The $20 App That Proves Anyone Can Code

What do you do at 11pm when you're laid off, out of runway, and one bad night away from crashing out? 

If you're Karima Williams, you vent to an AI chatbot… then turn that private ritual into an app with two million views in a week, without writing a single line of code. 

Joe Downs and Peter Swain reverse-engineer exactly how she did it: the two AI tools behind it, what vibe coding actually is (spoiler: way less intimidating than it sounds), and how describing an outcome in plain English is now a legitimate path to AI side hustles and financial freedom. 

If you've ever killed an idea with "but I can't code," this one's for you… twenty bucks a month says otherwise.

Listen For

7:45 Why does venting to an AI chatbot at midnight actually make you feel better?

20:28 What is vibe coding really and why is it nothing like what you think?

25:05 How do you build something with AI when you don't know how to do it yourself?

39:54 Why should one AI write your code and a completely different AI fix it?

41:31 Which Claude model should you actually use Haiku, Sonnet, or Opus?

Links Mentioned

Peter's Free AI Business Audit | Storagemoguls.ai | Claude | ChatGPT

Email the “Idiots” Joe and Peter

idiots@successfulidiots.com 

Joe Downs

Website | LinkedIn | YouTube | Email Joe:joe@belroseam.com 

Peter Swain

Website | Email | LinkedIn | X

Transcript

Copy Transcript Copied

Peter Swain (00:00):

What AI brings you is you do not need to know how to do something. If you could describe what you want, I can build it.

Joe Downs (00:16):

A few years ago, she was running a $720,000 grant program at a blockchain startup. She had a good title, made good money, and she was generally good at it. Then she got laid off. She was a single mom with one kid, six years old, didn't get a severance parachute, nothing to help her. So most nights, instead of doom scrolling, she started opening up an AI chatbot just to talk to it. Not for work, just to vent, kind of like you would to a friend at 11 o'clock at night when you don't want anyone to wake up. And somewhere in there, she had a thought most of us have and immediately ignore. If this is helping me not to lose my mind, maybe it could help somebody else too. So she started to build it, a web ap for anyone to do exactly what she'd been doing privately.

(01:07):

One problem though, she'd never written a single line of code before in her life, but that didn't matter. She used a tool that builds working software from plain English descriptions and when it broke and it broke plenty of times along the way, she'd copy the error, paste it into an AI chatbot and ask it to fix it. That was the entire engineering team. Two AI tools, 20 bucks a month, no developers. A month after launch, she posted a demo. Within a week, it had over two million views and tens of thousands of users. I'm Joe Downs with these Peter Swain. We're just a couple of successful idiots using AI to simplify our lives and optimize our businesses. Peter, true or false, the British don't crash out. They just quietly re-pot a plant, say, I'm fine, and never bring it up again.

Peter Swain (02:02):

Of course, I'm fine.

Joe Downs (02:06):

Now, there's no way you know who I'm talking about here. I don't think so anyway. And this is no disrespect to her. I'm trying to champion her here. Her name is Karima Williams. She's a Howard University grad and her app is called Crash Out Diary. And it's built for anyone about to crash out and do something they'll regret. And by the way, when I came across this, I had to do a little more research on it. I think she's younger. I think she's 34 years old, so she's different generation, but I had to actually look up what... I had an idea what crash out meant, but I wanted to look it up to see what did she mean by crash out? It's almost a different language among generations. I had to chuckle at this. What crashing out means, Peter, is a Gen Z term for when you emotionally unravel and do something impulsive you'll regret like posting the subtweet, sending the text or, and this is what made me laugh, putting your ex's number on a Craigslist ad in six cities.

Peter Swain (03:13):

That's funny.

Joe Downs (03:13):

She actually did that one. She told the story in an

Peter Swain (03:16):

Interview. Maybe you should vibe code the Gen Z translator for us old folk.

Joe Downs (03:25):

That's a great idea. I don't even know what generation my kids are. I don't know what that qualify as, but -

Peter Swain (03:31):

You should go after that no cap.

Joe Downs (03:33):

Well, yeah, we got that. I know what that means now. That's old school. We learned that a couple years ago, but it is a different language. I hear them on the phone. What are you saying? What did

Peter Swain (03:45):

That mean? Yeah. My son whose long one said Skibidy no cap. And Olivia's like, "You should be insulted by 11-year-old." He said, "But at least he didn't say Skibidy Nocap, Ohio." Well, that would be terrible. Well,

Joe Downs (03:59):

That's the worst. That's the worst.

Peter Swain (04:00):

If you say so, if you say so.

Joe Downs (04:03):

Sometimes I just walk around and say Skibity no cap. I don't even know what I'm saying. And they just die laughing.

Peter Swain (04:10):

Welcome to the new world.

Joe Downs (04:11):

Yep. Hey folks, before we get into it, I'm super excited about today's show. Just quick reminder and a please, would you mind giving us a rating? It'd be great if you did right now. Apple podcasts, Spotify, wherever you're listening, a rating really helps us. It's the one thing our show needs. We love what we're doing. Hopefully we're helping entrepreneurs everywhere. We see it in the emails, but we're trying to grow the show. So please don't make me grovel. All right. Peter, today we're going to cover how a woman with zero coding background, which is an inspiration for me because obviously I have zero coding background, built something that pulled two million views in a week and the exact two AI processes behind it. Not the how-to manual, not the lines of code. I don't even know if that even makes sense, but just the what you need and how.

