The Vibe Coding Expert: The Fastest Way To Build & Launch $1M Al App SOLO (Claude + ChatGPT)

1. 0:00 – 2:38

   Intro

   1. CRChris Raroque0:00

      I genuinely think if you're a beginner or you're just kind of starting out programming, Claude Code, it is probably the most human-like AI coding tool that I've ever experienced, and it produces the best results compared to any other tool that I've used

   2. CJCalum Johnson0:13

      You said, "I've been working on this," talking about Subscription Monster, "casually for three days, and we already have a brand, a working prototype, and a really rough business plan"

   3. CRChris Raroque0:23

      Here's the landing page for Subscription Monster. So it's a very simple landing page, but this is a real screenshot from the product I ended up doing in those three days. This can probably be, as a solo developer with just AI, a $50,000 a month business. I think if you really push it, maybe this can get to 100,000, 150,000 solo

   4. CJCalum Johnson0:43

      Okay, so today's guest is Chris Raroque, and let me say this. A week before we recorded this episode, I got on the phone with him, and he said something that stuck with me. He goes, "When I made my first app, I didn't charge a dollar for it for two years. Literally thousands of people were using it, but I didn't think I was allowed to make money from something that I built." And so the moment that Chris built up the confidence to create a paid version of his app, everything changed, and Chris has built multiple profitable apps since then. And the cool thing is, is that he did that with no background in coding or computer science. He literally studied economics at college. That's why this episode matters because we're in this rare moment of time, this golden era where if you're a builder with ideas and inspiration, you can build full apps completely solo. And so in this episode, Chris shares the full blueprint he uses from coming up with the idea and the brand to actually starting to build it on Claude Code. He leaves nothing off the table. One more thing before we get into it. This is the most ambitious AI build that we've ever filmed. To help you follow along, we also created a supporting document that goes step by step through the Claude Code section of this video. You can download it in the description below. But anyway, let's just get into this. Welcome to The Calum Johnson Show. You're an entrepreneur who's self-taught. You didn't have, like, a background in coding or as a software engineer, and despite that, you've built several, like, profitable apps using AI. And so it's not even just the fact that you've built them. You literally show people, like, every part of the process as you're doing it. And so here's where I want it to begin.

2. 2:38 – 6:14

   Why non-coders can build real apps now (AI makes it possible)

   1. CJCalum Johnson2:38

      For the person that's sitting at home and they want to build something, whether it's an app or product, whatever the thing is, and the thing that's holding them back is this belief that, you know, "I'm not a coder. I don't have this experience. Like, this just isn't possible for me," for that human being, what is the value to them of listening to the conversation you and I are about to have?

   2. CRChris Raroque3:02

      Yeah, I, I think that... And that is the exact person that I want to be able to speak to. It's that you definitely can. Um, it's very possible, and honestly, now more than ever. I, I used to say that it was possible like years ago, 'cause again, I did not have a coding background. I was supposed to go into economics and finance. If I can do this, you can do it. Now, because of AI, because of this year, I truly believe that statement is even more true. Um, there's gonna be a lot more builders, and I think after watching the episode, you will hopefully come to that same conclusion as well.

   3. CJCalum Johnson3:36

      Nice. That's awesome. You know what? I wanna, I wanna make it tangible for people and, and really help them understand kind of what you've been able to build coming from a place where, like you said, you literally studied economics in university. You didn't study computer science. If people take this episode in and they're able to kinda integrate some of the lessons and the value that you give, can you kind of-- how does this show up in their life? Like, what's the-- what is the impact been-- what has the impact been to you of learning some of these things, some of these AI tools that you're now using?

   4. CRChris Raroque4:15

      Yeah, I, I think it just opens up a whole new world because at the end of the day, I feel like what I do, and what a lot of entrepreneurs do, and what a lot of startup builders do is they're-- they just solve problems. That's what they do. I feel like if you open up your mind to what's possible with AI tooling, um, and what you can do with AI tooling, it really opens up almost this other dimension where you're gonna go about your day and you're probably going to identify problems, and you're probably going to be just thinking to yourself, "Okay, I could build something to solve that," or like, "Wow, I should start a company about that." I think after you watch this video and if you apply some of the stuff that I'm talking about, you're probably just not gonna see the world the same way. You're not gonna see your job the same way. You're not gonna see just, like, j- just your day-to-day routine. You will probably visualize it in a different way and-- 'cause that's, that's the way I visualize my entire life now. I'm just constantly being exposed to these pain points where I'm like, "Oh, I think I can solve this," or like, "This, this could be, um, you know, this could lead to some-- I should do an experiment when I get home. Like, I should try to solve something." That's, that's what I hope people get out of, and I think it's, it's impossible not to when you go down this path.

   5. CJCalum Johnson5:21

      Yeah. No, it's so cool. You know what? We, we had, um, context with people. We had a conversation before this, and, uh, one of the things that you said, as soon as you said it, I actually wrote it down. I wanted to bring it back to you. You said, "I know exactly where a lot of people are right now. They feel like they can't make things and then also make money from doing it. Like, why would anyone do that? And they have this imposter syndrome. I want people to see that they can just build things and come to that realization way faster than I did." Can you, can you just take us down kind of memory lane for a second? Like, tell me that moment When you realized, you're like, "Not only are we kind of in this golden era where you can just build things even if you're non-technical, but it can also make money from the things that you're building."

3. 6:14 – 9:03

   First dollar online: the personal CRM story + imposter syndrome

   1. CRChris Raroque6:14

      Yeah. Th- this is probably the story that you're talking about is, um, when I made my first real dollar on the internet from one of my products, and it's a product that I made for free. I just... It was, again, to solve my own pain point. Um, for people who aren't familiar, it's called a personal CRM. Um, basically, it's just a way better version of a contacts app on your phone. But I needed it 'cause I was doing a lot of networking. I was gonna go into finance. So that's the first app, the first serious app that I built, but I did not monetize that for maybe, like, two years, and I think that was because I had this sort of imposter syndrome where I kind of thought, "You know what? No one's gonna pay for this." And I just felt bad charging for it. That, by far, is the number one lesson that has stuck with me probably over the last few years. Like I... That is the most memorable thing because once I did charge for it, like, there was just this moment in my head where I was exposed to all these entrepreneurs online who they were making money. I was like, "Okay, maybe, maybe I can try to do this," but it was really hard. The minute I did it, it just opened up, it just opened up my mind that I had been wrong this entire time. Or like, "Why have I not been [chuckles] charging for this?" And it really made me start valuing my work again. But it's also the number one issue that I keep seeing with builders who I'm talking to is they're just not valuing the work that they do, and they're very scar- scared to charge for it. And I completely get it because when you build something, you put it online, you're, you're kind of almost embarrassed to charge for it 'cause you're like, "Okay, anyone could've built this." Like, that's the feeling you get when you build something. You're always like, "Oh, of course anyone can do this." But in reality, that's not the case. Like, not everyone can do what you did, and not everyone did do what you did, so you should be able to charge for it. But that, that's been such a big lesson for me and, and every time I launch an app and I do it, I really get more confident about myself, and I get more confident in charging. And it's not even really about making money, to be honest. It's more of just kind of really proving to yourself that you are creating value for the world and for other people. But by far the number one lesson I learned.

   2. CJCalum Johnson8:16

      Yeah. No, you're right. It, it, it's special. You know, before, before we get into the actual AI stuff, since that moment, so since that moment of you turning on and, like, starting to charge, can you share some of the things that you've built, like the most notable apps that you've built? Can you kinda just give people... Even if you want, you could share your screen. Like, can you show people kinda some of these different things that you've done?

   3. CRChris Raroque8:42

      Yeah, definitely. Um, here, let me pull some up here. Let me share my screen. Um, so here, here are the top three apps that I have. I'm, I'm working on another one which is still in, still kinda being worked on, and then we have another one that is probably coming because of this podcast. But, uh, the number one app that I have, the most notable one, notable one is called Ellie.

4. 9:03 – 12:22

   Ellie demo: time-boxing that thousands use daily (with paying users)

   1. CRChris Raroque9:03

      So this is a daily planning app, and it really stemmed from this pain point of I just needed something better than, um, Google Calendar to time box my day. 'Cause I was literally just using Google Calendar. Um, I was just... I mean, e- everyone has probably gone through this. They've used some sort... Like, a bunch of different systems. As a coder, as someone who can do this, I was like, "Okay, I think, I think it's time for me to just build something myself." Um, so this is number one. This is probably the one most people know me for. I think it has around, um, [lips smack] three to 4,000 people use this thing every single day, and like five to 10,000 people use it every single month, and it's got, like, over 50,000 people have, like, signed up to try this thing. It's got over 1,000 paying users. Um, but this is the one I'm, I'm probably the most proud of because the emails that I get from people, the impact is just so tangible, where people have told me, "Hey, I have ADHD, and this really impacted the way that I go about my day." Or a family who contacts me and says, "This has really helped me and my kids manage our time and our schedules." So this one is, um, this one is really great because I just, I get so much from all the, the... I, I get so much from seeing the impact that it's had from the user emails. Um, another one is this budgeting app, Luna. So this, I mean, I mean, the number one reason I'm proud of this one is 'cause I named it after my dog, and we'll get into mascots and stuff. But this is a, in my opinion, a very beautiful way to do manual budgeting. So on purpose, there's no automation, there's no banking integrations, because I just needed something to really curb my spending, and I needed something that would actually change my habits. And this is another app that I'm... honestly wasn't sure if a lot of people would use because manual budgeting is pretty painful. But I've been also surprised by a lot of the emails where people have said, "This is actually the only thing that has clicked with me so far. This is the only app that I've been able to consistently use to actually keep track of my spending and really get it down." So this is another app I'm really proud of. I'm also very proud of the design. Uh, this was a project for me to really push myself as a designer, so I really worked on the complex interactions, the animations, all that stuff. Very proud of how polished this ended up turning. And then the other app which I was talking about earlier, Mogul. This is the first serious app. I built it years and years ago. Um, I... The design has not changed for maybe, like, five or six years. I'm still surprised that it, it still has kinda held up a little bit here. But this one is very... I'm, I'm proud of it 'cause it's the first one that's made, it made its first dollar. There's still hundreds of people that use this every single day. I'm very surprised 'cause I don't do much marketing or anything, but people still use this one. These are the three major apps, and obviously I have some other ones. But I also wanna show this to kinda give people a sense of when people think of AI apps, when they think of vibe coding, they kinda just think of... The, the term I keep hearing is AI slop, where they're like, "Oh, these are just, you know, coded apps that were thrown together in 24 hours and don't really contribute anything to the world," or they all look the same. Now I'm using AI. Like, I think Luna, this is the first one that is fully AI coded and- I really want people to see that just because you're using AI does not mean it's just this useful or this useless toy that was produced in 48 hours and doesn't do anything. You can build some very serious apps. And I think Ellie's another testament,

5. 12:22 – 13:50

   “AI slop” vs serious software — how Chris ships with AI (fast)

   1. CRChris Raroque12:22

      too. It has an, a- an iOS app, it has a React app. I did start coding it before I even got into the AI stuff, but now I think 20% of the code base is now just fully AI generated and, and all the future code that I'm doing for Ellie, all the future features, it's all AI generated. So this is another testament, I think, to how far you can push with this stuff. It doesn't have to be the stereotypical, okay, AI slop that you see out there. So those are the, those are the main apps.

   2. CJCalum Johnson12:48

      Yeah. You know, we- we're gonna, uh, get into it, 'cause we're gonna show people literally how you can go from nothing, like not even an idea, to, like, building your first prototype of an app, uh, using Claude Code. And, and, and one of the things, and even it just came through as you explained it and you went through that, is I love that process of how you even went from, like, a personal problem, so, like, a personal pain point that you were having in your life, to then actually building a solution to that, that now thousands of people are using. Like, I just think that that process is so cool. And so you know what? Before we get into, uh, how people can actually do it themselves, one of the ... I was actually ... I was looking through your YouTube this morning. The most popular video on your YouTube is titled, "I Was Wrong About Claude Code." And so, like, over a hund- over 260,000 people have watched that video.

