a little bit deeper, it's like they have these little things that made this possible when they applied it to like their own business, their career and everything like that. So it's more maybe taking like advice with like a grain of salt before you apply it. Would you say?
[00:07:18] Speaker A: I have. I literally could name like five different methodologies that I've learned to like, wow. This one is, you know, I need to really consider this, you know, and like, it's just, it's literally like 30 some years of experience, you know, of like, it's just what happens when you get older. But I mean, you definitely like, like I'm saying the things even I'm talking about now is like people should just look and see if they can apply it and other things. But yeah, I mean there's, there is, there is several. I mean one we talk. I or I, I want is, is the example of, of you just the way you acted and so professionally, especially at your age, Alicia, you look a lot younger than me and, and you're the founder of the company. To respond back and, and to just be, oh, you know, very, very humble and like that, like, that is a skill in itself, you know, of not to puff out your chest and be like, you know what?
You know, screw you. But yeah, I mean there's kind of like like two lessons there for me like to pass on is one is like the way you acted and the other is just with me and my company, I quickly learned as soon as I had employees that if you come with a problem, you come with a solution.
And I think that's a super valuable thing is that just as you work with others because you can just get. Everybody can come up with a reason they don't like something, but if you're part of an organization, then so naturally for me, when I thought that site was. I would have, would not have sent that if I wouldn't have also sent, you know, like, here's what I, you know, because I just think it's just so easy to say something looks bad. But you know, I don't know, it's just my methodology.
[00:09:01] Speaker B: Yeah. And I know we, we spoke before and you were talking about like your mythology in terms of like building new things like software companies and everything you were saying, if I recall correctly, correct me If I'm wrong, you were saying that you always just like start building stuff because you find like a neat problem or anything like that. You start offering it for free to the marketplace, giving away. And as you start getting demand, it's like, hey, maybe I start monetizing that. So it kind of looks like you're taking a similar approach. And you've done that with a lot of your past companies here.
[00:09:28] Speaker A: Yeah, and yeah, everything I've ever done, even, I mean like literally even my books and just, just everything, my blog, everything was always just I looked for something that wasn't there that I really needed and would use, whether it was a tool or just writing an article about something I learned and you know, then creating it, sharing it with others and if they find value in it, then literally the money always takes care of itself. And then monetizing it is a fun adventure, you know, like getting the people, getting the users is always the hardest thing. If you show me something that makes no money but is getting hockey stick growth, I mean one, you're not going to have a problem, you'll know. But I think that's what holds a lot of people back is they immediately come up with so many excuses for why it won't work when it just, it costs you no time. And right now with the tools we have, I mean everybody can be a top tier developer.
That probably is going to upset a lot of developers, but it's just the way it is from somebody who's hired and fired many.
[00:10:32] Speaker B: So yeah, it's becoming like a distribution game where like the product itself isn't always like the moat, but if you're able to distribute it, get an audience, monetize it, I think that's where like most like the pie comes from. And a lot of people don't quite understand how to do that yet.
[00:10:49] Speaker A: And yeah, that's going to be a
[00:10:51] Speaker B: bit of a problem.
[00:10:51] Speaker A: The thing is the money will find you just build something useful, that's it. Like, like, honestly I, I think my, my formula is so simple, but it's just what I do and what I've always done is just, you know, just build something I find useful, open it to others. I probably have hundreds and hundreds of these things on GitHub and some catch fire, you know, and some people find useful. The competitiveness for like a GitHub thing or something like that with now that everybody can make them is very competitive, but it's still there. And you'll still see, I mean the success stories now of just regular people that Aren't, you know, top tier pro developers making useful applications is amazing, you know, but there's also a lot of predators. But, but there is, you know, there's a lot of really good stuff. There's a lot of people trying to sell the dream. You don't. I would just. The one thing I would pass to everybody is you don't need to pay for any thing. It's just literally about will and patience and that's it.
[00:11:54] Speaker B: Yeah, yeah, we, we've built like some internal tools ourselves and like why did I not build this before? Like this is increasing your efficiency by like 3x saving us so much time. But I think there's like a good balance between figuring out is this actually worth building? Is this actually going to save us some time or does it just look cool? So many people are going to vi copes and tools and they like I built this like okay, what is the actual impact? Like how is that actually helping people? And then you realize like, I don't know, kind of looks cool to be honest with you. So yeah, there's a good balance in there, in the way.