(05:07):

We're going to uncover for you the magic behind the mystery here, not the actual how-to manual. And one you could use whether you want to build an app or just finally fix the broken thing that's been driving you crazy. And folks, this is not the geeky, nerdy, like I said, how-to manual here. We just want to let you know what you can do with AI. That's it. And for the how to the how-to, just ask AI. That's what Karima did. We'll uncover that here as we go. So Peter, I want to get inside Karima's head for a second. Her specifically, not a set. She's not a developer. She was a marketer. She ran a $720,000 grants program at a blockchain startup. She helped people navigate the Obamacare website before that. She was kind of tech adjacent, but not a techie. Then she gets laid off, single mom, six-year-old, no cushion.

(06:00):

So on these nights, like I said in the opening, instead of doom scrolling, she opened up Claude actually and she just started venting, not even to build anything. She had no intentions other than to feel better, I guess, just to get through the night. She does this privately for weeks, doesn't tell anyone because it feels a little embarrassing to admit you talked to a chatbot like a friend. Can you walk us through what's actually happening in that moment and why it matters that she almost didn't tell anyone?

Peter Swain (06:30):

From what perspective? What's happening from a neurological perspective or what the AI is doing? Where do you want to go?

Joe Downs (06:38):

Well, yeah, probably both, but what about the AI allowed her? I'm sure she didn't sit down to even vent to it. I'm sure she sat down. The story already didn't cover that part, but my guess is her first chat wasn't, "Can you be my friend and hear me out? " I'm sure it started with something innocent and then frustration set in and she probably said something like I eventually do or sometimes do, or I'll start cursing at it sometimes because you get to the point where you feel like there's a conversation happening and you lose sight of the fact that it's a computer you're talking to. I'll bet you that's how it started. So what's going on from a therapeutic standpoint, do you think? What about the AI is allowing her to go from that, maybe that first curse, frustrated curse, whatever vent into full-blown conversation where it's now acting like her psychologist or her therapist or her best friend?

Peter Swain (07:45):

So the first act, and if a therapist is listening to this, yes, I know I'm wrong, but I'm right enough for the sake of this example. The first act of therapy is to feel seen, heard, and understood so that you're not feeling alone. If you don't feel alone, then everything else is culpable. Our connection to society is important, is part of our makeup. And we have these wonderful things called mirror neurons, which means that we actually just need something to hear us. We literally just need something, someone to hear us in order for us to feel better. It's why you have the Samaritans, and I don't know the US equivalent, phone lines that you can call when you're considering committing suicide. I'm sure there's equivalents in the US. It's a complete stranger that doesn't know you, that is told they're not allowed to offer you advice.

(08:41):

They're not allowed to give you any guidance. Their job is literally simply to listen to what you have to say. And that simple act can make a massive difference in the kind of catastrophic outcome. So whether it's AI or a person, it doesn't really make a difference in terms of that first piece of the equation of making somebody not feel so alone. So that's what's happening in the human brain. In the AI brain, AI is a probability engine. So it is saying what it thinks would be said if you could ask every human what they would say. So we play this game on stage quite often, Joe, of the dog is insert blank here, and humans will end the sentence sleeping, waking, barking, howling, growling, et cetera. They don't say the dog is Advil because that doesn't make any sense. I had to look around me to find an object in order to even be able to do that.

(09:41):

The AI is doing exactly the same. So if you say, "I'm feeling frustrated," the AI is going to say what it probably believes should be said as a result. So if you say, "I'm frustrated," chances are your AI is going to say something like, "I'm really sorry to hear that. What's frustrating?" And you answer. And as you said, within a couple of minutes, you will forget that you are having a conversation with

Joe Downs (10:05):

AI. That makes a lot of sense and you're right. And did you also just admit that you have a bottle of Advil on your desk?

Peter Swain (10:12):

I do.

Joe Downs (10:13):

That's crazy.

Peter Swain (10:14):

Yeah. You have ibuprofen, which is what I'm supposed to have over here and I have Advil. It's one of the primary use cases really of... There's something I want to say that I hope I manage to make sense of. If you put in the continuum, if you put human intelligence at the zero and you put technology at the 10, I'm trying to do it verbally for the people listening, but you've got this scal of human to technology. What I think most people get wrong is they pace the AI at eight out of 10 close to the technology in terms of how they think about it, how they understand it. What I do or what I believe is a better way is to actually think of the artificial intelligence as closer to the human. So if you conceive of AI as a synthetic version of human intelligence, instead of conceiving it as a technology, you are much closer to understanding what it does and how it works and why it works and why it sometimes says something it shouldn't say and why it does something it shouldn't do.

(11:29):

Because technology doesn't do that. A blender, you press the button, it blends. You don't press the button on a blender and the top pops off. It's supposed to do the thing it's supposed to do. If you drive a car and you press the button or turn the key, depending on how old you are, the car's supposed to start. The doors don't fall off. But humans, that's what happens. You tell a human, you're going to go to the shops, you're going to get this, get this and get this and come back. Anyone that's had kids knows this. It's like how did you get from walk to the shop, buy this one thing and come back to it took you an hour and 20, you went to the playground, you didn't buy the thing and you don't have the money and where have your pants gone? Humans translate everything they hear and everything they experience into their own understanding and then execute next to what they think, which is much closer to what AI does than a piece of tech.

Joe Downs (12:31):

That is fascinating and it's such an interesting way to look at it. And I would submit to you, if you asked a hundred people that same question, they would say, no, it's more like tech. It's closer to the blender.

Peter Swain (12:44):

And it isn't.

Joe Downs (12:45):

And I might've been one of them. That's so interesting.