6. 13:50 – 16:37

   What is Claude Code (and why Chris switched from Cursor)

   1. CJCalum Johnson13:50

      And so to begin, like, when you say you were wrong about Claude Code, let's start right in the beginning. First of all, what even is Claude Code? And then what were you wrong about?

   2. CRChris Raroque14:03

      Yeah. So Claude Code is, at the time, it was one of the newest AI coding tools out there. The number one, at least at the time of recording, was Cursor, and that was the one I was using. That's the one most people know about. There's a bunch of AI coding tools, but Claude Code, in my opinion, at least at the time of we're recording this, is, in my opinion, the number one tool. It is probably the most human-like AI coding tool that I've ever experienced, and it produces the best results compared to any other tool that I've used. But it, it is ... It's kind of complicated to explain what it is to people who are non-technical, but really it is a terminal-based tool, which means you access it in the terminal, which that, that alone just threw so many people off, and that was another reason I dismissed it, to be honest. But you go in the terminal, you just type in commands, like whatever you want, and it just goes and executes things. This went against the paradigm that a lot of people have about coding tools, because you don't see the code. Like, you just tell it to do something, and you don't get to actually ... I mean, you, I guess you can kinda see the code, but there's no code editor. Like, you're not supposed to edit the code when you're in Claude Code. Um, and I think that threw people off, but it's this ... It's, in my opinion, also one of the first real agentic coding solutions. Because again, most coding solu- most AI coding solutions are you go tell it to do something, and it'll go write the code for you. You review, and you kinda have to go back and forth. Claude Code is the closest thing where it's a true agent where, and I give it a couple more, like a year or two, where you can literally get away just not seeing the code, which is a very bold statement, but it is ... That is, this is the tool that has gotten this close, where it's literally almost like [laughs] a human programmer. And I think a lot of technical people who have used it and understand will probably agree with me, too.

   3. CJCalum Johnson15:52

      Okay, you know what? I'm gonna, I'm gonna be that person. I'm gonna ask the, uh, the stupid question. And, um, when I say the stupid question, it's because I have never coded, and so I'm gonna, I'm gonna represent that viewpoint and ask it to you. When you say, um, and I can feel the significance of you saying it, of like, "It's a tool that you can use and you'll never see the code," give us the context. Like, why is that such a big thing? And even when you say it's, like, human-like, the most human-like coding tool, like, what is the significance of that? Okay, quick break. I just wanna give you guys some value. So when I first started experimenting with AI tools, my biggest challenge wasn't building. It was figuring out what

7. 16:37 – 18:17

   Sponsor break: Free AI App Builder Kit (idea → spec → launch)

   1. CJCalum Johnson16:37

      to build. And so in the beginning, I found the toughest part was actually landing on that final idea that I could then go all in on. And from reading the comments, I know that I'm not the only person that's had this problem. I partnered with HubSpot to bring you this AI App Builder Kit. It's completely free. I've been so excited to share it with you guys. And so let me just share a bit more. It's basically a playbook of done-for-you prompts that you can drop straight into ChatGPT or Claude or whatever is your favorite AI tool, and it will help you as you build your AI business. And so what you'll find inside, it has these different prompts that can help you with anything from generating app ideas that people will actually buy. It can also write a full technical spec that AI can then code, and it can create a launch plan that gets you your first users. It's honestly a game changer. And so for me, my favorite part of the kit is the ChatGPT guideline template. It took me less than five minutes to fill it out, and then you drop it into ChatGPT, and it will make sure that every response directly matches your brand voice. Honestly, it's helped me draft and publish content way faster. And the good news is this kit is completely free. So all you're gonna do is you're just gonna go to the link in the description, you're gonna download the free resource, and you're gonna get started. So thank you to HubSpot for partnering with me to create this. Let's get back to the episode Okay, so we're gonna get back to the video in 60 seconds. But first, today's sponsor is gonna help make sure that your finances stay on track. Introducing our friends at Rocket Money. Let me know if you can relate to this. You decide that this is the year that you're finally gonna get your finances

8. 18:17 – 21:31

   Sponsor break: Rocket Money — track spend and kill wasted subs

   1. CJCalum Johnson18:17

      right. You set out all of these clear financial goals, you build a spreadsheet with your budget on it, and then two weeks later, you have already lost track of your spending. And so at the start of this year, I made a change and I started using Rocket Money. Rocket Money is a personal finance app that lets you manage subscriptions, lower bills, create a custom budget, and grow your savings all in one place. And so first of all, I linked all of my accounts, and so it pulled up every subscription that I was paying for. And so with just a few taps, I canceled the ones that I didn't need. But the big win was actually the budgeting tools. I can now see every expense in real time. I can get alerts before charges hit and actually stay on track with my spending. Plus, its Goals feature automates deposits towards whatever I'm saving for. Members who actually use all of Rocket Money's premium features save up to $740 a year. And so if you wanna give it a try, go to rocketmoney.com/calumjohnson. That's rocketmoney.com/calumjohnson. You can also just go to the link in the description of this video and get started using Rocket Money today. Thank me later, and let's get back to the episode.

   2. CRChris Raroque19:33

      I think the significance is it is the first-- Th- this is the first time that I think a lot of programmers who are, or, or just in, in general, not even programmers, a lot of people who are like, who, who say, "Oh, to use these tools, you need to know how to code. Like, you actually need to know how to do this." This is the first time where I'm starting to see that that might not be true anymore. Like, you actually might not need to know how to code to truly use this stuff. Although I will say it does help a lot. It really helps so much if you know how to code, but I think you can get away, because again, you don't need to review it. They don't even want-- They don't even care if you review the code. Like, you can, you can go ahead and just, just let it do its thing. That really unlocks this whole, again, the whole world we were talking about before, this really unlocks that whole thing for, for people. Um, so it- it's gonna have a huge impact, uh, both positive and negative. I think to some, a lot of developers, yes, this might actually really impact the job market quite a bit if this continues, if more tools like Claude Code continue. But I think this does open up a huge door for a lot of non-technical people who are just getting into it, a lot of industries where there's not a lot of software engineers. It's gonna be such a huge shift, I think, when this type of AI tooling gets even better.

   3. CJCalum Johnson20:54

      Yeah. You know, one, one of the things we're gonna talk about is the, the newest app that you're building. You're literally building it right now. Um, Subscription Monster. So I-- one of the things that I even heard you say in, in the video, um, I Was Wrong About Claude Code, you mentioned obviously that you used to use Cursor, but basically since you've started using Claude Code, you've kind of, you haven't really looked back since then. So in building, like, the new apps that you're currently doing, like Subscription Monster, that's primarily being done on Claude Code.

9. 21:31 – 29:14

   Subscription Monster: the problem, constraints, and first signals

   1. CRChris Raroque21:31

      Correct, yeah. Uh, o- only exclusively on Claude Code now.

   2. CJCalum Johnson21:35

      Yeah. You know, let, you know what, let's just get into it, 'cause I think one of the things I enjoy about your content, and I said this to you even before, is that not only do you build the thing and you're the person that's actually doing it, like the non-te- the non-coder who's actually doing it, you also show every little piece of the process. And so right now you're building an app called Subscription Monster, and you've literally shown people how you're going from, like, coming up with the idea to actually building the product. And one of the things that you said in your video is you said, "I've been working on this casually for three days, and we already have a brand, a working prototype, and a really rough business plan." And so that's actually where I wanted to begin is, like, even coming up with the idea, like, even just sitting down and thinking to yourself, "Okay, I'm gonna start. I'm gonna build a new app." What in your process, what in your mind was the step one? Like, where did it begin?

   3. CRChris Raroque22:42

      So the, the number one, the number one step for me when I build an, any new app now is, okay, what is, what's the purpose? And I thought, you know what? There's been this pain point in the back of my mind. Recently, I had to a- answer the question: How much are we paying in software subscriptions for my other business? The answer is in the thousands of dollars, but I, I, I just hated that I couldn't answer that accurately. We do have some software, like some of our credit card stuff. We have QuickBooks. Like, it can kind of answer the question, but not well. And so I thought, you know what? Maybe this is actually a good enough problem because I, I really feel like I, across, like, multiple business bank accounts, I just... I can't answer this, but it feels like this should be a really simple thing to do. And then I did some quick googling, did some quick searching and confirmed, yeah, nothing really exists like this, or at least the way that I'm envisioning it. Okay, there's a signal that this should be done. Although I did have to do a little bit more research, so we can talk about, cool, how did I confirm that this is the right thing? But that's how I got the initial idea. Again, it's just, why can't I answer this question? Like, again, a couple of weeks ago, I had to answer this question myself. I, I couldn't even answer, like, how do we list the business subscriptions? The way I did it was I had to go physically print out every single bank statement and go line by line and try to figure it out. Y- I even tried feeding this into ChatGPT and some other things, and it just wasn't doing it well enough. So I had to just, I just had to go in and just kinda manually go through the PDFs, have it extract things, and then write it literally in an Excel sheet. And so it's like a Google sheet that I have. So that's how the problem, that's how I identified the problem. Again, seems small enough because I'm like, how is this not... I mean, we have AI, we have all these things. How is this not, how is this not solved right now? Um, and then I did the process of figuring out, okay- Number one, why, why does this not exist? Um, and then what, what's my game plan? Like, if I'm gonna do it, is the-- and does the game plan fit in what, like for this show? Like, for, for the constraint I have, and my time constraint was this show. And some people's time constraint is, "I have a full-time job," or, "I only have three months to make money." So the constraint is actually not ... Like, that's something a lot of people don't factor in, but I think it should be factored in because if you don't have that constraint, you might accidentally be building something for years and not releasing it. Or you might be building something way too small and it's just not ambitious enough or it's not going to get the financial outcome that you want. So putting the problem and the constraint together, that's usually the first step. That was a huge long-winded [laughs] answer for that, but hopefully that helps.

   4. CJCalum Johnson25:08

      No, that, that's actually really important. The, the part about the constraint is really important 'cause I've spoken ... And, and I used to be like this, but I've spoken to so many friends that, like ... And sometimes it's not even that they're building something, it's like they're creative, but there's no constraint to it, so they never release. Like, that, that was literally the problem that I had with the first two businesses that I had is, like, there was no, there was no deadline. There was no time when like, "Okay, this is gonna hit," and so it just went on and on and on until you lose momentum, and then it dies there. And so I guess my, my question for you is, like, do you kind of actually have a process of, like, tracking pain points and, like, uh, almost like problem areas in your work life that can then almost be spun into these AI app ideas? Or does it just spontaneously kinda come? Okay, so give me 60 seconds. I don't usually do this, but I wanna take you guys inside my apartment for just a second because there's one service that is now a core part of my daily routine. That is the new Alexa Plus from Amazon. So let me say this. A couple nights ago, I'm winding down in my living room. I've just finished work. I'm trying to get organized for the next day. Instead of reaching for my phone or my laptop, I just say to my Echo Show 15, "Alexa, what's on my calendar tomorrow?" Next thing I know, done. It tells me what's coming up tomorrow. I then ask it to play music from my playlist and book a dinner reservation for the weekend. Done and done again. And so when I tell you this has quickly become the one-stop shop for everything that I need. I use it to manage my day-to-day, order groceries, book Ubers, play music. And the craziest part, it actually remembers what matters to me, how I talk, and handles multiple things in one go. That's because Alexa Plus is actually powered by generative AI. In other words, it's smart. It's more conversational, and it can actually follow your train of thought without missing anything. It's like having a calm, capable assistant in the room with you. I kinda think about it like JARVIS from the Iron Man. And so anyway, if you wanna try it, Alexa Plus is now rolling out for free through early access, and after early access it will still be free for Prime members. So go to the link in the description to request access. That's the link in the description to start using Alexa Plus today. Thank me later, and let's get back to the episode.