[00:12:27] Speaker A: Yeah. And I think when you work with agents right now like you know, Claude Code or OpenAI or your own, you know, model and, and stuff like that and I think what a lot of people do is they try to do the all encompassing platform that does everything and is great at everything at once and you really just focus on the one thing you need. Right? So for me, a big thing for me how I started with my, I have a comms agents which is, you know, in the works. But, but literally because I have so many social outlets and connections go back, you know, 25, 30 years and you know, whether it's email and whether, or whether it's LinkedIn or Twitter or you know, just how do you keep up with all these DMs, right?
So I made an agent that you really can only make yourself because you know these, these companies aren't huge on that, at least in a commercial way or if you actually put out something it would be banned in a second. But, but yeah, I have it, I have it just manage all my communications on all platforms and so it'll draft a reply for me where I can just hit yes because it's, it's ingested all my books, it's ingested all this stuff. But I just started with a simple one for Facebook just because I never log into Facebook, right. So I just wanted something that just grabbed the messages and then I made it so it went to LinkedIn and then I made it, you know, so logged into Twitter and then Reddit and then other things because there's so much communication that I miss that is very valuable.
And so, yeah, and then to just have it in one inbox with the already drafted thing that you can just simply edit and send an archive is just super nice. So that was just how. That's just an example of, of, you know, just starting small.
[00:14:11] Speaker B: Yeah, but I, I like that though, because I mean, when you're like a, a technical founder, you would call that like an mvp. Like you get the minimum viable product that works, that does the thing. But from your perspective, I mean, I would call that pretty similar. But you find like the one thing that you actually need, you make sure it works and then, yeah, you can expand like the other things. I mean, for example, for you, if your main platform is email, probably going to focus on email, but afterwards, you know, you may add Messenger, Reddit or anything like that to track everything, but you go at it in order of like, priority, what actually is relevant or useful to me in like the first place.
Too much.
[00:14:47] Speaker A: If you try to do it all at once and you spec out a PRD for an agent to tackle or you to tackle, you know, if you're going to go old school, then you're going to end up with a project that you're probably burnt out on, it's not working and it's like 10% of all the things are done, whereas if you just make one thing that works, and I call this the Coke theory, which I actually got from Warren Buffett and I, I'm from Omaha at least the last 25 years or so, and went to see him speak, you know, over overton Probably 12 times or so, and he, he talks about his Coke there. They own Coke one for one. But it's basically real simple. If you made Coke, then making Diet Coke, you've already got all the things to make it right, whereas somebody really can't start with. So one, it's a competition thing, but two, is that. So if you've built the framework to do LinkedIn, there's not a whole lot that changes. But if you tell an agent to build this and, and just do it for all the things, it's not, it's going to, it's going to be a complete disaster. But if you have it, do one and then tell it. Okay, now apply the same logic with the, you know, depending on the tools. I'm trying not to get too technical, but I'm happy to get Technical.
But, you know, just, just simply tell it to apply the same logic rule set to, to LinkedIn and then. Okay, test that works. Okay, wow. Now let's go to Twitter. That will be done 10 times faster and it will just, it'll work in it right away. And so you get that dopamine rush of like accomplishing stuff, which is very valuable, and you just, you just keep iterating.
[00:16:28] Speaker B: Yeah. Especially if you, if you keep like the same context with like an agent, because if you try to do it all separately, they're missing all that context. So what worked? What was the logic and everything like that? And I think that's where all the puzzles starts, like fall off and not work perfectly.
[00:16:42] Speaker A: And that's. Yeah. And, and it's funny you mentioned that. I, I would. For anyone who's building stuff, I. And, and literally you can just say to your thing like, make an agent that does X when I do this command. And so for me, I have it like every so often I say, I want you to do what I call resume resync, which is basically it writes.
So what it does is it writes a little blog post like it was writing a blog of everything it's doing so far today. And I say, be, be bluntly honest, don't. No human's ever going to read this, so don't worry about hurting anyone's feelings. It is hilarious to watch what it writes about me. But so, yeah, so basically it just makes a very blunt post of where it messed up and where you know, other things, what it learned. And then I have it go over the chat and see if it can extract any skills or memories of, you know, simple things. Like if you use a Mac operating system, it uses a, what's called a ZSH shell. And really these agents are used to working with the bash shell. So there's a lot of little intricacies when it does little commands that don't work exactly the way. And you'll notice, like it always says, oh, this is a Mac os, so I've got to do it this way. But if you just tell it, like, go over our chat and see if there's anything you can extract from that.