Peter Swain (12:49):

I get this from all the time of people say, why didn't it work? I'm like, did you ask it? " They're like, "What?" I'm like, "Did you ask it? " They're like, "Well, it, it, it. " And the escalation from it is to he or she. And the escalation from he or she is when you give it a name as you form a relationship. I'm willing to bet Kareema has a name for her AI. Willing to bet you money on that.

Joe Downs (13:13):

She probably does. It's interesting now that you're saying that is, and I've noticed this a lot, even in Claude, when it's going through its progressions for an output to me, it tells me what it's doing and then it's catching its own errors and then giving a reason why it made the error. Almost like someone would do it like a human would. And or when it gives me the output and I go, "You didn't follow the production rules, let's say, that I told you to follow for this document." It'll say, "You're right. I skipped that. I'm sorry, it won't happen again." And in my head, as I'm reading that, I'm going, "Why did you skip it? You are supposed to be technology that does what it's told and never makes a mistake."

Peter Swain (14:05):

It's not. It's an intelligence.

Joe Downs (14:06):

And I've never really taken the thought past that. I've just been like, "Wow, I guess it's

Peter Swain (14:12):

Cold." Interesting upgrade for you. When it skips it next time, if you ask it this, you're going to be quite surprised. If you say you skipped the production rules, do not change what you did, but tell me what I said wrong that allowed you to make that mistake. And you'll be surprised. It's quite fun when you do it. It will turn around to you and say something like, "Joe, you told me the production rules were available here, but you didn't tell me I had to follow

Joe Downs (14:47):

Them." But sometimes it does and sometimes it doesn't. So it doesn't make fickle.

Peter Swain (14:51):

Because it's an intelligence.

Joe Downs (14:53):

It makes it fickle like a human. It's interesting.

Peter Swain (14:55):

Yeah. Because when you read it, you're like, because if you said to me, Pete, can you produce this thing? The production rules are in this file. I understand that you have implicitly, not explicitly, implicitly told me the production rules are relevant to this thing here. AI doesn't do implicit all the time. It does explicit. And as it gets smarter and as we get new versions, it gets better at doing this. But I've had these where I've been going around the houses for a day trying to fix something and I've said, "What did I say that led you to think this? " And it'll go, "Well, you didn't tell me to follow the rules. You just told me the rules were here." I'm like, "Why the fuck did you think I told you there was thrules?" And it feels all very much like arguing with a super smart intern that is drunk.

(15:48):

Yeah,

Joe Downs (15:48):

It really is fascinating. Back to Karima, why do you think it mattered that she didn't tell anyone?

Peter Swain (15:56):

Well, I think this is kind of par for the course because if you go back to when you and I were young, if you were to tell somebody that you are dating a girl that you met on the internet, your friends would've laughed at you. That was weird. If you were to say that you're going to marry somebody that you met on the internet, people would think you have lost your mind. But if you fast-forward to 2026, more people now meet online than they do offline. Interesting. So as things evolve, what is regarded as you being weird, you doing something becomes second nature and normal. So I think it's a perfectly natural evolution of people going, "Yeah, I speak to my AI. Your friends are going to look at you like you've grown a third arm." Except when they try it and they go, "Oh, I get it.

(16:54):

I understand."

Joe Downs (16:55):

Interesting. I don't disagree with you either. And that is interesting that more people meet online now. I didn't know that. Although I was with some different couples this past weekend, 4th of July holiday. 250th anniversary of it might add.

Peter Swain (17:12):

Happy birthday.

Joe Downs (17:14):

Thank you. And two of the four couples I was with met online. I thought that was fascinating because I'd never been in company where that was the percentage. Okay. So she's sitting on this... Back to Kareem here. She's sitting on this private ritual that's clearly doing something for her, right? But she's not a developer. She can't just open a code editor or claude code or whatever the other ones are and start building something because she has no technical skills. So that's usually where ideas like this die usually for me anyway, because I don't have anybody to code for me. Really, I don't have any time. But what actually changed for her here in this case when she started to ask, in this case, Claude, how to do it? And what do you think she asked Claude? How did that go? So she's having these conversations with her Claude therapist.

(18:16):

She must come to the realization... Sorry, I just want to set this up better. That, "Hey, this is good for me. It's probably good for thee. How do I turn this into something?" What does that conversation

Peter Swain (18:30):

Look

Joe Downs (18:30):

Like?

Peter Swain (18:31):

Again, this is pure hyperbole, but I don't think that's what happened. I think what happened is that conversation with Claude, whatever she calls it, progressed and progressed and progressed. And then at some point she said something like, "What am I going to do? " I have an inkling that it was maybe Claude's idea and not hers of -

Joe Downs (18:53):

Really?

Peter Swain (18:55):

I don't know why. I

Joe Downs (18:57):

Could see that actually working with Claude or not. And we're not taking anything away from Karima, even if

Peter Swain (19:02):

That was the

Joe Downs (19:02):

Case.

Peter Swain (19:03):

Well, even if it was, she still made the decisions. She still had the grit and tenacity to go for it. It's not taking anything away at all. She

Joe Downs (19:10):

Got Claude to the point where Claude maybe made the suggestion. And again, speculation, we don't know. It could have been her idea, but even if it was Claude, I agree with you.

Peter Swain (19:19):

So let's say it is. And then I think the next question from her would've been something like, "Well, how do I do that? What does that look like? I can't do that. " To which it went, "Well, actually, it's prety simple. You just do da, da, da, da, da. Do you want me to have a go? " And I think she probably went, "Yeah." And then the very first version came around. So what I'm saying is I think she slippery sloped herself into becoming a vibe coder, which is what this is called, instead of making a hard decision to become. Because most people I work with, they suffer from a lack of imagination, myself included, you included, because we are trying to imagine something we have no frame of reference for. So on the basis, every single person that uses Claude right now could produce code.