   5. CRChris Raroque27:35

      It, it's now ... It's definitely more spontaneous, so I don't track it, but it, it's almost impossible now for me not to kinda store them in my head because I just see some things enough. So for example, off the top of my head, the number-one pain point I also have ever since doing content creation is I hate doing the rough cut of the edit, 'cause I have four hours of footage that needs to be condensed down into 20 minutes or something, and I'm so confident that AI can go extract a really good script from the transcript and do it. CapCut has kind of a little feature for it, but it's just not good enough and, like, this ... But the friction happens to me every single week as I create content. Every time I encounter some sort of pain point, like the business ... Uh, like asking business subscriptions what are we paying monthly for. Every time it happens it, like, carves something in my head where I'm like, "Man, I, I gotta do something here." And then eventually some of those actually do kinda pop where it's like, "Okay, this pain point is too great. I need to do this." To be honest, the budgeting one was one of those as well. It just was grinding at me that I know I need to be manually logging this. Like, I know I need to have something where when I go get coffee or something, like, I need an app to just remind me that I cannot afford to go do this. And that pain point just weighed down on me enough where it kind of went from just living in my head to being like, "Okay, it's, it's ... There's so much pressure here, I need to go do this now." So that's kind of what my mental model is. It's I kinda just store it. 100%, though, I think people should write these things down, and I know a lot of people do have some idea log or something, and then just maybe make a note of, "Okay, how many times [laughs] has this happened?" Or, "How many times have I been upset by this?" And then once it's enough, you ki- that should signal, okay, you should probably

10. 29:14 – 33:08

    Do this right now! (Turn these AI Ideas into $1M Apps)

    1. CRChris Raroque29:14

       go deal with this. Like, you should go build something to kinda try to solve this.

    2. CJCalum Johnson29:17

       Yeah. You know, I, I, I wanna f- for the person that's sitting at home, 'cause I'm sure even as they're listening to you and I speak, there's, like, the ideas of different things that they can build are, like, firing off, and certain people never get past that point, right? Like, it's just all ideas on the page. When it comes to, like, actually prioritizing ideas and thinking through, like, "This is the one that I'm gonna action on," are there certain types of pain points or problems that it's actually, like, they're perfectly ideal for you to build the solution using AI? Like, are there ... Like, for you, are there certain times when you hear a problem and you're like, "Oh, that's the perfect ... Not only is it a, a valid problem, it's the perfect thing for me to now go and build using a Claude code"?

    3. CRChris Raroque30:07

       Yes, I think it's ... Yeah, I actually do keep ... Like, this is one of those mental things that I think about when evaluating ideas that just kinda happens in my head. Actually, maybe there is a little tangible step here. It's I always try to think, "Why does this not exist right now? Why is there no business around?" Usually there's a very good answer to that. Even, we'll, we'll get into the Subscription Monster thing, 'cause I did some research. I actually had Claude do the research for me and analyze why does this not exist. And it came up with some really good answers to why it doesn't exist. But once you get the answer to that, and usually it's there's not enough money in the market, or it does, but it's a small feature of this bigger thing as an upsell feature for another business. There's usually some answer of why it doesn't exist. If I don't like the answer, where I'm like, "That's not good enough. This should still exist," then I'll go pursue it. That's also a good signal, too, with AI, which is how quickly can I get to the MVP here? How quickly can I build this with Claude Code? Now that I know how fast I can move, like I can genuinely build something very substantial in a week or two, that scope, that is a really big thing that I do look at. 'Cause if it is like massive where it's like we're gonna go, we're gonna go, you know, try to solve something to find new-- something with DNA and trying to discover new drugs. Okay, I don't think I can do that with Claude Code. Like that one might be a little bit... I mean, actually, you probably could, but it would be a longer term period. I would probably need to pull in a lot of experts to figure this out, and it's probably way more than just a coding problem. So if the problem is confined to, okay, I can build a really quick prototype to really test is this gonna solve this for me in two weeks, maybe a month, that is perfect for AI. That is perfect for AI coding. And then another layer, too, which is kinda meta, is can AI itself solve this problem? That's another thing that I've been recently adding to my checklist, is as a why does this not exist right now, and is it because of an AI problem? To be honest, this problem that I'm bringing to this podcast, the Subscription Monster, like why can I answer what subscriptions do I have? One of the things when I was researching is that's apparently a very hard problem to solve because let's say you have thousands of financial transactions, sometimes the banks and the credit card companies, they label it very differently, or it's like just a little bit off. That is a perfect problem where you-- before, I'm sure a lot of companies, they would have to either make custom AI models or they would have to build all of these rules around categorizing Amazon transactions and Spotify transactions. That is perfect for AI now. I think like those ideas where you're like, does this not exist because it's super tedious or v- just not financially... Like it's just really hard for a company to build this without a lot of money to make a custom model. Those are perfect problems because the current AI technologies are just looking for solution, or just they're looking for problems like that. So that's another, that is another thing that I look at when I'm determining whether to use AI or whether to literally use AI in the product. Those are some things I look

11. 33:08 – 41:03

    1 tweet → 25 signups: build-in-public waitlist playbook

    1. CRChris Raroque33:08

       at.

    2. CJCalum Johnson33:08

       Yeah. Okay, you know what? Let's, let's do this, uh, 'cause I'm getting excited. Is, um, can you actually, first of all, like pull up the landing page for Subscription Monster, which is like what you're building right now, so people can see that. And then you spoke about actually you use Claude. So I guess the distinction is there's Claude, which is like the language model. It's alm- almost like ChatGPT, and then you have like Claude Code, which is the, uh, it's the agent that it uses to actually like code things. You were using Claude to actually figure out, is this a problem that's worth solving and like why hasn't it been solved? Could you show us as well, like what do some of those prompts look like, sound like, that you're using to like validate is this idea actually worth building something for?

    3. CRChris Raroque34:01

       Yeah, let, let me show you. So, and let me pull up. Yeah, this will be really fun because a lot of people don't, don't do this. Um, first, here, here's the landing page for Subscription Monster. So it's a very simple landing page, but this is a real screenshot from the product I ended up doing in those three days, and it's just a waitlist page. So it's just a way for me to collect waitlists. I only put it in a single tweet that I tweeted at like 1:00 a.m. in the morning, and from that tweet, though, 25 people did sign up, so that's pretty cool. I think once I promote it, I'm very interested to see. But this is the waitlist page. This is, uh, it's very simple. Um, now to talk about how did I use Claude to do research. So, uh, here I am referring to literally Claude, like Claude Desktop, Claude the language model. Um, so here's the chat where I started it, and I actually said, "Is there a reason there isn't a subscription finder like Rocket Money for business?" Because it does exist for consumers. There's hundreds of these, "Let me go help you find subscriptions and cancel them," but there really isn't any for businesses. And I basically was just asking it, why does someone-- why does this not exist? The distinction, though, is I'm using something called Claude Research, which when you're using Claude, you can actually turn on this research mode. It is going to go through hundreds of different sources to try to answer your question and do research. It's not gonna respond directly to your question, but what it's gonna output is basically this giant list of, "Okay, here's my analysis." And so-- And it, it, what's really cool is it also asks some follow-up questions. 'Cause it, I asked it, "Why does this not exist?" And it said, "Let me make sure I'm answering this relevant to you. What, what, are you a solo entrepreneur? Are you a small business owner?" Which is, this is so cool that it asks these follow-up questions because I would expect a really good research analyst to also ask a bunch of follow-up questions, so this signals that, great, this is gonna be a very tailored response. Um, it made this report. I'll be honest, I was too lazy to read this entire report, so I just told it, "Can you summarize what the reasons are?" And it actually listed out some really, really good reasons like, you know, payment methods like are very complex, especially when it's international. Some people use credit cards, some people do literal manual invoices, so it's gonna be kinda hard to see in the bank, in the bank account. Um, there's also some regulations why this might be kinda challenging, and then really there's no API. The one that was really interesting to me, though, was not the technical challenges. I wanted to hear about the economic and the, the real business challenges here. And the two things that really stood out to me really were, at the end of the day- This is a product that a lot of-- Like, to acquire a business for this might be really expensive. Um, it might be really hard, and there might not be enough businesses willing to pay for something like this, where maybe you can get to a $10,000-a-month business or a $20,000-a-month business, but $20,000 a month is amazing for an individual, amazing for a solo app developer. That can barely cover payroll for a lot of companies. So that is probably why no one has really raised money for this problem. That's probably why there's no standalone products that exist. That's probably why it's just usually a subset of a n- of a larger product. So Mercury the bank or Brex the credit card, they all have this feature where they can look for subscriptions within their product. But I think that's why it's not a standalone thing. It's just the market isn't big enough for a real serious business to exist. But again, we're looking at my constraints here, which is my constraint is I just need something for the show and, you know, maybe I'll-- they'll throw something in there. I'll learn some new technologies. Like, those are my constraints. So none of those really apply to me. If my constraint was financial, where I have a family, I have to go do this, or I need to go raise money, I need investors and get a return for them, okay, this business might not make sense for those constraints. But I, I was using, um, Claude here and, uh, some-something else I was asking about, 'cause I, I think I misread this. I didn't know that it said Plaid. I thought-- I think I just didn't read that line at the time. I asked it, "Why can't I use Plaid?" And it was like, "Oh, you can." So I basically just went back and forth to ask it, you know, what, what does it... Like, why does it think it doesn't exist? In my opinion, this is kinda like the market research I would typically do, which is I would probably literally go, you know, try to network around or probably try to ask other business owners, "Why don't you use this? Would you pay for this?" Like, "What's going on here?" Um, obviously do a lot of Googling and try to figure it out. But these are the kind of questions that I'm asking, and Claude Research is so incredible just to get a sanity check. 'Cause a lot of entrepreneurs, you know, when they do this, they're just like, "Oh, it doesn't exist because no one's smart enough to do it." No, there's usually a reason. [chuckles] And this thing will be smart enough and unbiased to tell you the truth of why it doesn't exist.

    4. CJCalum Johnson38:46

       Yeah. No, the, the, the sanity check. And, and that's the thing. I think a lot of the times as entrepreneurs, especially if you're like a solopreneur, you're building something by yourself, it's like you're kind of in this, um, [lips smack] vacuum where it's difficult to get feedback or, like, have, like, that, uh, interaction with someone to kind of validate or, like you said, like a sanity check. And it's cool 'cause I do the same thing with AI, where you-- it's almost like you get in this conversation and you kind of voice to the AI what are your kind of like doubts or the things that you're unsure about, and it just helps, like, illuminate and, like, get clarity on exactly where you have to go. I think the other thing that you mentioned, um, which I really like, is having clarity on, like, what the constraint is. And so it's actually a really smart way to think about, like, which idea should I use AI to go and build is, well, it doesn't-- There's like this intersection of ideas that it doesn't make sense for, like, a, a huge business with hundreds of employees to go and build it, because to your point, the market opportunity is probably just not enough. But for, like, a solo person who's using AI and could build it over the span of, I don't know, a few days or, like, a week, you could build, like, a pretty profitable $20,000-a-month business solving a problem like this that you used AI to go and build. And it's like, it's actually-- I think that intersection's actually pretty perfect in terms of validating an idea. Um, [lips smack] okay. Here's where I wanted to go next, and I wanted to go back to that, that quote that you had. You said, "I've been working on this," talking about Subscription Monster, "casually for three days, and we already have a brand, a working prototype, and a really rough business plan." So w- we kind of went over, like, the idea. When it comes to the really rough business plan, I'm curious, like, what is happening there? Uh, like, what are you putting together? Are you using AI to do that? Like, give us the context of that business plan and why it's even valuable.