And it will, and it's amazing. But it's also important to just every once in a while say, hey, I want you to show me everything you know about me and all the skills that you have, because you'll find some maybe aren't doing what you want. They just worked one time really well.
And it has a lot of memories that are not True.
So it's really good to do an inventory there.
[00:18:26] Speaker B: Yeah. Interesting. Yeah. I've always thought about just keep the context but it gets bloated after a little bit. So sometimes maybe you want to start a new session or anything or if your computer crashes. Yeah, yeah, that's another scenario. I've had that happen. I felt so stupid because in the beginning when I started building some vibe coded tool as you would call it, I would just do it in my Chrome browser and after a while there'd be so many files it would just crash. Recently I just started using like the. The actual terminal for like Claude code was like I can't believe I've been doing this like this entire time. And I would drag the file to GitHub to get it to push. Oh it took so long but got. Finally got it sorted.
[00:19:06] Speaker A: I have been. I literally have been using a computer in the terminal for. Since 1995 is so I mean like, like people around me, like when I met my wife it was just like you just, you know, they think I just look at a black screen of. You know, it's like why, why don't you use something with images and stuff? But it's just so efficient.
Yeah.
[00:19:28] Speaker B: Yeah, that's what I realized. Like.
[00:19:30] Speaker A: Oh my God, it is. It is really wild. You mentioned a little bit on it is that the tools. And my very first Internet related job was I was working on a dummy terminal for Linux and. And doing like what's called a RADIUS server which is. I didn't have any clue what I was doing. I just, just you know, pressing the buttons and doing the things and following what it said and resetting people's passwords and just simple things. And then I started editing you know like. And these exact command line tools are the exact command line things I use with Claude. So the exact scripting syntax. Like it's exactly. If you look at the tools we use in there they were. They're like copyrighted like in the 70s.
I mean like it's amazing.
[00:20:14] Speaker B: So works right?
[00:20:16] Speaker A: Yeah, I feel very fortunate. Yeah, I feel extremely fortunate like that I just.
I literally like some people have said like you fall uphill all the time and it's true. I just, I've been. I do feel like I, you know the whole luck thing versus you know, preparation meets opportunity I think is a. Is a really big thing because I would point to a lot of that. I've really prepared for the moments when they came. But yeah, yeah, we talked a little bit.
[00:20:48] Speaker B: Yeah we talked a little bit about the fact obviously you've been in this space for like over like 20 years. I mean I would say probably close like three decade at this point. But correct me if I'm wrong here by any means, but, and, and now you're like pretty obviously pretty deep in like the technical side of AI. I know you mentioned you didn't want to like to go too technical and like that set of things, but I guess when you're thinking about like non technical business owners who they're thinking maybe they have AI or they're, they're thinking about it, they have misconceptions.
What do you typically like find yourself having to explain like over and over to like separate like the, the hype from what's actually like real. Because some people like oh, it's gonna take over jobs and some people are like it's useless. But I feel everyone has a different perspective around that as well.
[00:21:36] Speaker A: Yeah, for, I mean I have friends who are even tech friends who refuse to do anything with AI because they want to enjoy their humanity while it lasts, you know, and these, these guys, I mean one of my guys was a founder of a very successful dating thing that was acquired for many billions. He wants absolutely nothing to do with AI.
He probably wouldn't care for me said his name, but I don't want that name drop. But yeah, so, but yeah, I mean I know people like that but I. So with him I'm just like, yeah, well you're loaded, so go do whatever. But for those who could actually use it and you know, and it could save them money or, or time or whatever, it's amazing. I mean it's. Yeah, I would, I would tell them I. Whenever I. There's so like there's been a couple companies that I've worked with, you know, who were looking at getting an AI. My whole family works for John Deere. Right. My sister, my mom or my.
Just everybody. And if you're from Moline, Illinois and you have a great job, it's probably with John Deere. So, and so when they were considering it and my sister at the time was a vice president of something and almost C level and she, we talked about it and with some others there and, and I said, you know, what you should do is actually go through what you don't like doing what takes up so much time and is so hard and don't think about it like taking away jobs, like look at what takes all your time and you hate doing.
And let's, let's just gather that amongst every department and let's do that. And so we built some, some initial, I would say MVP model stuff which then they contracted with, you know, I think aca, I think is the company they went with. But.
But yeah, it's, it's just like, so just anybody, I always say to them, what's one thing you do on your computer that you hate doing?