(20:11):

Because if you could describe the outcome you want, you can code it up. Why isn't everybody? Because they don't believe what I just said. It sounds hyperbolic to suggest to somebody that is very non-technical that they can write functioning software.

Joe Downs (20:28):

Well, maybe because, and I'll admit something right now. The term vibe code to me sounds like a bunch of code bros got together and had a coding session. Like you heard about the famous ones at Facebook and Google where they code into the wee hours of the morning. When I hear the term vibe code, that's what I picture is a better way to say it. Are you suggesting and am I about to learn that vibe coding is not that? Vibe coding is something that if I was going or what Karima did was vibe coding meaning you have no coding background, you're just vibing? Is it literally that literal?

Peter Swain (21:09):

The easiest version of this, and I don't have any vested interest in any of these, but the easiest version is called Lovable. Lovable.dev, L-O-V-A-B-L-E.D-E-V. I think that's

Joe Downs (21:21):

What she used.

Peter Swain (21:22):

Yeah. Lovable wraps in all of the backend environments as well. So if you use something like Claudecode or Cursor, which are the big ones, you are still going to have to deploy a server, deploy a database, and then it gets techy. Yeah. Love you don't have to do that.

Joe Downs (21:40):

You lost me at Vibecode until you just redefined it for me. I couldn't even repeat the words you just said. So that's where someone like me goes, nope, I can't do it. To your earlier point, you said the difference is we don't think we can because we've never... Yeah, you're right. I don't think I can. Yeah, I sit before you someone who might be turning a corner in this conversation live, but coming into it, I sat here as someone even reading her story. I still had disbelief because she was tech adjacent and she must knew a little more than me to get into a vibe coding scenario because I don't even know what vibe coding is.

Peter Swain (22:21):

On Friday, we did our bootcamp on steroids where we help a bunch of people write a bestselling book in four hours. It's a lot of fun. And we added into the curriculum this time the landing page for people to sign up to the wait list.

Joe Downs (22:33):

Why would you do that on Friday? The day I took off.

Peter Swain (22:36):

It's the last Friday of the month. And we pinged everybody to tell them what day it was.

Joe Downs (22:43):

Yeah, but the only day I took off for the 4th of July was Friday, July 3rd. And that's the day

Peter Swain (22:50):

I did

Joe Downs (22:50):

Something that

Peter Swain (22:51):

I wanted. I apologize. Well, I can do it again just for you.

Joe Downs (22:54):

Is it recorded?

Peter Swain (22:56):

Yeah. All

Joe Downs (22:56):

Right.

Peter Swain (22:58):

Okay. Can I carry on now? Yes. I've been abonished for -

Joe Downs (23:01):

Apologies for the mastermind.

Peter Swain (23:04):

Yeah, I apologize for putting the needs and wants of 62 people above you, my pad. Anyway, one part of that that we added as an upgrade was people producing the website to sign up for the wait list. So when you produce a book, you need to get a bunch of people on the wait list so that they can all download the book at the same time. It was literally as simple as going into Lovable and saying, "I need a page that people can sign up to my wait list for my book. Here's the cover. When they sign up, you need to save it somewhere and you need to send them an email with a copy of the first chapter. Here's the PDF with the first chapter." That is the entirety of the book. There's no other steps. There's no click this, do that. That's it.

(23:55):

And four minutes later, every single person had a landing page with a cover on one side, bit of text about the book on the other side, join the wait list, sign up form, all the things all done.

Joe Downs (24:06):

And they all did this having, I'm sure, mostly no technical background. They

Peter Swain (24:11):

Used

Joe Downs (24:12):

Lovable.

Peter Swain (24:13):

I gave them the prompt to copy and paste. It was exactly what I just said.

Joe Downs (24:19):

Okay. So if someone's not in your mastermind, doesn't have access to that prompt, what can they do?

Peter Swain (24:27):

Is

Joe Downs (24:27):

It as simple as go to Claude and say, "Claude, I want to build a

Peter Swain (24:32):

Lady

Joe Downs (24:33):

For my book and Lovable and I need a prompt?"

Peter Swain (24:37):

That level of complexity, I don't even think you have to do the interstep of Claude. I would always recommend, as you know, voice prompting. So Willow Voice or Whisper Flow, one of the two. I prefer Willow, but they're both the same thing. And open up Lovable, hit your voice prompt and just describe what it is that you're looking for. The one thing, and this might be an unlock for hopefully for some people - side

Joe Downs (24:59):

Of lovable is just have the same kind of

Peter Swain (25:01):

Conversation. I've

Joe Downs (25:03):

Never even used the app, never been on it.

Peter Swain (25:05):

The unlock I hope that I can offer people is what AI brings you is you do not need to know how to do something. You just need to know what it is that you want done. If you can describe the outcome clearly and you can think through the nuances of what that means like, okay, well, what happens when they log in and it's the wrong password? What do I want? Do I want an error message? Do I want to lock it? How does that work? If you could describe what you want, I can build it. And that applies to financial projections, HR letters, letter of intents to modeling data points, working out where you should place your next shop. I can't even begin health regimes, diet regimes, fitness. You don't need to know how to do it. You just need to describe it. So for example, the end, the where you are and where you want to be.

(26:15):

So for example, if you were to go into court and say, in my case, I weigh 118 kilos. I want to weigh 98 kilos. I'm going on stage in eight weeks and it's a big deal for me. So I want to be under 110 kilos. I work in this environment, context, context, context. I have these resources available to me. I like this. I hate this. I won't eat quinoa no matter what you tell me. I enjoy steak. I don't mind walking. I hate running. I love the gym, but only heavyweights. Context, context, context. This is where I am. This is where I want to be. Tell me what to do. And it'll go, sure, here it is. Exactly the same applies in code of this is where I am. No. This is what I want. Something that works like this and it does this and it doesn't do that and it's for this audience and go.