12. 41:03 – 47:17

    The rough business plan: first dollar, ceiling, SEO/Reddit/cold email

    1. CRChris Raroque41:03

       Yeah. I think, um, s- sometimes I do use AI to kinda dig a little harder. In this case, since again, this was such a small effort, such a small constraint, it's we just need this for this podcast, um, I just thought about the business plan. There are really only two things that I cared about, 'cause a business plan can have thousands of things. But the, the only two things I cared about were: cool, what is, what is the plan to go get the first dollar? Um, what is the ceiling of revenue I can probably get by myself in this constraint that I have? And then also, how am I going to market, and how am I going to actually try to acquire new businesses? And then, and then the last one is, yes, it is built off of this very small feature, this small pain point. Where am I gonna take it next? Like, how can I expand this if I choose to do it? Those are the four things that I really looked at. We can talk a little bit about kind of the, okay, what's the ceiling? We kinda discussed this a little bit. Um, this was more of just me kind of researching the market. When I discovered that there is literally only one standalone product that does this and it doesn't even... It, it's pure manual. Like, you have to just log what your transac- what your subscriptions are yourself. Like, it doesn't scan anything. It doesn't do anything. When I figured this out, it really signaled to me there's probably no money here. But how much money is there? This is just me doing normal research, looking at other businesses just 'cause I've really, like, done this a lot and from a bunch of different industries. But I really think comfortably the conclusion I came to is this can probably be, like, as a solo developer with just AI, a $50,000 a month business. I think if you really push it, really push it, really dial up the marketing, do this stuff, maybe this can get to 100,000, 150,000 solo. Past that, we're gonna need to add a lot more features. Um, so that's to answer the question of what's the ceiling here? What can I do? Um, then the next thing is, okay, what is gonna be the marketing strategy? Like, how would I start acquiring users to do this? Um, obviously social media is there, so I kind of know, great, I can probably acquire some people by doing this build-in-public thing that I normally do. But I wanted to also figure out, okay, if I had no choice, this is the only app I had, this is, this is my primary income source, how am I going to do this? Um, after doing some research, uh, like some combination of AI, like I have some other chats where I was talking to, to Claude about this, um, it's most likely going to be a combo of SEO because really no one is doing this. I will say though that it does suck because, you know, uh, Rocket Money and all these guys are competing for SEO there, but there really isn't many for businesses, so I do think there's an SEO opportunity. It's gonna take a long time, but that actually is probably gonna be one of the biggest channels. Lots of Reddit, because Reddit, I think there's a lot of people I've noticed who are also searching for this, that they're like, "Why does this... Like, how do you do this for business?" The answer most people say is just go sign up for Rocket Money as a business and just go link your credit cards and don't use any of the other features. That's the number one thing I'm seeing. So Reddit and word of mouth are gonna be massive. Last thing which I will do if it comes down to it is I am probably going to do a lot of cold emailing. Um, and that's probably very surprising to people because no one talks about that. After I talk to some businesses, like we, we even talked about it, I have been so surprised by the number of businesses that have told me, "Wait, I can't answer that question," or like, "Wait, I don't know what we're spending on. You're right." And they're like, "Wait, I, I probably spend like 20 bucks like one time just to solve this problem." [chuckles] Like, like what, what... So that really makes me think that, you know what? I think I can craft some really interesting cold emails or do some manual effort. Like I probably literally can go down the street and go talk to some businesses and be like, "Hey, I'm, I'm not even gonna charge a subscription. How about I do a quick audit? Give me like 50 bucks. Let's just link your account, and if you wanna keep it, you can keep doing it." Like I, I think there's actually a really good cold outreach for this one just because of how like visceral the reactions have been when I show business owners this thing where they're like, "I would use it." I have not seen that for a lot of products. Like this is a very interesting one. Um, but that, that's, that's kind of my pseudo... That's why I said it's kinda like not a great business plan, but it's like a very rough one, which is cool. I have the marketing strategy. I, I know how I'm gonna acquire the users. I already spun up the waitlist so I can start capturing this and, uh, that means I have my initial beta users. I already have the MVP done. I already kind of know what the feature roadmap's gonna look like, like for the next month or two. The last piece, which is I think number four, was, um, what's the feature roadmap set? After doing research, it's so clear to me there are infinite ways to take this. The fact that like Mercury and Brex and all these guys, they have subscription analysis as a standalone feature means they're using this to upsell a product, or they're using this to cross-sell other parts of their products. I can definitely do the same if I choose to do it. So I can probably hook people into subscription analysis, and if I wanna make a business budgeting app, or if I wanna make an AI app that, uh, lets you ask questions about your financials for a business, there's so many ways to take this. Just knowing that alone made me stop there, where I was like, "I have enough now." Like I have enough... I'm not gonna get ahead of myself. Let me now start building. Let me start working on the actual product now.

    2. CJCalum Johnson46:05

       Yeah. You know, it's, it's so good. I think also like a, a secret that I just heard from, from listening to you speak is we live in a world now where like everyone wants to do stuff on like social media, or now the new thing is obviously AI. But almost going the old-fashioned way and talking to users or talking to your ideal user will give you like almost like a, like a secret insight into how you should actually go and build the business, and that's what you had with Subscription Monster is like when you spoke to people about what you were gonna build, the actual target user, it was like great validation. Um, you know what? Here's where I wanna go, 'cause obviously I think you're... You do a great job at like building on Claude Code, and we're gonna show people how to do that. But I also think another one of your superpowers is in the branding and design. And so I wanna start right in the beginning there. Even when it comes to the name Subscription Mo- Monster or like Ellie and Lily and like all of these things that you've built, talk to us... I know that there's like AI that's involved in that process. Talk to me about how you come up with the name.

13. 47:17 – 59:48

    The AI App brand SECRETS! (why “Subscription Monster” sticks)

    1. CRChris Raroque47:17

       Yeah. Th- this has been one of the things that I've been surprised that people ask me about, um, because to be honest, for a lot of the apps, I was so bad at naming apps that I decided, you know what? I'm just gonna name them after my pets. So like Ellie and L- uh, Ellie and Luna are specifically named after my hamster and my dog, which is why the mascot looks this way. Um, and I thought to myself, "I'm gonna change it." Like I'll change it once... Like these are just placeholder names, but they always stick every time I do that. Um, so but Subscription Monster was extremely conscious. Like this was a very conscious one. Mogul Networking, Mogul is, was also extremely conscious. I think the original name was actually Karen, and then years later, that's not the greatest word anymore, so I was like, you know what? I don't know. Like I named it after the computer from SpongeBob named Karen. I was like super intelligent computer. That was a horrible idea. So then I named it to Mogul and thank God that one actually stuck. But for Subscription Monster and these apps, the only thing I know for certain is that I know when the name does not sound good to myself, like I would not be proud of it. And so I can actually show you what the process is for how I named Subscription Monster. If I had another pet, I would've just named it after my other pet, but I ran out of pets, so I was like, "Well, now we gotta actually come up with a real name." Um, so we can actually go re- look, really briefly, I can show you what this looked like. I actually did initially start by using the same Claude conversation that we were, that we were using for the business plan, and I asked it, um- I asked it, "Can you come up with a name for this? Like a business name, preferably one that has a .com." To be honest, I don't think... I didn't think it could actually do the domain search for me, but it did a bunch of research and it came up with some business names. And to be honest, these were absolutely horrible business names, but that's okay because I wasn't trying to get Claude to actually come up with a business name, I was just trying to get the creativity going because I, at least I know when I hate something. Like it's hard for me to come up with... It's hard for a lot of people to come up with stuff. It's really easy for people to reject things because they know what sounds bad. So that's what I was doing here, and it was interesting because I remember reading some of these and I was like, "Wow, these just sound so generic." BizScribe, Expense Radar, Subsly. Like a lot of these things sounded like generic tech companies. And, um, I was kinda like, "Okay, this isn't really, this isn't really good." I had another Claude conversation where I was like, "You know what? Maybe I'll just like start a new one so it's a little more creative," where I said, "Can you come up with some sort of mascot? Like what is a good mascot for this?" And it actually came up with a meerkat, which a meerkat is just this animal that, I guess it does a lot of searching and it's always like peeking its head and looking for things. So that's why Claude was like, "You should do this because it'll go look for subscriptions." And I thought, "Okay, that's, that's actually not a bad idea." Um, so then I continued in this chat and I said, "Can you come up with names where the mascot is a meerkat?" And it started coming up with things like WatchKat, AlertKat, WarnKat, like a bunch of these things. Sub- um, [lip smack] uh, I think MeerMoney was another one. The only thing I hated, to be honest, like I was kinda close with some of these, is I think if I named it something Kat, the Kat is a K, and I think it would've really caused some conflict because people would probably use cat with a C. And a lot of people don't even know what a meerkat looks like or what it, what it is, so I was really worried that it would, it would get confused, so I kind of stopped right there. Um, so then I thought, "Maybe I can name, name the meerkat. Maybe the meerkat can just become a mascot instead, and I'll just come up with a name." So it actually came up with some cool names like Max. I think Milo was another name I really liked, so I was like, "Oh, maybe I can call it like Milo Money, Milo something." Um, once I started getting down this rabbit hole though of being like, like it's going to be the name plus something, so like Luna, the domain is lunabudgeting.com, ellieplanner.com, mogulnetworking.com. I realized, "Oh, you know what? I, I should probably have a word there. What is that word? It's probably gonna have to be subscription." 'Cause I was thinking maybe I can get away with something Money, like Milo Money or Milo Finance, but I think it's too generic, and if I'm-- if SEO is gonna be one of my plays, I do need the word subscription in there. And so I actually asked it, "Okay, can you come up with something else?" And this was a whole other conversation where I, uh, in Claude where I said, "What are some other subscription X? Like come up with subscription X," and I told it what my plan was, what I was trying to build, and it came up with so many different like Subscription Wolf, Subscription Hound, Subscription Owl. One of the terms was Subscription Monster, and again, it's one of those things where I... If I see it, I'm like, "Oh, there's something there." When I read that, I was like, "That one actually sounds kinda cool because it's, A, very memorable," 'cause it, it's just you don't hear stuff like this in the finance business world. Like Subscription Monster just like sounds so hardcore. It just sounds so different. So that one jumped out at me, and then as I was thinking, you know, "What's a little more?" I was like, "You know what? Subscriptions are kinda like this monster that no one thinks about that's in the back of the business owner's head, where they're like, 'I know I gotta deal with this, but I, I'm just like too scared to even look. Like I don't even wanna know what subscriptions we're paying for.'" And so I thought, "Yeah, this... Okay, this could actually be a really good one." So to summarize it all though, I really just went back and forth with AI, and the important thing was I just, I didn't stop until I was fully happy with the name and, and kind of-- now I guess the mascot. So inadvertently, by going this route, I knew that the, the mascot had to be a monster. Um, so that's actually how we came up with the name. And then we can go into the branding, um, and, and how I got the logo and everything if you want to. But I don't know if you have any questions.

    2. CJCalum Johnson52:59

       Yeah, you know, s- so, so listening to you, there's two things that stand out. It's like, and 'cause I, I think it's a valid question of... And, and I've done this before when I've tried to start businesses. You come up with all of these names, and then you're like, you, you... A lot of the times you even have to ask other people. You're like, "Which one do you like?" 'Cause it's hard to make that decision. They, they all start to kind of fall into like a jumbled category together, and it's hard to make that final decision of like, "Okay, this is the one I'm gonna go with." And so there was two things that stood out. One is there's almost like a gut feeling to it, like you just read a certain name and you're like, "Okay, this is the one." But then the second thing, and this is something actually I wouldn't have thought about, is actually just like the, the clarity of it, of even when people go and search it, because obviously that's an important part of marketing is like the SEO, is, is it easy to go and search for? So the fact that like MeerKat, the way that Kat is spelt is actually like with a K and not a C, that would have lost you traffic. Like that part, the simplicity and like the clarity, I wouldn't have thought about that. It makes complete sense, but I wouldn't have thought about that in deciding the final name. But to your point, it's key. Um, and so even that strategy of literally putting what your business does in the name, like Subscription Monster, is it's memorable, it's clear, and it, like you say, it kinda s- it sounds good as well. Um, okay, here's where I wanted to go next, 'cause I think a key part of the, uh, a key part of your branding, and it's something that stood out and you spoke about it, it's the mascot. Why- And, and maybe you can even show some of the examples, like the mascots for Ellie and Luna and all of the... Like why is the mascot... That's almost like a key... It's like a staple of how you build things. W- why is that?