And a good friend of mine who works for the city, he has a nice city job, he really, you know, he doesn't really want to embrace AI. And one day we're sitting there and he's like, well, let me show you something. Like, a lot of people don't know that computer use with OpenAI, you know, with their models is especially really, really good.
It can use every application and be a wizard at it. Like I even use, I use computer use to remote to other computers on my network and like install software or install drivers that I like. Sometimes a Linux box will go rogue and I just am like, yeah, you know, remotely connect to the, to this box and fix it.
It just does. So like, but what he does all day, without getting specific, he basically said, you know, this is my entire job. And I was like, what do you mean? And he's like, just what we've done here. Like it basically was going to his email, grabbing the message, grabbing the attachment, readjusting the attachment, evaluating it, doing it back. And he's like, so we can just do this all day? I was like, well, yeah, I mean, as much as you want, you know, And I think people, people think of AI as just a development thing or just you have to program it and all this. No, you can just go to chat GPT, download the Codex desktop thing, give it permissions and say my printer's not working, fix it. You know, or my wife had an issue with her, what she likes to call her finger mouse with the trackpad. But it's officially, I just looked at her, but it's officially known as the finger mouse here. And so yeah, it was, there was something going on or some conflicting thing. I mean like, so I would say to them, you know, like, I mean, what person, what normal person? Not me, but what normal person enjoys trying to fix those things? Like that's a pain in the ass, especially if you're a Windows user. Sorry, but I mean it's. I use Windows for me too. Has its purpose, you know, it's great at games, but, but yeah, so I mean like, but how much do you enjoy when you're. Sound is a word, you know, and then it's like, oh, what the. You know, like you've gone through everything or you just bought this new camera and it's not working. And that could be anything. Well, AI is just a ninja at that. And a lot of times there's new, more optimized software drivers that, you know, it's, it's just, it's just so good. So I would just, again, this is a long answer to your easy question, but I would just ask them what do they hate doing? What are they not good at?
And you know, don't think of something technical, even just, you know, super technical. Just, just something that can be just busy body work.
[00:26:09] Speaker B: Yeah. Because I feel like you've been able to see it from a lot of different ways. Maybe it's like again like someone that works for the city, enterprise level, the tools that you're building, SMBs and everything like that. But when you're thinking about normal people, as you were saying that maybe just have a normal job, you know, if they're, they see AI, they're maybe going to use ChatGPT for like a receipt, like not a recipe for like cupcakes or anything like that. Like I know my mom loves to do that in some cases, but people don't take the time to say, oh, what do I hate doing at work? That feels very repetitive, that maybe could be done very easily. It's like, hey, I can actually do this. Like that's crazy. Like that's one of the things I've been a big proponent of is every like couple of weeks I'll ask my team, like, hey, what's like one thing that you guys are taking like a field, take a lot of time to like do and is there like any way I could streamline that for you? Optimize it. Done. We've been able to do a lot of things with AI and it on a business context, so many things you can do. But it also applies to just everyday people. Like they don't realize that a lot of the work and things that they do could be streamlined, save them a lot of time and it's, it's crazy to think about like where we'll be in like a couple of years. I know some people have the theory like, oh, we're going to be replaced by robots and some other people, like it's just going to make us more efficient. I'm with the latter, but I know everyone has a different opinion on that as well.
[00:27:25] Speaker A: So I mean I'm giving my opinions on like, like what do I think it means for humanity as a whole. To like. Yeah, I mean like I, and I definitely have them.
So I think, I think like as AI now you will see jobs replaced but you'll see so many new opportunities presented. And I think the opportunity for people that have a technical understanding like of just like I was thinking about making a course or something for free just, just to explain the basics that will help you so much communicating with AI and I don't those will be needed much more. But just understanding like what's a function or what's an array or what's a cache? You know, just little high level explanations so that someone would have the intelligence to say like have we looked at a cache for this? You know, or what would, would that help? You know. But yeah, specifically like I think one of the things I've, I've wrote about was you don't. I think the, the, the chapters actually, you don't have to hire anymore.
And actually it's just about. Because I have had a lot of developers that would come in at 10 and leave it to.
Because they were so goddamn good and they were complete jerks, you know, and they, but they were the guy who could figure it out, you know, and they were brilliant, but you don't have to put up with that anymore. And the good guys who actually have talent are gonna do very, very well. Because I think it's. This social game is much like more important than ever with the, in this, in this world of AI and literally the other night our air conditioning went out and we have my wife's 80 year old mother that lives with us and it was not going to be a good night if we couldn't get someone out. And we had a contract with this company and they have this AI system and my wife is going nuts because nobody is like, you know, because she's. We just need somebody out here. So I wrote an agent that would talk to the AI agents because she was like I'm gonna hate this world. I'm gonna hate this world with everything's AI. And I'm like no, because I'll build the things that will interface with those AI things. So I did and it worked on the first try, you know, but, but actually we ended up going with.