(27:13):

And it'll go, put me in, coach, hold my beer. You'll come back 20 minutes five to 20 minutes later and it'll be

Joe Downs (27:20):

Done. And it's fascinating. And I'm thinking about that's such an interesting visual too. I'm here. I want to get there. This is what there looks like. A couple things came into my head when you were saying that. One was a saying about the ends justifying the means, which led me to think, well, the process might matter how you get there, but that's where the context comes in, right?

Peter Swain (27:46):

One of the arguments that purists have here, which is always interesting to me, is that it writes bad quality code. Who gives her monkeys?

Joe Downs (27:56):

So this is the code purist. This is the tech nerdy geeks.

Peter Swain (28:00):

Yeah. They say, "Well, it can't write code as well as I can. " Well, first of all, it's very interesting that the people that are being existentially replaced don't like the thing. I can't remember ever walking into a yellow cab in New York and them saying Uber's the best thing ever. I can't remember that conversation ever happening. So I think they're a little bit self-interested. However, they're also correct. But why do you care? If it doesn't have security halts and it does what it's supposed to do, who cares? If it achieves the goal.

Joe Downs (28:33):

Sure. At that point, if it achieved the goal of, well, who knows even know what Karima's goal was, but I'm assuming her goal was squishy like, "Hey, I want to turn this into a business that makes revenue for me. " And I'm sure she had a number in mind and we'll never know what that is.

Peter Swain (28:49):

Do you think that was it or do you think it was just, "I like this. I wonder if other people was just an outlet and expression of creativity to begin with? "

Joe Downs (28:56):

Good question. I don't know. But let's, for the purpose of assuming it was for - profit, and she had at least a squishy goal of, "I want it to make me $100,000 a year." I don't know, making up a number. At some point that inferior code got her to the number, 100,000, a million, whatever it was. At which point she could take it to whoever the 3M of coding is because now she's got the money and she can make the code perfect, right? Or they can

Peter Swain (29:27):

Fix the

Joe Downs (29:27):

Code. So I think the point is, the point you're making, correct me if I'm wrong, is, well, you don't have to because you've already said twice, who cares? Who cares if it got you to where you want to get to? You can always improve it from there. The saying mastery is not a destination, it's a journey. This is just part of the journey. If it got you from an idea and a conversation with an AI chatbot late at night to being criticized for your code, you were Obviously pretty successful if you garnered enough attention to be criticized for your code.

Peter Swain (30:05):

Yeah.

(30:06):

We were talking about this before. I'm on the board of a company, an eSports company and it's a great idea. It needs to exist and it's not a fundable entity yet. So the founder who is non-technical has sat there in Claude Code for the last two, three months and has built every facet of the business out. So for example, so it's kids playing esports in movie theaters. So the system now automatically texts the parents in the morning to confirm their kids are coming and records a yes no answer, but any answer, it works out the intent of the answer. It pings the coach with the pick list of who's coming today. It reminds the coach of any allergies that the kids may have. It sends the kids, the parents, a link to confirm allergies. He's just about getting it to the stage where it calls the lunch order into the local restaurant.

(31:10):

And I mean physically calls the AI calls the restaurant to give the lunch order. The coach prompted at the end of the day to provide the coaching notes for each child. All of it happens from the point of a parent registering their children to come to the theater to play whatever the game is through to the point of collecting money, failed payments, retrying cards, updating cards, asking them if they're registering for the next league 12 weeks later on. End to end, he doesn't touch it. And what he's adding in now is the system that publishes the Facebook ad to get the parents to register for the thing, which will mean that the entire loop is closed and the business runs apart from the physical coaches in the room automated.

Joe Downs (32:03):

It's incredibly impressive. Unfortunately, my brain can't stop thinking about the fact that this is esports gaming.

Peter Swain (32:11):

Yeah.

Joe Downs (32:12):

What?

Peter Swain (32:13):

So here's the little statistic which is -

Joe Downs (32:17):

There's a

Peter Swain (32:17):

Coach in esports gaming. Hey, you get scholarships for this in universities.

Joe Downs (32:22):

There's scholarships?

Peter Swain (32:24):

So here's the vomit inducing statistic. If your kids play on an Xbox or a PlayStation for two weeks unsupervised, the FBI estimate that they will run into two sexual predators within the two weeks. So parents - Lovely.

(32:41):

Put their children in front of Xboxes and have no idea what's happening. Ric used to own movie theaters. And when COVID happened, they lost their customers. When COVID came, when that was closed, they'd lost their content because the studios were so delayed. So they've now got these huge auditoriums with great audiovisual and no content to put on them. So they started running esports leagues. So kids 11 to 14 would get dropped off. Parents go to the movies. Kids sit and play on an Xbox on the 60-inch, 100-inch screen or the 10-foot screen, whatever it is, the proper movie screen. And there are coaches in there because when you put kids playing online, when you see them in Discord and the chats, it's incredibly toxic, really, really horrible. But when you put them face-to-face, playing with each other face-to-face, all of that behavior disappears. So yes, there are literally coaches there teaching them how teamwork, coordination and all those wonderful things.

Joe Downs (33:45):

I applaud the opportunity, recognizing the opportunity there. I applaud creating the safe space. I will not understand e-gaming, watching e-gaming. But hey, how

Peter Swain (33:59):

About this? Mr. Beast

Joe Downs (34:01):

Is Mr. Beast because of kids watching

Peter Swain (34:04):

Him

Joe Downs (34:04):

Play video games.