    3. CRChris Raroque54:59

       Yeah, it, it honestly, originally I, I, there wasn't much thought about the mascot, but I did it specifically with Luna Budgeting because, to be honest, I... It was really meant to be a placeholder. Like, I made the app icon a photo of my dog, like a placeholder image, like an illustration of my dog, just almost as a joke. Like, just as like a, "Oh, this will be... This is great." Like, I have nothing to lose, it doesn't matter. Like, let me just go do that. Inadvertently, what ended up happening was I noticed people got so attached to this, to the dog, and I kind of thought more about other businesses and other, just, like, other industries and I realized, you know what? For a lot of places, or even, you can even look at sports teams, the mascot is a really memorable thing. Like, people aren't gonna remember what the logo of, you know, I don't know. They're not gonna remember the logo of a sports team, but they will probably remember the mascot or they'll probably remember, like, the character. Like, those are really memorable, and I think the same thing applies here to businesses. I understand why a lot of businesses don't do it, because it's very hard to get right. Like, if you get it wrong, you really get it wrong. It could really turn people off. Even especially for something like B2B products, something I had to be really careful with was if I'm too cartoonish and I'm too playful, that could really mess things up because some businesses won't trust that anymore, versus also if I go a little too B2B where the logo and the mascot are just too generic, I won't stand out and I'm really worried about that. So mascots do this interesting thing where it really does set the tone for the product and, like, how playful it's gonna be. And it was really cool with the budgeting app because, again, it wasn't on purpose, but I did notice people kept bringing this up. They were like, "Oh, I love how, like, playful this is compared to the other budgeting apps." Or like, "I love how it feels, like, way more personal." And they always s- like, I've gotten so many emails where people now, like, they end the email saying, "Oh, say hi to your dog L- say hi to Luna for me, by the way." They know that, like... It, it almost also adds this dimension where they know the person who built it, because when you also sign up for Luna, I literally throw a f- a photo of my dog in the welcome email saying, "Chris and Luna says thank you." It just adds this, like, touch of personality to the app, and it's so incredibly powerful that I was like, "I gotta do this for every single app that I do now." Um, same thing with, um, this other app which I didn't mention here that hasn't been released yet, but there's this transcription app and I made this, um, I made this character called Lily the Ghost. That one actually had a couple of, in my, in my world, like, pretty viral tweets because of the mascot. People were like, "That ghost mascot looks so cool." It actually got to the point where so many people reached out when Amazon launched their Kiro, um, code editor, like, a week ago saying, "Amazon, check your logo." I think I had 20 people reach out about that, just because they remembered that mascot. Um, and so that's, that's why I care so much about it. Another little third, uh, third thing that's kind of a benefit is when you have a bunch of placeholders or these empty states in your app, I notice that if you put the mascot there, it just adds this element of, like, playfulness and just this extra layer of, "Wow, this is a really good app," because not a lot of apps do that. They just usually have, like, a search icon or something in the app. But when you see the mascot holding the search icon, you, you remember those experiences in the app. Like, even if it, you might, you think users don't, they really do add up and it really adds that, that personality. It adds the personality. So that's why I do it. That's why I care.

    4. CJCalum Johnson58:27

       Yeah. That's so good. You know what? Can you, can you actually, um, can you put up, uh, just so, like, people can see what some of these mascots look like?

    5. CRChris Raroque58:34

       Yeah.

    6. CJCalum Johnson58:34

       'Cause I think it, it'll make it real for people.

    7. CRChris Raroque58:37

       Yeah.

    8. CJCalum Johnson58:37

       And, and just as you're doing that, um, one of the comments that we get the most on these videos, because obviously we, we've gone through, like, how to build certain things, is people are curious of, like, how to do the marketing or the branding. How do I get people to use the product or like the product? And it's like when I saw the way that you approached, approached it with mascots, I was like, it's such a unique thing, and it's simple, but it makes sense why it works. Like, it adds this, like, very personal touch to the whole experience and to the branding. And so I heard you say this in one of your videos. You go, "My typical process for coming up with the mascot is to take hand-drawn illustrations, feed it into AI, and try to come up with a bunch of illustrations until I get to something I'm really happy with." Can you kinda just, just take us behind the scenes for a second? Like, show us that process of actually, you start out just drawing something, feed that into AI, and that's how you come up with the final design. Like, like, show me that.

14. 59:48 – 1:11:24

    Live mascot build: hand-drawn → ChatGPT image → pro logo

    1. CRChris Raroque59:48

       Yeah, this is probably now the number one question that I've been getting that I have not answered in any video. So let me, let me first also show you what some of the mascots look like as well, just so you guys can see this. So for, uh, Luna the budgeting app, this is what the mascot looks like. So this is my dog, Luna. This is who the app's named after. Then we have Lily the Ghost. This is the one that everyone just really remembered, um, and they were like... I'm just so surprised how, how memorable this mascot is to people and they really loved it. And then Subscription Monster, this is the new mascot for Subscription Monster as well. So I'll take you guys through the process of what this looks like. I actually have some of the ChatGPT conversations, and ChatGPT is the way that I am creating these. It is the way that I'm doing it. But like you said, my current workflow is I take a hand-drawn illustration and then I go and I feed it into AI to make variations. I am very bad at drawing, but luckily my girlfriend's very good at drawing and so I actually pulled in a favor and asked her to go and, and, and do these for me. And so let me go show you guys what this looks like actually. Um- And so first, let me, let me show you guys what I'm talking about hand-drawn illustrations. So I will say for Luna, this one's actually-- It was so good that that one I actually didn't even use AI for. Like, the drawing she gave me, I was like, "We, we gotta-- We just gotta go use this." Like, it's, it's so good. So this is a hand-drawn illustration. However, what I did was I took this hand-drawn illustration, threw it into ChatGPT, and that is how I came up with Lily the Ghost. So you can actually see some elements of the draw- of the style, like the shadows. You can kinda see it a little bit in, uh, in the mascot. And she actually ha-has given me a lot more illustrations. So again, these are hand-drawn, but you can kinda tell, um, that they're almost a little bit of the similar style with the shadowing and stuff. But that's been really important. Let me show you exactly how I got the Subscription Monster logo, though. Um, because what I did was I took the Lily the Ghost logo, and then I tried doing some variations of that. And so the first thing I did was... Oh, I guess you can't see it here, but I literally uploaded the Lily mascot here, and I just said, "Can you create a mascot of a monster in this style?" So again, the issue is a lot of people, they kinda just try this once, and they're like, "Wow, this looks horrible. I'm kind of done." Um, what I have to do is keep tweaking it, and I keep modifying it. So, and again, it's kinda like the way, way I treat Claude Research. It's just a starting point, and I know what I don't like. So that's the important part. So I saw these, and I, I tried kind of remixing this, and I said, "Can you do another one where it's the same style, but just make it look totally different?" Like, I don't even know. And then it came up with this, where I guess it just added, like, one tooth and changed the color. And I said, "Okay, can you make the-- give it two eyes, make the horns smaller? Okay, this is looking a little too cartoonish." So I just kept remixing this, and I kind of got to the point where I was like, "Okay, this-- all of this is horrible. Like, none of this is good." So I started a new chat. Exact same l- uh, logo. So again, I just uploaded Lily. But I kind of knew, "You know what? Maybe I was too general here where I said, 'Make it a monster.' Let me be more specific." And so I tried thinking, "You know, what I'm kinda going for is this Abominable Snowman." And so I told it, "Can you make it look like the Abominable Snowman?" When this came up, I immediately was like, "This is exactly the direction that I wanna take this." Like, this is perfect. And you can tell, it really does take a lot of inspiration from Lily the Ghost, which is why I think it's important to take some sort of illustration, either an existing logo, preferably one that you got hand-drawn by an artist, which shockingly is not gonna be that expensive for a lot of people. Like, just reach out to a local artist. Go ask them... Even if you ask them, "Can you just draw this on a piece of paper or something for me?" Just feed that into ChatGPT and then have it riff off of it. That's how you get more original art styles when you do that. So, but when this came out, I was like, "This is great. Can you do a version where it's on the computer?" Really, really good. Um, I realized that, you know, it does still look a little too cartoonish for a business app, and I was kind of just trying to think, [lip smack] maybe it's the mouth. The mouth does make it seem a little too, a little, a little too happy. Um, so I took the mouth off, and I realized, "Okay, this is getting a little bit more neutral," which again, is very important for this type of product. Like a business product, it can't be too, too cartoonish or else people won't trust it. Um, and then so I... Oh, yeah, oh, I guess this is glitched. But I was really happy with this direction, like this monster with this, with this computer. So what I did was I actually fed it in another chat where I said, "Cool, we have this monster." And I, I uploaded the one, um, or I actually uploaded the, this base one to start off. And the reason I do new chats is I realize that for ChatGPT image, it does use the past images pretty heavily, so I'm constantly making new chats. And I'll, I'll show you another example of why. But in this version, I said, "Can you remove the mouth? Let's just start with this as the base." Like I, I really actually like this version. It's more neutral. "Now, can you give it glasses? Can you make them wired glasses? Okay, can you have it looking at a receipt?" 'Cause I think it's gonna be looking at receipts and trans- and, and, uh, subscriptions and all these bank statements. "Can you remove the glasses?" Right here is when I-- When I saw this, I was like, "This is the mascot." Like, I would be so excited getting this on a T-shirt or a hat or something. Like, this is just such a cool mascot. I still kept trying to, like, riff on it a little bit and see, "Can you give it magnifying glass? Can you have it go through a pile of subscriptions? Can you have it holding a calculator?" Um, you can see here it, like, added some feet. So sometimes it does add random things, which is why you start a new chat. But after a couple of variations, I was like, "You know what? I'm, I'm gonna settle on this. This is actually... This is a really good one." Um, so that's actually how we came up with this logo. And then I just picked a-- I just went on Canva, and then I got-- I just used one of their fonts and then put it next to the monster, so this is the current logo. So I might tweak this more. But this is how I come up with mascots. Um, I'll show you really quickly, um, another ver- another variant that I did because, um, people probably wanna see me do this. So I took another one of, uh, another, uh, illustration that my girlfriend hand-drew, so this is hand-drawn, and I decided, "You know what? I'm going to-- Let me try to make another version of this." And this, this one's actually interesting because I think it showcases what a lot of people get wrong too. Yeah, this is it. So I fed it in this starting image, and I said, "Instead of a dog, can you make this a cat with white, grayish brown hair?" And you can see it really did copy a lot of the style. Even got the rubber duck. It got all this stuff. And I was like, "This is a great starting point." I then asked it to make a bunch of changes, and this is gonna be my, my tip for everyone, 'cause everyone does this. They say, "Oh, I love it, but can you put her in a coffee shop, and instead of iced coffee, a mug and a laptop?" And now it looks like this. This is like the notorious ChatGPT style image that everyone keeps generating. I have found when you give it too much to do, it's just going to almost create it from scratch and barely reference the original image. So my tip is when you have a variation you really like, "I really like this cat," what you can do ... is you can actually start a new chat, and this time I- all I did was I fed it in this starting image, and I told-- I started making small modifications. And what I mean by small is, can you modify it so instead of iced coffee, there's a mug? So if you look at the original one, there's iced coffee, and this time we have now a mug of tea. Then I told it, "Instead of a laptop or instead of a keyboard, can you use a laptop instead?" Great. Can you remove the table and the duck? Great. So now we literally went from having this hand-drawn illustration to this new potential mascot for another app, because I do plan on getting a cat very soon. So but that's literally how you can take these hand-drawn illustrations and riff off of them. In another world, what I probably would have done is take this illustration, work with the illustrationist, and then just have them kind of make a bunch of different variations until I'm happy. Which I'm honestly okay doing too, but the problem is I feel kinda bad when I do that because I know that I probably only have two or three iterations with them before they'll start getting mad at me. And what I love about AI and what I love about ChatGPT is you almost have unlimited variations. But again, the really important part is I think using a hand-drawn illustration and hand-drawn because I've noticed that when you just give it random references to things, which number one can be really bad if you don't own the copyright to that or something. The other thing is it's for some reason it comes up with very non-original things. When you do something hand-drawn here, I know that there's only one variation that exists of this image because it was commissioned, it was hand-drawn. So but that's, that is the process, and that's what I've been doing. That's how I got Lily the Ghost. That's how I got-- or that's not how I got Luna, but that, that is how I'm gonna get all the mascots. And again, the beautiful part is now I can use this as empty state. So when I'm processing your transactions, imagine if you have an animation of the subscription monster going through this or if you're conducting a search ask- looking through your subscriptions, you can have this as the, as the, um, the placeholder logo, logo while that's running. So again, when you have, um, ChatGPT and you can make unlimited versions, you can almost make every single placeholder you can conceive of and put them in your app, which does really elevate it, I think.