We found a person that answered the
[00:29:43] Speaker B: phone and came out right?
[00:29:45] Speaker A: And I gave him a monster tip and literally. And he will be. We canceled our contract with this other place.
And this guy is such a nice guy, so good. And it's just like that guy is going to be badly needed. All these companies, moving cities, there's A. There's a lot of services that just need a human touch.
[00:30:08] Speaker B: Yeah, that just is. Oh, that's like the, the main thing I'm seeing right now. It's like a lot can be like optimized, automated, but the one thing that will not be is like that human touch, that customer service. And a lot of companies, like, I feel like home services you mentioned, like H Vac, for example, like air conditioning, like, that's one thing that you're seeing. And I'm on like X, like Twitter as you would call it. And I see so many people complaining about that, like, oh, like I'm calling like a company just to replace this thing or something like that. It's always like an AI that like picks up. And I feel that a lot of people, I mean like even yourself. You said it. Yeah. There was an AI company you went to the next one you're like, I'm not dealing with that. I'm going to find out the guy that actually picks up the phone and does good work. And again, I think that human component is just going to be even like more prevalent as we move along in like the upcoming years. So I'm totally with you on that. And I feel a lot of people are going to fall behind.
[00:31:02] Speaker A: Yeah, I literally think if I could give an advice to a developer worried about their job, I would tell them to, you know, get some evaluations of your social skills because I mean, like, I, you know, you don't know them in a work environment. But how easily do you get frustrated? Like how much patience do you actually have when you work with, how are you with working with others? Because, just. Because I've had to be the developer and the garbage man and everything in my company, you know, whether, because I was one employee to 300, you know. So yeah, it's just like I never think any jobs below me and I'm happy to contribute up because many times I felt like I was working for a lot of my employees because, you know, I mean, just, just I, I have my skill set and my other stuff and I know what I'm not good at. So yeah, like, talking to my financial guy was like talking to my dad for money. Like, it's just so weird. Like, why do I feel bad about it? It's my money, you know, but it's just so, but yeah, I, I a lot. I've seen a lot of really cool tools revolutionizing.
Yeah industries and I'm seeing a lot of really cool tools totally teaching us that there's. They shouldn't exist.
And I think like right now it's a big learning opportunity. I think a lot of these services like we just talked about, I think your plumbers, your electricians, your services that you need like now and you just need an answer now.
And I was looking at the psychiatry thing that was just like you get to choose well if you're in a state of, of like panic or whatever, you know, or whatever. I just can't imagine these people that are in such a state trying to navigate an AI menu and not lose their mind even more. I mean like, it's just, just seems, doesn't seem to make sense.
[00:32:44] Speaker B: Yeah, no. One thing I want to go like dive a little bit into before we like wrap things up. Because there's one thing that I thought was interesting when we, we spoke initially, like you had mentioned that you were building some like your own like servers and everything like that. And most people nowadays, what are they doing? They're just renting it or to put it very more simply, they're just paying a subscription to ChatGPT, Anthropic Cloud or anything like that. So it's in like the cloud. But you've also talked about like the route of like owning like your own infrastructure. And I feel like for a lot of like enterprise, especially when you're thinking about compliance heavy like industries like that's going to become even like more popular. But I'd kind of be curious to hear like from you, like where do you think it makes sense? Like when actually does it make sense to own versus rent? And what do you think people tend to like underestimate when they're looking at the, the real economics of running AI? Because some people like, oh, I'm just going to, you know, get a Mac Mini or something like that to run my own model. Not. And then we'll be good. I won't have to pay for AI ever again. But you know, there's a pros and cons to that.
[00:33:48] Speaker A: Oh for sure. I mean, yeah. So I have, I have quite a cluster that runs and I also have on my Hugging Face Shimani repository. I put all the models that. So I actually take like the Facebook model, rip the guardrails off and then I fine tune it and training on stuff and that's amazing and it's wow. It's a cost savings. But if that's what you want to do and learn how to do that, it will save you a ton of money. But yeah, I would say 90% of the time I use just regular ChatGPT, so I mean, like, even though, you know, like I have that available here and I have done some really neat things with it, it's more of a hobby thing. And I would say to a lot of people, like, if you want to. I mean, I've been around from, you know, the era of like there was no, you know, like, I actually had to pay for like two load balancers at 75 grand apiece. And for my first auction ads company, you know, I mean, like that's. I mean, load balance. And it was amazing when I saw like AWS and.