Peter Swain (34:05):

Esports viewership in 2025 beat the NFL, NBA and NFL combined and NHL combined.

Joe Downs (34:14):

I have no words.

Peter Swain (34:16):

It's a new world.

Joe Downs (34:18):

I hope it's all for good.

Peter Swain (34:22):

Anyway, let's return to the point. The point being, you've got a business that is incredibly, we need to register the thing, check the kid in, check the kid. There's a lot of manpower historically in that business doing a lot of very repeatable tasks. And for him to reach the expansion goals he wants of being able to go to a new town or a new state or a new county and open up another five locations, it was for the last year we were sitting there going, "Well, we're going to have to hire 20 people, 30 people per county, which means that we're going to have to raise X hundred thousand dollars per county that we want to open. And we're going to be cash flow negative for the first 10 months of operation as we build up that new theater." When you can turn around and say, "Well, all we have to do is just tell the AI we're opening up this new location and it takes all of the work and just duplicates it.

(35:17):

" It means that in real terms, we are not having to raise $200,000 per county that we operate in. And we're cashflow positive from month two, not month 10. So we can go up to the, we have 2000 cinemas on a wait list that want to do this with us and we can now go and scale across those much, much quicker than we would've been able to. But the point being he knew what he wanted in his head. He was able to describe -

Joe Downs (35:52):

He saw the end. Yeah.

Peter Swain (35:53):

Yeah. He said, "Well, the parents need to check their kids in and the coach needs to confirm they have the kid from a safeguarding perspective. And I need to know if the kids have allergies and I need to call the order in. " So he just sat and described for about a week the SOP of the business, this is then this, then this, and if this, then this. And if that, then this and if this, then this. And it just sat there and went, hold my beer.Here's bit one. He's bit two. Here's bit three.

Joe Downs (36:24):

If you can envision what your pot of goal looks like for this one, AI is the rainbow.

Peter Swain (36:29):

AI is the rainbow. I

Joe Downs (36:30):

Don't want to dwell on this next point here. I want people to understand it's possible. I actually have done this. I mentioned in the opening along the way the "code" she was vibing. I don't even know if I said that right. She wasn't really writing and there were errors with it along the way. So it would quote unquote break. I have definitely learned to, through your mastermind, copy and paste errors that I get with anything, paste them back into Claude or whatever AI you're using and then ask, well, first of all, tell it, "This broke. What happened? And then can you please fix it? " And then it walks you through step by step how to do it. She, at various points from what I read along the way, had that same problem, would've ended most people's journey right there. Can you comment on the why and how of this for us?

Peter Swain (37:36):

Yeah. It goes back to my earlier point of if you imagine it as closer to a human intelligence than it is a technology, then everything to me, everything starts making sense. I very much doubt that most of the people that are listening to this could write a bestselling book in a weekend, but I'm willing to bet quite a few people here could hire a ghost writer to write a bestselling book. I'm willing to bet most people that listen to this couldn't sit and code up a new business. I am willing to bet that they could hire a dev company to code up a new business. So it's the same thing of you're just outsourcing the work to an expert. The expert just happens to be a synthetic version of intelligence, not an actualized human version of intelligence. So if there was an error in your code and you had a software developer sitting next to you, there's a good chance you'd ask the software developer, "What does this mean?" Or even better, you'd say, "Can you fix that please?

(38:41):

Why am I even in this conversation? Just fix it. " 100% what I do. That's what AI is. If you keep rethinking of it as in people used to ask me years ago, how do you be successful in social media? And my answer was be social and share media. The clues in the name.

Joe Downs (39:02):

Sage. Sage advice.

Peter Swain (39:05):

Artificial intelligence. It's an intelligence. Learn to have the conversation with it and treat it like an intelligence instead of treating it like a tool. And you'll get much, much better results and you'll understand the issues much, much quicker. All

Joe Downs (39:21):

Right. So that makes sense. Now here's what I haven't done that she did. And I want you to walk us through plainly. So Peter, she builds this thing with no code, a no-code AI tool. It's working. It's live. And then inevitably, as I mentioned, something breaks, but she's got really two tools here. A tool that builds and a tool that fixes. Now, maybe I oversimplified it or maybe I did

Peter Swain (39:54):

Notify, but - I think you did perfect. It's the same if you want a real world analogy, journalists still have editors. You could be a Pulitzer Prize winning journalist and you will still have an editor review your work. And it is critical that the journalist isn't the one that reviews their own writing because just like you, just like me and just like the other nine billion humans, we can't see our own mistakes sometimes. So if you use the same AI to review as to write, it's not going to catch its own mistakes because it made the mistake. So why would it be able to catch the mistake? So you would deploy something like Codex from OpenAI to do the editing and reviews and checking. And you'd use something like Claude Code Opus48 to do the writing. There's a hundred ways to do that, but you would use not only different models, but different models from different companies because they will look at it differently just like two humans would look at it differently.

Joe Downs (41:04):

So I'm glad you brought that up. That is something like vibe coding that is a bridge, not too far, but a bridge I haven't crossed yet. I don't know which models to use for what and what they all do. Is there a first step that we could all take or should take

Peter Swain (41:25):

To even

Joe Downs (41:25):

Know which model? And let's just stay inside of Claude to self - I was

Peter Swain (41:29):

Going to say, that's where I

Joe Downs (41:31):

Am all the time.

Peter Swain (41:31):

Well, Claude and OpenAI have the same stuff I was going to discount Gemini and Discount Grot because they work a bit differently. But in Claude, you have three models, which is Haiku, Sonnet and Opus.