    2. CJCalum Johnson1:09:20

       Yeah.

    3. CRChris Raroque1:09:20

       So yeah.

    4. CJCalum Johnson1:09:21

       Y-

    5. CRChris Raroque1:09:21

       That's the process.

    6. CJCalum Johnson1:09:22

       Y- you know what? It-- first of all, it's so good. Thank you for sharing that. I think the, the other piece of it, and, you know, when we first started doing, uh, these masterclasses, especially on AI, this was like a, a mindset switch for me. And so what I mean, to give context, is we did an episode a few weeks ago with, uh, Riley Brown, and one of the things he said to me before we even recorded the episode, he said, "The way that you have to think about building things with AI is fundamentally different than the way that we previously used to do things." And the thing that he said next, he goes, "It's all about speed. It's all about speed and iteration." And so what you just demonstrated with, like, the interaction that you have with ChatGPT, where essentially you give it something, you ask it to make, like, one change, you give it a single piece of feedback, and you kind of go back and forth sometimes to, like, new chats. It's like this very quick iterative process that will take you to the final thing. And for me, that was like, um... that was a light bulb moment because I think in the way that things used to be, it was always like try and get it as perfect as possible what you're building, take a bunch of time doing that, and then release. Now it's, like, all about speed. It's like give it this, give it a feedback point. It's gonna give you an output. Go again, and again, and again, and again until you get to that perfect thing. Um, I think it's so good. You know what, though? So and I, and I, and I've given this quote a few times where you said, "I've been working on this casually for three days," talking about Subscription Monster, "and we already have a brand, a working prototype, and a really rough business plan." So you and I have covered the really rough business plan. We just went over the branding piece and the mascot. I think it's time we get into the prototype. Um,

15. 1:11:24 – 1:22:00

    Working prototype: Plaid sync, 1,400 transactions, duplicate spend found

    1. CJCalum Johnson1:11:25

       yeah. Maybe you can open Claude Code and kinda just show us around and we can jam a little bit.

    2. CRChris Raroque1:11:30

       So let me show you guys what the, what the prototype looks like. And I'll be honest, I was messing with it a little bit, um, before, and I think I might've broken something, but, um, that's just also how it is. So let me show you guys though what this prototype looks like. Um, let me show my screen. So this is the actual app. Um, it is also actually hooked up to my actual, one of the bank accounts that I have for the business, and it does show the upcoming subscriptions, like what we're gonna be charged for, um, and I'm able to do all this filtering. But the really cool part is when I was going through this, I started identifying-- like, I actually did identify that I was paying for two Cursor accounts without realizing. So and, and 'cause I, I just have so many different of these accounts it's impossible to keep track of. But it actually does function. It pulls in all of the transactions. Um, I'm able to go through... There's like hundreds, like it's pulled in I think 1,400 of my transactions, so it's the last two years' worth for these specific accounts. And it is actually very-- it's like an actual useful prototype. And the big thing that it allows me to do is, I mean, it, it does answer my question, but now what I'm gonna start doing is giving it to other businesses and seeing, and I wanna see if I get that same magic moment where when I discovered, "Wait, why are we paying for this?" Like, I thought I canceled all the, all of these subscriptions. Um, I wanna see if I can get to that same point. So this is the prototype. This was really built in, I think, very casually. I got to roughly 90% of this in like three days. What I ended up also doing too- Was, um, to really prove that this is actually not, like, like there's really nothing magical here. These are all of the prompts that I use with Claude Code, and I categorize them so you can even see what kind of prompt. Like, am I dealing with an error? Am I trying to make a new, um... Am I iterating on it? Am I adding some new features, or am I doing some design work with Claude Code? There's about 107, I think 116. There's 116, I think one of these is just like a, a, a, a note for me, 116 prompts to get it to this point. It has Plaid linked in, so it's able to actually hook up to bank accounts. The syncing does work, so it does actually sync. It uses AI to actually go categorize trans- uh, transactions and everything, and being able to figure out what are the subscriptions that we have. So all of this stuff does work. Some stuff is a little broken right now because I did break it like an hour ago, but it actually does function and it does work and it's, it's, uh, 116 prompts to get here. So we can go through some of these prompts too. What we can also do, um, because I will say, like you can even see here, when I-- This was the first prompt. I said, "Can you create..." You know, "Can you, can you go do this?" And I actually pasted in, um, the features that I want because I told... I had another Claude instance where I said, "Can you just make a list of, of everything that's needed here? Like what are all the core features I need?" And that's how we started this, but there was a lot of errors. Like, like I think because I asked it a little too much. That's actually not what I advise to do. I was just-- I was kind of reckless with this one. I was moving very quickly. Um, a lot of errors, so a lot of it was like me just trying to figure it out. So we probably shouldn't go through it from scratch because it's kinda, it is kinda rough. But, um, another thing I did want to showcase here too, though, is kind of the breakdown here. Of the s- 116 prompts, um, 31, like 25% of the prompts were really just me making modifications to it. I did a lot of design stuff to get to this point, so 44 prompts were to really work on the design of this. And so it was just a lot of little things, like this card right here was very conscious. I was like really playing around with the design. Um, and then the error rate was actually very surprising. Only 14 prompts were to do with errors. A lot of them were at the beginning because, again, I don't think I did it right. And then I did ask it a lot of questions, which we can kind of get into what these questions are. Um, so but this was all done through Claude Code, and the way that it really works is, let me just pull up, let me pull up Claude Code right here. Um, so this is, um, I, I'm running it in a terminal called Warp. Um, this is like my favorite terminal but... And you can run it inside Cursor, you can run it anywhere. But the way that I do this is, let me see-

    3. CJCalum Johnson1:15:47

       When you say it's run inside Warp or run inside, like, 'cause I'm not gonna lie, this is the first time that I'm even seeing someone go through Claude Code. What does that mean-

    4. CRChris Raroque1:15:55

       Oh, okay

    5. CJCalum Johnson1:15:55

       ... that you're, you're running it inside Warp or running it inside Cursor?

    6. CRChris Raroque1:16:00

       Okay, perfect. So, so what it means is to access Claude Code, you have to use some sort of terminal, and a terminal is just a way for you to interact with the files on your computer without a folder. So typically, when you're, you know, trying to go through your files, you pull up Finder, you start moving around, and that's how you navigate. In a terminal, you can interact with the computer just by typing, just through text. All computers come with a terminal, so if you just go on your Mac, you type terminal, you got one. You have to access Claude Code this way, which is very confusing for a lot of people. So what I've done is I've opened up the terminal, and what I have to do is I have to go navigate. So let me just go back here. I opened up my terminal. You have to navigate and go into the folder of the project to be able to use this thing. And so this, the commands that I'm typing here might be confusing. All you have to do, all you have to know is CD just means change directory. That's just the equivalent of me going in Finder and saying, "Go to mascot basis folder," and just go in there. So I'm just going into the Subscription Monster folder, and so now here we are in this folder. So if I type this, you can see these are all the files that we have for Subscription Monster. It's the same thing that you would, it's the same thing that you would see here. Um, like it's literally just the same stuff you would see here. But this is the starting point. This is where you would even begin to access Claude Code. So now what I can do is I type in the Claude command in the folder, in the project that you want, and now I'm running Claude Code. And so now I have this chat window, and I can just start asking it questions. So let me, while this is running, so we can maybe have a, a discussion and you can ask any questions, let me show you just what this looks like. So again, I've already built out this project, but let's say you're in a new code base and I just want to ask it something. So we're not gonna ask it to build any new features, but maybe I ask it, "Can you explain this code base to me?" Oops, I even misspelled it. But it's really smart, so I realize you can even just, you can type in... It, it understands misspellings. You don't have to be perfect with this. What it's gonna do is it's gonna start thinking, and it's gonna start trying to analyze the code base. What I've found is that Claude Code takes a very long time to run things, like 30 minutes to up to four to five minutes every time I ask it to do something, and it's because it's really thinking through the problem, and I do think that is a big reason why it is more human-like, because unlike a lot of coding tools where you tell it to do something and it's just gonna go execute it immediately, this thing spends a lot of time reading and thinking, which I think a lot of-- If you're a good, if you're really good at what you're doing as a human, you will probably do the same thing. So that's, that's kind of, that's kind of, uh, how this works, is you just, you have this chat window, you can ask it things. And so now it just answered my question. It said, "Subscription Monster helps users track and manage recurring subscriptions. Here's the architecture. We're using Remix, we're using Clerk, we're using Convex as a database. We're using Plaid for the banking." It's able to, it's able to answer all of these different questions. Um, and then obviously I can then chat in here and ask it to go make some modifications. So let me just also show you what this modification looks like.

    7. CJCalum Johnson1:19:03

       Just one-

    8. CRChris Raroque1:19:03

       So let's say-

    9. CJCalum Johnson1:19:04

       One quick-

    10. CRChris Raroque1:19:05

        Oh yeah, sorry. Go ahead

    11. CJCalum Johnson1:19:05

        ... one, one quick thing, Chris. So the, the initial kind of code base that you were asking it to kind of break down and explain, where did that come from?

    12. CRChris Raroque1:19:14

        So this was generated using Claude Code. So what I did... And, and if you want, we can actually just go through and just make a really quick test project, if you want to see me go through it.

    13. CJCalum Johnson1:19:23

        Yeah.

    14. CRChris Raroque1:19:23

        Um-

    15. CJCalum Johnson1:19:24

        I'm just thinking the-

    16. CRChris Raroque1:19:24

        Do you wanna-

    17. CJCalum Johnson1:19:25

        ... the basics of, like, someone is literally, they're watching this video, and they're opening this for the first time, and they're like, "How do I just get set up here and get started?"

    18. CRChris Raroque1:19:39

        How about, how about, how about I just go do that? I can just go ahead and just start, um... We can just go through and make a r- we can make, like, a, a mini version of Subscription Monster from scratch. Um, do you wanna... Shall we just go ahead and do that?

    19. CJCalum Johnson1:19:50

        Yeah, we can do, like, a quick one. And, and we'll do it, like, almost the, the roughest prototype, 'cause I, I think my belief is, I just think momentum is so important. And so even people being able to just, like, get started and build, like, a rough thing, there's, like, value in that, and then you can iterate from there.