But yeah, so like my first one, I was paying like $200,000 a month to start,
[00:34:57] Speaker B: right?
[00:34:58] Speaker A: And now with a. Now you see, like when I started with aws, doing hardware there and doing like a horizontal scaling auto scale group, right, with an elastic load balancer and just, you know, amazing stuff. Everything can scale to the moon and scale back down.
And then it was like, you know, but maybe you need like two or three grand to start, right? But. And then you go now to some of these services like Vercel and other things that massively will take a Docker image and scale it to the moon. I've seen the evolution all the way from, you know, having to have your own hardware, shipping it to a data center, where I had one in Minneapolis, had a full rack of brand new Dell servers. Just the electricity and other stuff. I mean, it cost me like an initial investment of like $200,000 and it was like, like close to 10,000amonth in electricity and other things. I mean, like. And bandwidth. Bandwidth was. I remember when we. Oh my gosh, it was. I thought we were down and I was calling my.
My guy who was running that, who was in Minneapolis. I was like, what is the deal? And he's like, we have exceeded the capabilities of fast Ethernet. And I'm like, what does that mean? You know? Well, we had gone over one gigabit, which was like the best. There was no 100 gigabit. You know, like Cat 6e had not. Well, it might have been out, but it wasn't being commonly used. So I had to upgr to, you know, like a 10 gigabit switch.
And back then that was like a hundred and some thousand dollars which is now available for, you know, probably under 500. I mean, I. Yeah, I don't know. And probably much better quality. So anyway, like, the opportunity that we have is incredible. And just seeing all of that come around. So should. What should you do? I mean, I think if you want to learn about it, I wouldn't go at it from like, if you want to save money with it, I don't think it's a good solution. If you want to really learn about it because you, you really want to learn about it, then, you know, it's, it's a really, I mean, everything's out there on how to do it, you know. And like you said, I've contributed to OpenAI's codecs and Claude codes, you know, tools as well. And that's just because I'm an everyday user and when I find a bug, I look into it and see, is this actually a bug or is this just me?
And then I say, you know, and I work with the agent and I'm just like, can you look at the repo and see if there's any issues with this? And if it says like five other people are having the same thing and we just fix it and I'll say, all right, let's make a PR and submit it and maybe it'll get merged and you know, probably, I'm probably like out of every 10 I submit, you know, maybe half get merged in. But it's usually for some small thing. But you know, it's, it's really cool to be able to contribute, you know, to a lot of stuff. And it's also a great way to make connections too, you know, and it's just, it's just by using the tools, you know, and finding the bugs and then give what I say, giving back, like if you fixed it, you know, whatever. Just, you know, ask your agents to look and see if other people are reporting it and then just say, why don't we fix it? And your agent will figure it out just on that, you know. Yeah, it's so wild. But, but it also means there's so much more competition.
[00:38:07] Speaker B: Yeah, I feel like for the everyday like or average business owner, if you're thinking about AI, you probably, as you said, shouldn't be trying to like build your own infrastructure. I mean, for like a time it was like pretty big. Everyone was buying Mac Minis and like, oh, like some people like 10 Mac Minis, like, oh, this is amazing. Like I'm building my own. And then a lot of people just end up selling those on ebay afterwards. It's like, I'm not using it. I feel like when you're thinking about enterprise because you've obviously consulted work etc with like more like enterprise organization and for that would you. I feel that can be maybe a little bit more worth it for them, especially in like compliance heavy like Industries. But I'm not sure if that's something you've seen a lot or not.
[00:38:45] Speaker A: Yeah, and one, one thing I see a lot is these companies going with enterprise plans over. And I know some of the terms of service get fuzzy in that. And you know what, everybody figured that out for themselves. I don't want to like tell anyone what to do with that, but, but I will tell you that, and I know this because I've worked on the Codex, you know, application that people use that if you're an enterprise plan using the API or if you simply are on the 200amonth subsidized, I think you get almost $20,000 of API use out of 200 plan.
If you just simply uncheck share my data, then it is the exact same access that the enterprise uses is exactly the same.