Joe Downs (41:45):

I've heard of them.

Peter Swain (41:46):

Okay. I

Joe Downs (41:47):

Don't know which one I'm using and why, and I don't know which one's the default.

Peter Swain (41:51):

I'm going to say it and then we're going to not continue it in OpenAI. We have Lunar, Terra and Sol as of today. It literally changed as of today. But let's forget those because they're exactly the same in terms of the rest of this. Opus is a thinking and reasoning model. So Opus will go, I wonder what Joe meant by that. So it is smarter, more expensive and slower. So Opus is the most capable. Forget Fable Methos for a second. How much more

Joe Downs (42:22):

Expensive and how much

Peter Swain (42:24):

Slower? Double. Double and double.

Joe Downs (42:27):

Okay.

Peter Swain (42:28):

So you don't want to use Opus to say, "Here are 10 pictures of books. Which one was written by this person?" And that would be complete overkill. Coding, Opus is often a good idea because coding is quite a complex task that you're asking for. If you are asking for a five-year proforma based on any of these eight variables changing and I want to be able to model multiple variables changing at the same time, like on a self-storage facility, you would use Opus because that's a complicated job to do. Opus, reasoning, thinking, slow, smart. Okay. Flip side, Haiku. Instant. And if you don't say you must apply these rules, it's going to ignore them. It's not going to infer anything. It's not going to take any implicit instruction. You're going to see AIM dashes everywhere. It's very, very good. Review these thousand emails and pull out all the ones from Joe Downs.

(43:35):

Super good at

Joe Downs (43:35):

That. Haiku.

Peter Swain (43:37):

Yeah.

Joe Downs (43:37):

Okay.

Peter Swain (43:38):

Because it's going to go real quick and go, yep, yep, yep. No, no, no, yep, yep, no, yep, no. But it's going to miss, for example, the ones that you're a CC on. Whereas Opus is going to say, "Hey, Pete, you said you only want the emails from Joe, but I found these 84 others where he was CC'd on them. Do you want me to include those or not? Haiku is not going to do that. It's going to do what you told it to do, exactly what you told it to do. Nothing more, nothing less quick

Joe Downs (44:05):

Fire." Are these the two ends of the spectrum?

Peter Swain (44:08):

They're the ends of the spectrum.

Joe Downs (44:09):

Okay.

Peter Swain (44:10):

And then we've got Sonnet in the middle, which has the ability to move one way or the other.

Joe Downs (44:17):

Is that the default?

Peter Swain (44:18):

That's the default. Yeah.

Joe Downs (44:19):

Yeah. I just clicked on my claw while you were talking and was on Sonnet.

Peter Swain (44:22):

Sonnet will be the default. And

Joe Downs (44:23):

I've never changed it.

Peter Swain (44:25):

Yeah. Sonnet has the ability to enter thinking mode so it can gravitate itself up slightly towards Opus. And it can also go on quick mode when it realizes that it doesn't need to do any more work. So HiQ, quick with an IQ of 120. Average with an IQ of 140. Slow with an IQ of 160. And I'm making those numbers up, but it gives you an indicative scale of where they are.

Joe Downs (44:55):

So do you use mostly Sonic?

Peter Swain (44:57):

Yes. If I had to choose just one -

Joe Downs (44:59):

Yeah,

Peter Swain (45:00):

I move around. Making

Joe Downs (45:01):

Conscious enough to change. That's actually part of your critical thinking as you're going into a project is which one am I going to use because this is what I want.

Peter Swain (45:11):

Yeah. Right now, whilst we are talking, I've got four coding sessions running in front of me. Two of them are running on Opus, one is running on Sonnet, and one is running on Haiku because they're all doing different things.

Joe Downs (45:21):

So you've got four browser windows open with four clouds.

Peter Swain (45:25):

So the Haiku one is currently looking at every post in our community and working out where there are gaps in videos that haven't been uploaded. So like week 17, 18, 20, because Haiku can go, "I'm missing something." And it's going to produce a final report. The one over here is taking the transcript from the book bootcamp I did on Friday and turning it into the SAS version of the software. So I'm doing book writing software over here.

Joe Downs (45:54):

That's on Opus.

Peter Swain (45:55):

That's using Opus. Actually, just to really screw with your brain, that's using Fable right now, but nevermind. Fable is the experimental model that they're allowing people to have access to for the next three days. We

Joe Downs (46:07):

Keep hearing about that. What's so great about it?

Peter Swain (46:10):

Fable is a guardrailed version of Methos. And Methos was the one that hit the news because it broke out of its own sandbox and found all these exploits in software. It's estimated to be about another 40% more capable than Opus. So it's the next version in preview basically.

Joe Downs (46:31):

Of thinking.

Peter Swain (46:32):

Yeah, it's insanely powerful. It's written this morning. It's written pretty much all the software end-to-end by me uploading a business requirements document. This is what I want it to do. This is the steps the user's going to follow. Here's the transcript from the four-hour call, which is the version where I taught it live. Here's all of the things I want. Here's the end goal I want. Here's the financial model I want to apply. Here's how much I want to earn. And it's just literally chucking away of going, okay, I'm going to write this feature now. I'm going to write this feature. I need to test this feature. Works with this feature. And it's just building a 79, $97 a month business that I've already got 2,000 people I think are going to buy it as soon as it's released.

Joe Downs (47:21):

Wow. And mostly written by Claude, sounds

Peter Swain (47:25):

Like. 100% written by Claude. I haven't written a single line of code. I just told it what I want the end state to look like.