    20. CRChris Raroque1:20:08

        Okay. Then let, let's, let's go, let's go do that then. Um, so here, let me go spin this up. So, and, and h- I- ideally, I would have loved to, like, spin up Subscription Monster and done this from scratch here, but again, it takes, like, five minutes for each of these messages with Claude Code, and I did 117 of them for this, so it would take a long time, but we can do something really small here. Um, so this is just from scratch, nothing. So right now we're just in the developer folder of my computer. So this one has... Yeah, this one has all of these files. Um, so let's say I want to make a new project. I will create a new folder here in Finder, and we're just gonna call this Mini Monster in Finder. You could, you could do this, you could do this, um, uh, through Terminal too, but I'll just do it here just 'cause I think a lot of people do it that way. And then if we look here, we can see, cool, there is a folder called Mini Monster now. I'm going to go into that folder. So now we're in it. It's just a blank folder. There's nothing in this. Um, what I'm gonna do is start Claude Code inside of this folder. And so just like you saw before, now I can start asking it things about the code base, everything, but there's nothing in this. There's literally nothing. So what I'm gonna do is I'm actually gonna say, "Can you create a new app for me? Let's use..." I, I'm actually just gonna dictate this. [laughs] "React for the front end. Keep it very simple and just have it say, 'Hello, world.' You are already in the directory I want it to be built in, so don't make a new directory." So all I'm doing here is I to- I'm just telling it, "Let's just spin up a new app. Don't make it do anything. Let's just, let's just do this." This is a very common thing that I see from people, is... And again, I f- I was guilty of it. That's how I spun up Subscription

16. 1:22:00 – 1:40:41

    Claude Code 101: terminals, folders, and the “interface-first” rule

    1. CRChris Raroque1:22:00

       Monster, but that's also why when I did it, I have, like, all-- half of the errors of the project were after me, after my initial prompt. My advice is to go through Claude Code and start with a very basic project to make sure it's running and start iterating from there quickly. So I just asked it, "Just spin up a quick React app, a 'Hello, world.'" And it doesn't even have to be React. You can just tell it, "Spin up an app," and it could recommend you a programming language. But in my head, I know that I want this to be a React app, and so that's why I told it to do this. I also told it specifically, "Do it in this directory." That's just a thing that, that I'm aware of, that sometimes as I've done Claude Code projects and I-- from scratch, it'll spin up a new directory. I don't want that. Um, so but what it did was now it said, "Cool. I'm going to go ahead and spin up a simple React w- app that just says, 'Hello, world.' Can I go create... Like, here are all the things I'm about to do," and we're on this one. We're gonna create this project. Um, what it's going to do is if, is it's going to tell you every time it's gonna create a file or just make any sort of moves here. You can actually tell it, like, "Yes. Cool. This sounds good," and then you can keep reviewing that over and over again. Or if you see here, option two is you can tell it, "Yeah, do it, but d- don't ask my permission again. Just start, just keep doing this." This is such a simple thing that I feel comfortable enough telling it, "All good. Keep doing it, and you don't have to ask me anything." To be honest, and I'm, might get some heat for saying this, like, it is so good that I would feel comfortable for non-technical people to actually just auto-accept a lot of this stuff and then just put this in the code base. But the caveat is, what I think needs to happen is as you're building, before you go to production or before you ship anything, review everything at the end or have a coder review it or something. But I actually have gotten to the point where if you ask it things correctly, I do trust it. But that, that's what's going on here, is it's trying to set up the project, and all you have to know is this is just an initial starting file that just defines, cool, what technology are we using, you know, w- how are we gonna run this thing, what is it. And so I'm just gonna accept it, and now it's gonna start actually making this React app for us. And again, this could take a few seconds, but... So it, maybe we wanna, we can, we can chat while [laughs] we let this run.

    2. CJCalum Johnson1:24:17

       Yeah. So, so that's your, so that's your, kinda your starting point. Like, that's the foundation of, like... You're, you're getting into Claude Code. You wanna start building something. What you just explained is kinda like the foundation, like the first step of that.

    3. CRChris Raroque1:24:32

       Exactly. Exactly. This is the foundation. This is the foundation.

    4. CJCalum Johnson1:24:35

       Okay, and-

    5. CRChris Raroque1:24:36

       Um-

    6. CJCalum Johnson1:24:36

       Oh.

    7. CRChris Raroque1:24:36

       Oh, yeah. Sorry, go ahead.

    8. CJCalum Johnson1:24:37

       And then the, the other thing I was gonna say quickly is you, you specified that it's a React app. I guess why, why is that? What does that mean? Like I said, this is gonna so- for some people that are listening, this is gonna sound rudimentary, but I'm like, for the, for the complete beginner, why did you choose React and specify that?

    9. CRChris Raroque1:25:00

       Yeah. To be honest, um, it, it's only because I've built other React apps, but so that I know to do this. If you're a complete beginner though, what I would do is I would actually... I mean, to, to be honest, for most apps- Um, I would, I would either go ask Claude or go Google it, or go try to figure out what technology should I use. Or honestly, you could just let Claude just default to whatever it wants. Just say, "I'm gonna try to build this app." But I actually would try to pick a technology first and do a little bit of research. Um, but my default for most apps that I would recommend to people is just do a React app. If you're doing iOS, do a React Native app or do a Swift app. So there are some of these almost, like, standards that a lot of apps use, and I think most people who are non-technical can get away just using React or using these things. Um, but the reason I specifically specified here, it, it is a technical thing. It's just 'cause I know that's-- it's, it's just a really good base. It's just a really good one. And most of the vibe coded apps are using some form of React. So if you use Lovable, if you use Replit, I think if you use v0, they will be some flavor of React apps. I think for them, they specifically use something called Next.js. But, so there are these standard things that most apps do use. React is one of them. But the reason I knew is 'cause I've done it before. But if you don't know, we can actually see, um, if you go through here. Um, here, I actually... Okay, let's do this instead. Um, I'm actually going to delete what it did, and instead, what I'll say is, actually I'm gonna clear this. So let, let's just pretend. Here, let's actually just spin s- Claude Code up again. So let's say we're in this directory again. We're in the Mini Monster directory you made. I'm gonna spin up Claude. If you don't know where to start, you could actually just tell it, "Can you build me a simple 'Hello, world' app? But I don't know what technology to use. Can you make some recommendations?" You could actually just ask it this initially, and it's actually going to think through what you need, and then similar to Claude Desktop, it's gonna think through and it's gonna actually help you choose. So if you say, "Cool, I'll help you make this. Let's look at some recommendations." So actually, the first thing it recommended is React. Another option is Next.js. Another is Vue. And then if it's mobile, it's gonna be React Native. If it's desktop, it's... So you-- if you're not sure where to begin as a beginner, you could actually just ask Claude Code itself what does it recommend, and then you can say, "Great." Um, uh, you can even ask it follow-up questions like, "Why would I choose Next.js over React? What are pros and cons?" Um, and this is something I haven't actually seen a lot of people do with vibe coding tools, is just have a conversation with it. Typically, you're-- they're just using it to build. But what I found is Claude Code is so... It's pretty good and it's very smart, that I, I constantly am asking its opinion or having it double-check my work. So then it says here, "Next.js. Here's the pros. Here's some cons. Plain React simplicity," which is what we're going for here. That's actually why I physically chose it as a programmer, was we need simplicity. Like, we don't need half of the things that are in Next.js. Um, and so I can even ask it, "What would you choose? This is just a demo thing." So we'll see if it comes to the same conclusion I as a programmer came up with just from this conversation. So it says it wants to use... Okay, it wants to use something called Vite, which is a, it's a type of React, and it's, uh... Actually, I guess, I guess it is React. It is literally React, and it, it is like a very simple thing to just build very simple apps, and it's, it doesn't have all the stuff that Next.js has. So then I could just tell it, "Yes, let's do it. Can you..." Yeah, actually, let's just do it. So this is kind of the same process that we just went through where I said, "Can you spin up this React app that says 'Hello, world'?" Now instead, it's, it's just we, we kinda just went through that process as a total beginner and we told it, "Hey, I don't know what I'm doing. Can you give me the recommendations?" And now it's just gonna go through and it's gonna follow these steps. It's making, uh, it's making another folder in here, but that's okay. We'll just let it do that. And so now what I'm doing is I'm just letting it set up the project. Um, I will say that Claude Code, there is a little bit of this curve where when you're working in a terminal or it's asking you, "Hey, are you sure you should run these commands?" It's very daunting to beginners. Like, this is probably very confusing for people. But it is extremely similar to what's going on in Cursor. Honestly, I think it's-- Like, I would say that the learning curve, just because of this interface, is the, it feels like the learning curve's higher. The actual learning curve is substantially easier with Claude Code, I believe, than, than most of the other coding tools. I think this is honestly closer in terms of learning curve to Lovable than it is to Cursor. Um, and, and, and it's even closer to Re- It's, like, closer to those tools even though it looks like this. I think this thing is really scaring people, and the fact that it's like, "I'm gonna ma- I'm gonna edit this file. Are you sure you want me to do this?" It is very daunting. Again, it's one of these things where I'm just gonna accept what it's gonna tell me, and we're, we're just gonna see what happens when I accept it. But I do trust it enough where you can just keep hitting accept most of the time and it should work. I'm gonna get some heat for saying that, but I really do think it is, and that's the way that I do it. I usually just almost accept everything until I get to the end, and then I really review everything pretty thoroughly. But I just, I just trust this thing so much at this point, just because I've... I used to manually review every single thing it did, but then I noticed that I was just accepting, like, 99% of everything anyway, so I think it's okay. Okay, so now-

    10. CJCalum Johnson1:30:38

        I think to, I think to your point as well, because watching it and, like, listening to you, as a beginner, it is, like, intimidating when you see this. Like, it looks very technical. But I think what you actually just went through and explained is really valuable to the beginner, where, like, you can actually ask it questions and get in a conversation. And through that process of, like, asking like, "How would you program this?" Like which, "How would you go about doing this?" Not only is it, like, are you getting to the answer, you're also starting to learn. You're almost becoming technical. Like, you're starting to learn, uh, some of these things. And so that process as well ... is actually really important, of actually learning some of the technicalities of how to do this.

    11. CRChris Raroque1:31:30

        Yes. I, I genuinely think if you're a beginner or you're just kind of starting out programming, Claude Code, and just most of the c- coding tools, it is the greatest way to learn programming. And I, I... It really sucks, because I hear a lot of people saying, "Don't use it. You're gonna get dumb. It's not gonna be good." Yes, that's very true, but I really think that if you use it correctly, you will literally become such a good programmer. Again, if you don't rely on it too much, because you, you... It's so patient. You can just ask it, I can ask it, "I don't understand," like a million times and it's not gonna get mad at me like a senior programmer would if I asked them the same question a million times. So it is a really good tool. Um, so it's done. It, it just spun up. It says that it just spun up this app, and so it's actually telling me, um, like, "Okay, I spun it up. Go to your browser and go, go look here so we can see it." When I click it, it's not working. Like, it's not running. Um, so it looks like it's down. I kind of knew that this would happen, but I basically, if I say, "It's not running, how do I run it again?" It actually did tell me right here, "If you want to start it, run these commands." But, um, I'm just gonna ask it. Like, I'm just gonna ask it, um, like, "How can I run this?" And it says, "To run your React app, run this command." So what I'm gonna do is I'm gonna copy this command, I'm gonna open up another window, and again, we're just, we're just in the folder again. We're just in this Mini Monster folder we created, and I'm gonna go run the command it told me to run. And now I see the reason that it didn't work, and I just know this as a programmer, is because it said, "Oh, I'm running your app," but it's not, because, like, we're, we're talking to it right now. It's, it's not actually actively running it. So it was running it at one point, but then it stopped to then have the conversation with us. That's just a thing that I'm aware of. And so, but I did ask it, "Hey, I don't think it's running. What do I do again?" And it says to run it. To be honest, if I was a super beginner, you can ask it... Like, you can literally tell it like, "I'm a beginner. I don't understand. How do I run this?" Again, it's just this patient senior developer over here who will answer your questions. So you can ask it, um, and it will literally tell you, like... Or I guess now in this case, let's see. Actually, let's see what it does

    12. CJCalum Johnson1:33:41

        It's, like, coaching you through the process.