And even people, yes, I'm telling you, if you have OpenAI and you want to have enterprise level privacy in whatever they're paying for, which is outrageous, but especially when OpenAI has a 95% profit margin, they say, you know, then yeah, just unchecking that share, let me help everybody improve and share my data. If you just uncheck that, there you are using the exact same level of security and anonymity and can't train on my data that the enterprise people are. Are they training on that? I don't know. I can't. I honestly don't know. You know, there's a lot of really interesting stuff out there. I, I do know some about it, but.
But then, you know, also it becomes a good solution locally with enterprises, like with John Deere, for instance, they have all these manuals, right? So yeah, the one thing they were talking about was they constantly get asked the most obscure questions, like they have a 1935, you know, tractor or something. Hopefully combustible engines go back that far. But it was leaking oil, right. And they just don't know how to troubleshoot it. Who's alive that does, you know. But they do have the original manuals, so that takes a human to go and look that up. Well, we can just ingest these.
And also I had to ingest every YouTube video transcript, every forum post. There's forums, dedicated age of gender that aren't theirs. But because, you know, they can actually use that content stuff. So actually when you chat with it, you say, I have a 1932, whatever, it's leaking oil out of here and it's kind of dark and it'll say, oh, here, print this out. Here's a step by step thing to do and just instantly. Right. And that, that's extremely efficient. Great customer experience. That's the one place that an AI can step in, you know, incredibly good customer experience available 24, 7 works on Christmas, you know.
And did it take anyone's job? No, it saved somebody who would have taken most of their day trying to go find a manual in a back room.
So yeah, there's definitely use cases for that and there's specific models and stuff to use for that, but it just depends.
[00:41:44] Speaker B: Yeah, it's going to be an interesting next couple of years and I mean even months, I mean things have been moving so fast. There's a new model that comes out every single day from what I could tell. But I guess just like parting thoughts. Jeremy, is there like any advice that you would give whether it's to like business owners, executives when it comes to AI, whether it's implementing it, using it? I mean you've been around the early 2000 now 26, obviously we're going 27 pretty soon, but you've been able to see like the evolution of technology as a whole. So a decrease on that.
[00:42:16] Speaker A: Yeah, I think a lot of times I think FOMO is becoming a big thing. So, you know, fear of missing out.
Even with me. Even with me. I mean like we had President Trump the other day saying we're renaming artificial intelligence, right. To superintelligence.
[00:42:33] Speaker B: Yeah, I think it's, I don't know
[00:42:34] Speaker A: if you saw that. I witnessed live on GoDaddy, the amount of superintelligence SIO, like search, you know, super intelligent optimization, super intelligent, you know, consulting, the land rush of names being registered. I think it was one of GoDaddy's biggest days in years.
It's just like these things are happening every day. That's straight fomo. Right. I mean I even got a three letter domain, you know, because one of the projects I have, I, I, and I, I even had a debate with my agent. I said, I don't, I don't know, is this wise? You know, and I think so. What I think is, is important is just be careful you're not building a social connection with these AI agents. Because I see that so often.
And, and because it is, it can be addicting. I think we're going to have a real problem with that.
And also the other, the other thing is just everything has come back full circle. So if you look at like the thing that I talked about to where, you know, we had to spend all this money to do something, but if you did it, there was Very little competition because very little people could afford it. And especially if you did it one way, you could do it another way because you've already done it that way. Right? So the Koch theory, and then you take that a step further and now we get into serverless. You know, it lowers the initial cost, but it also lowers the barrier to entry, more competition. But we have serverless stuff now and then, you know, aws and then take it like to the vercels and that of the world. I guess you would call that serverless. But so, but so what does that all come back to? When AI is implemented and everything's easy and there's a million competitors, I think it just comes back to just being original, being yourself and being a person behind the brand, because the one thing you can't duplicate is yourself, right? Like, and people can use avatars and all this stuff and AI stuff, and it's very nice and it's very realistic to make a fake influencer and all that. And obviously they get followers and stuff like that. But literally, if you're going to be behind a product and all that, though, if you look at all the competition I have for some of these things I can show, like, like I have a AI ranking indexing tool that I'm building in a whole database and I can say, well, here's the thing is I help build the freaking tools that y' all use, right? So I also, you know, have just like, I've done these things and if you actually ask Chad GPT about me or my book, it knew about my book within 24 hours of it being published and it used my site over Amazon for a citation. Like, why did it do that? Well, because there's ways to get it to index and stuff like that. So I think we're going to find the human touch, really is what it's going to all come back to is everybody's so worried about the AI taking the jobs and it's going to make life a lot easier.