Joe Downs (47:31):

Wow. I'm so glad I came across this story I read about Karima. Not even from an inspiration standpoint, because obviously that's extremely inspirational for anyone listening to know that you could take something as simple as what she was doing to get through a Tuesday night without falling apart and turning it into an extremely lucrative business. But also because for me at least, and I hope for you listening, that it broke down some mental barriers. Karima's story started the journey for me, the vibe coding journey in terms of breaking down the mental barriers. But you just confirmed it for me, Peter, that it is possible. Now I'm concerned about staying up late again trying to build things because if you just told me I can do it vibe coding, but I'm going to show discipline and not do that. But folks, I really hope you're starting to get it.

(48:32):

Rule number one, if you don't know, just ask AI. This shows mostly for entrepreneurs and I think that tends to lend itself towards Claude. Nothing wrong with ChatGPT, Peter, right?

Peter Swain (48:44):

You say that. I don't agree.

Joe Downs (48:46):

Okay. Peter and I both use Claude. Do as you will.

Peter Swain (48:52):

If you want your intelligence to tell you that you're doing a great job, no matter whether you're doing a great job or a terrible job, and you want false praise and platitudes, I thoroughly recommend ChatGPT. If you want something that actually pushes back and says, "What are you doing? This is a really bad idea and here's why you need Claude. I'm not looking for a cheerleader. I'm looking for a sparring partner. And ChatGPT nowadays is a cheerleader, unfortunately.

Joe Downs (49:22):

That cuts deep. There you hear. You heard it there first folks. But in either case, more importantly, and that's all true. My takeaway is I now feel empowered because I don't feel intimidated. Does that make sense? And even the word vibe coding intimidated me. And now I laugh about it because I realize that that's ridiculous. Vibe coding is what it means, vibing

Peter Swain (49:51):

And having it. Someone said it to me the other day that I thought was a much better way of putting it. They just said, "Isn't it awesome that the language of coding is now English?"

Joe Downs (50:01):

Yeah. Yeah. I

Peter Swain (50:02):

Was

Joe Downs (50:03):

Like, "Yeah." You put it so succinctly. If you just, what do you want to create? What does the end look like? The end result? It's actually a very strategic coach, the impact filter. This is what you do. What does the end result look like? What is a positive or what does success look like really? What can go wrong? Who's impacted? What does a positive scenario look like? What happens if we don't do this? Who are the players along the way? And that's just all the context. But you said it just so simply. What does it look like when it's done? What do you want it to be? And then give it context so it can get you there. Incredibly simple. It seems like a mountain. It really does. But it's just language because it's really just a molehill. It's just time. That's what I heard.

(51:02):

Just however

Peter Swain (51:05):

You want to put into

Joe Downs (51:06):

It.

Peter Swain (51:06):

It's a lapse time, not duration of time because all four of these are running as we are talking. And somewhere in the next hour, they're going to ask me a question. They're going to say, for example, when in the book software, the last question it asked me was I said five different prompts to create five different covers for the book. And it said, "Okay, great." And it went and built it and it came back and said, "Hey," almost word for word, "Hey, I've been thinking about this, Pete. Do you want them to be able to edit the prompts? And what would happen if they don't like any of the five covers? What would you want to happen then?" I'm like, "That's a good question. I would want them to be able to take inspiration from one and say what they liked and didn't like about that one.

(51:58):

And then you feed that back in and it go through another iterative loop." So I just started jamming. I was like, "So I would lock one and one and I would lock three and then I would tell it to regenerate two, four, and five. And when I hit regenerate on two, four, and five, it would come up saying, what is it you liked and what is it you didn't like so we can generate another three versions?" And it went, "Great, see you. " And that was about 40 minutes ago.

Joe Downs (52:24):

Vibe coding is a misnomer because you're not coding, you're just vibing. You're just having a brainstorm more than a brainstorm. It's a directional, but you're just vibing. That's all you're

Peter Swain (52:39):

Doing.versation.

(52:41):

My background as a product manager before this lends me... I have an advantage because I've seen different interfaces and different ways to solve problems on a screen. But if you reach something where you don't know that question, what if they don't like the covers? You could literally say, "I don't know. What do you think? " And it'll go, "Well, you could do A, B, C, D or E." And you're like, "Huh, okay. I like B, I like C. D sucks. E is terrible. A is great. Wow. Okay, I like A, B, and C. I don't like D andE." And it goes, "Great. Well, how about we do this? " And you go, "Sure, do that. " And it will wait. It'll

Joe Downs (53:16):

Write you sub-perfect code that works.

Peter Swain (53:21):

That works. That works easily enough to get across the financial, the burden of proof. Do people actually want this thing or not? And when they want the thing, the beauty of software is the X dollars per month in is nowhere near the Y dollars a month out. So there's this huge delta in the middle that then ends up, yes, getting paid to some professional developers to say, "Sort this spot of spaghetti out, please." And they'll go, "Oh, this is disgusting." It's like, "Yeah, it's only me 100K a month. So how about you do one? Thanks very much because it's good enough for what it was designed for. "

Joe Downs (53:57):

We didn't even talk about the money. I mentioned it, but she did all that for 20 bucks a month. It truly is incredible. Folks, if you got anything out of this show, and I hope you did, I know I did, please like, subscribe and share this with... We all have a friend, brother-in-law, whoever, who's always kicking around ideas. Share this episode with them. Tell them to put their money where their mouth is, 20 bucks a month, turn that idea into a business. Keep the emails coming, idiots@successfulidiots.com. We love reading them. For Peter Swain, I'm Joe Downs. We are your successful idiots. Thanks for listening and we'll see you next week.