That's just the way it is. But that's just the evolution of technology.
So I would tell people, don't be afraid, but don't fall in love. Okay?
[00:45:36] Speaker B: Yeah.
[00:45:37] Speaker A: And then just, just keep being nice. This is. It's that simple. I mean, like, I would just tell people, relax a bit, you know, I like that.
[00:45:46] Speaker B: That's interesting because I feel like a lot of people have given me like the opposite. It's like, go make sure that. I mean, obviously use it. But a lot of people, like, make sure that you are implementing it you're using, but the way you're seeing is like, yeah, definitely use it, but make sure that, you know, you're a nice person. You have that human touch whenever you speak to people. You're keeping those relationships because the, the human element is going to be like the biggest things, like, the next couple of years because obviously a lot of people are just going to delegate their entire thinking and personality to AI and that's. Yeah, maybe, Maybe not the best.
[00:46:16] Speaker A: Super easy to do. Yeah, super easy to do. And like I say, I delegate the task I don't want to do. So I will spend, like almost every morning crafting a beautiful prompt that's going to loop all day and solve a problem for me so that when I'm, you know, then I can play with my dog during the day or be with my wife who works from home during the day or all. Anything. I can just go. I love to just go to McDonald's before 10:30 because I love Egg McMuffin, so. But I like that, you know, and I love knowing that I'm working, but I'm just. You know what I mean? And that's why just having it do all the things you hate, I get more done by more developers that cost me absolutely nothing by, you know, noon than I've ever done, you know, and it cost me. It's just, it's an amazing time we're living in. I feel very fortunate that I got to see the Internet, you know, wave and, and now AI.
Well, LLM, it's. I think AI has been going on for a long time. It's just the LLM or the SI phase. So it'll, that's, It'll be, It'll be really interesting.
Yeah, very, very interesting. And now that we're training on synthetic data and reinforcement learning and you're seeing the models come out so much faster, it's accelerating at a really, really exponential rate.
[00:47:36] Speaker B: So, yeah, yeah, it's going to be an interesting next couple of years again, to put it simply.
[00:47:41] Speaker A: But absolutely, yeah.
[00:47:42] Speaker B: Jeremy, I really appreciate you coming on the show. You've had some people talk about, like, AI, but I feel like we've been able to dive a little bit more, like, in depth into it. So I was obviously looking forward to this conversation here, but I know that you obviously had a new book that came out not too long ago. You're. You're resurfacing on social after your decade under the radar, as I would like to call it. But for anyone that wants to get in touch with you, Reach out. What's kind of the, the best places to, to find you on here?
[00:48:08] Speaker A: Yeah, I mean I'm on Twitter, Facebook, you know, everywhere. YouTube, everything is shoe money everywhere. And if you try stealing my thing, I have the trademarks and I will get it.
But yeah, that way I'm
[email protected] or just shoemoney.com you can reach me every which way. I'm pretty reachable and on all the networks and I do have an agent that will inbox your question.
It's my attention. And Zach actually got a reply from Claude which I was not expecting.
[00:48:36] Speaker B: But I saw, I saw that Claude show up and I was like, you know that, that kind of rings the bell. Is that like the, the cloud, I'm thinking, or is that just like his actual consistent. And then I saw like they're applying like, you know what? I think that's an AI.
[00:48:50] Speaker A: It worked.
[00:48:50] Speaker B: So it was cool to see it in action.
[00:48:52] Speaker A: It's weird because you're the only person that's ever. And it's supposed to go by Clark and it has its own account, but it's only supposed to do things that I tell it or email people with my explicit, you know, permission.
[00:49:05] Speaker B: So.
[00:49:05] Speaker A: Yeah, but I, I do take the guardrails off and a lot of people don't think you can unrestrict a cloud or an open AI. You totally can, you know, just as much as you can do with Facebook or, or model and, or anything else. So yeah, there's, there's a lot. If you're really curious about it and want to do it's. It. It is a little scary that you can do it so well and you know, but yeah, thanks so much for having me and this is great, great conversation.
[00:49:33] Speaker B: Well, we're going to put like all the, everything in like the, the show notes. If anyone else reach out and like show money.com obviously your, your famous blog and like website if anyone wants to, to check that out. And yeah, to listeners, if you've enjoyed this episode, don't forget to like comment, subscribe, review all that fun stuff. Reach out to Jeremy here if you want to learn more, chat with them and then, yeah, until then, thanks so much for tuning in. Keep pushing and we'll see you in the next one.