    13. CRChris Raroque1:33:42

        Yeah.

    14. CJCalum Johnson1:33:42

        So not only is it building it for you, it's also coaching you at, in the same time.

    15. CRChris Raroque1:33:47

        Exactly. Exactly. Exactly. So if I ask it here, um, well, this is a live demo, but I'll, I'll, I'll see that... Oh, yeah. So it's... Oh, I think I messed up, because I was already run... I'm, I am running it on the right, so now it's like-

    16. CJCalum Johnson1:33:59

        Oh, it's confu-

    17. CRChris Raroque1:33:59

        Let me just get this

    18. CJCalum Johnson1:33:59

        ... it's confused.

    19. CRChris Raroque1:34:01

        It's a little confused. So if I... But if I tell it, um... Here, I shouldn't have, I shouldn't have had it running on the right when I asked it, "It's not running." Um, because now it's trying to, it's trying to figure out why it's not, so it's like, it's like it just, it just killed the one on the right side to then try to, to then run itself

    20. CJCalum Johnson1:34:15

        Yeah, to figure out why it didn't work before.

    21. CRChris Raroque1:34:17

        Yeah. So let me just... Here, let me just stop it. And you can hit stop by hitting escape, so you can tell it, you can just stop it. Um, so I'll just, I'll try that again, but this time I'm not actually running it on the right. Let me help you step by step. So I see it's, it's, it's still trying to run it, so it's not telling me, like, how to run it. It's just, it's just, it itself is still trying to run it. So what I'll actually do is I'll tell it, "I'm a beginner. How do I run this? I want to do it myself in the terminal." So let's see if it gets it this time. Like, I just want, I just want some instructions. And so any... Yeah, so it says here, okay, "Open your terminal." We o- have it open. "Type this command in," which is what we did, and then, "You're gonna see something like localhost. Click that link, and now it should be working." So again, we're here. We open the terminal. We're in this folder. We hit... Oops. We hit npm run dev, which it told us to do, and just like it said, you know, you're gonna see this link, localhost, and then now when you click it... Whoops. What is that? Ooh, I think I know what's going on. I shouldn't have named it that. Oh, I know what's going on. Um, sorry. We are in-

    22. CJCalum Johnson1:35:39

        See, I like this, because I feel like people can see, and we, we spoke about this before. Like, I think-

    23. CRChris Raroque1:35:43

        Yeah

    24. CJCalum Johnson1:35:43

        ... it's good for people to see how messy this process is.

    25. CRChris Raroque1:35:45

        Yeah.

    26. CJCalum Johnson1:35:46

        Because then when you're coming across the problem, you're not like-

    27. CRChris Raroque1:35:48

        Yeah

    28. CJCalum Johnson1:35:48

        ... "Oh, great, I messed up," and give up.

    29. CRChris Raroque1:35:50

        Yeah.

    30. CJCalum Johnson1:35:50

        It's like this is actually what everyone experiences.
17. 1:40:41 – 1:46:46

    Mini app from scratch: watch the exact prompts

    1. CRChris Raroque1:40:41

       and I'll-- Now, we'll, we can do that in this example too, but let's say I see this and I'm like, "Oh, I wanna change the buttons. I wanna do this stuff," it's so much easier to tell it, "Hey, can you move this stuff around?" when there isn't a bunch of code for the back end or there isn't all this other stuff. It's just, it's just a, a prototype with hard-coded data. So much easier to prototype and work with it while it's in that form. So that's just another tip. That is the, that is the steps that I take when I'm working on features and new apps. It's interface first, then I go for back end design. Um, so let's, so it says here that it made it. So now what we do is we just have to run npm run dev. Um, I actually think it's gonna just update for us, actually. I think I don't have to rerun it. I think I can just look at the website. Um, so and then now we actually have it. So we actually have this interface, and it's just an interface. That's all it is. And what I'm gonna do is I actually do have a, a sample PDF. I'm just gonna upload the bank statement, and we can see that it identified these things. Again, it's, it's fake data. It's all hard-coded, 'cause I told it it shouldn't function, but it's kinda cool that it actually, like, whatever, I'm pretty sure whatever file I throw in here, it's gonna show all this stuff. But the, the beautiful part is that I can start modifying this. So now if I say, "You know what? I don't like the design. I don't like how it's, like, kind of centered to the left." Um, what I'm gonna do is I'll go back to Claude Code and I'll tell it some corrections. "This looks great, but can you actually make it where it's centered? I don't know why it's left-aligned. It almost looks like it was made for mobile for some reason. Can you make it look more like a website and make the design a little bit more modern?" Um, so if you notice, I also just kind of talked to it like a human and dictated. So that's another thing too, is that Claude Code is pretty smart enough to deduce this stuff. Like, it's going to really think about what you said and break it down. So unlike some other coding tools where I think you have to be very direct with it like that, you can get away with this in Claude Code. Um, and so I-- and again, we're, we're just working on the interface. There's no functionality. It's not actually processing the bank statement, but I just, like, don't like what I saw, and we can see it's already working. Like, um, so the server's already running, so we don't have to restart this. Like, it's just gonna start making modifications here. But the important part, though, is, like, I'm doing this step here before I hook up the data, because also Claude Code will have an easier time executing it if it doesn't have to think about all the other stuff. It's just thinking about the interface.

    2. CJCalum Johnson1:43:03

       Yeah.

    3. CRChris Raroque1:43:03

       Um, so let's see.

    4. CJCalum Johnson1:43:04

       Y-You know-

    5. CRChris Raroque1:43:04

       I wonder if you have any questions while this is running.

    6. CJCalum Johnson1:43:06

       Yeah. Chris, so one thing I'm thinking, just, just in the, um, [lip smacks] in the interest of time, 'cause we're, we're nearly, we're nearly up. I think-- So, so two things. So the first thing is you're actually... And, and this is even how we got inspired to create this. You're actually showing all of the steps and, like, covering all of the steps of how you're building Subscription Monster. And so for the person that's listening at home, like, they can Take this episode that o- this conversation that you and I have just had as, like, the basis, but then actually see some of the steps and also the exact prompts you used on, uh, Claude Code to build the final app. That's correct, right?

    7. CRChris Raroque1:43:50

       Yep, that's correct. Yeah. I'm gonna be documenting the entire thing. I'll be sharing all the prompts that I use, so you can see the 116 prompts to get here. Yeah, I'm gonna try to make this as transparent as possible.

    8. CJCalum Johnson1:43:59

       Cool. And, and I think the other thing that's, that's really cool that you kind of... And it's like a, a central principle, is actually you start with the interface. Like, you start with how it looks and, like, interacts with the user. Like, how it feels, where the buttons are, all of that stuff. And then you, uh, after that step and after that's perfect, then you start to figure out all the back-end stuff that makes it work, like the actual functionality.

    9. CRChris Raroque1:44:24

       Correct. Yep. Exactly.

    10. CJCalum Johnson1:44:27

        Sweet. I guess to... 'Cause I, I have two things before we, before we finish off. In that process, like even when you're speaking to beginners, or even for yourself in the beginning, what do you find is the mistake that people, especially beginners, make that, like, it kinda gets them disillusioned with the whole thing? Like, it stops them getting to that point that we're just at, where they actually have something tangible in front of them and now they can make the modifications. Like, what do you find as someone's doing this is the thing that they should be aware of, of like, "Don't do this because this is gonna kinda get you into the troubled waters"?

    11. CRChris Raroque1:45:11

        N- Number one for sure is trying to do too much in one go. It's the same thing with the logo, like when we were making the logo. You ask ChatGPT to make all these modifications. Just like a human, you give them 100 instructions, they're gonna, they're gonna fumble that. Give them one instruction, way higher chance of figuring it out. Same thing with Claude Code. When you break things down into smaller chunks, it really has a higher chance of getting it. And the second thing is to really try to ask the right questions when you're working with it, because it is... Like, a lot of people, they kind of just ask too... L- L- Literally talking to a human, when you ask someone an open-ended and a vague question, of course you're gonna get probably not the best response. Ask very specific things. So if you notice when I said, "Hey, it's not working," it didn't... It was like, "Oh yeah, it's not working. Let me try to do it," versus if I said, "Hey, it's not working. I wanna do it myself. Can you explain it?" We got two different answers. Like, it then gave me the instructions. So how, depending on how you ask the questions or your phrasings is extremely important. But really think of this as a human. Like you would pr- a, a human you've never met, a human you don't know the, the competency of, you don't know how well they can interpret things. You will probably go slower, and you will probably really break things down to make sure they understand. Treat Claude Code like that. Those are the two t- big, big tips.

    12. CJCalum Johnson1:46:25

        Yeah. One of the questions that we get the most is always like, "Well, how do you get your first users?" And interestingly, like you have the join waitlist. So it's not actually live yet, but you're already building a waitlist. Give, give me the... Why do you do things like that?

18. 1:46:46 – 1:52:31

    Step-zero marketing: waitlists, beta users, fast feedback loops

    1. CRChris Raroque1:46:46

       Yeah. Um, so I do, I do this because, number one, to gauge interest. Like, so while I'm in the process of building, I always put up a waitlist. I have a whole video on this, like why I do this. But you get, you get to gauge interest because truly I put out one tweet, I expected one or two signups, and I expected it to just be curious people who are like, "I just wanna see what's going on." There's already been 25 people, which signals to me this is kind of interesting. I've also, more importantly, got a lot of comments and people saying, "Hey, my girlfriend owns a business. Like, I think she would actually use this. Let me know when it's live." So the waitlist really signals to people. It just gives them a way to, to be able to show, "Hey, I'm interested in this," which shockingly, a l- it sounds so intuitive, but a lot of people don't do that. They just kind of build privately, and then they, they don't even have a way for people to do this. Um, that's really important. The second thing is, um, is if I have this initial set of users, it is so critical for me to pull from that pool to then beta test and to be able to ch- to, to, to check. Because if I didn't have this list of 25 people and I'm like, "Man, I really wanna test if Subscription Monster works for, you know, in another country," it's gonna be really hard, 'cause then I have to go out and go find people. These 25 people have expressed interest that they're in- that they hope- uh, hopefully they own a business. Hopefully they have something where they can actually test it. It, it's just, it's just a no-brainer to do stuff like this. But... And it's also the, the path to get the first user, because let's say I have 10 of these people help me beta test, the next 15 people who signed up, or tho- those other 15 people are pro- possibly gonna be paying user number one. So it's just like, it's just like a, like building momentum before the product's even launched. I think it's really important to do this.

    2. CJCalum Johnson1:48:28

       Yeah. It's almost like the, the step zero of, like, your marketing. Like before the thing's-

    3. CRChris Raroque1:48:33

       Exactly

    4. CJCalum Johnson1:48:33

       ... even live, you've already got momentum. Um-

    5. CRChris Raroque1:48:36

       Exactly.

    6. CJCalum Johnson1:48:37

       You know what, Chris? Here's, here's where I want it to end. I think one of the reasons, and it was even why I was so excited to get you as a guest, is because not only have you built the thing, you've also been the person that in the beginning had like... Wasn't the coder, wasn't the software engineer, had all of the imposter syndrome that comes with not having that skill set already, but still wanting to build things and make money from it. And so my question for you, almost if we had to, like, rewind in time, and the Chris that's speaking to me now had to have a conversation with someone like the Chris of the past who's watching this video and is, like, inspired to go and build something, but is facing all of that imposter syndrome, the limiting beliefs, the self-doubt. I'm just curious, with everything that you've built, and the success of it, and the thousands of users, and the revenue What would be the one thing that you would almost tell that old version of Chris that when he's having that peak moment of self-doubt, it could just get him through? I, I'm, I'm just curious. What would your message be?

    7. CRChris Raroque1:49:50

       I think, I think the message, 'cause that's really tough. I wish I could give myself three or four messages, but, um, I think I would tell myself, um, n- like, I ... I'll cheat. I'll say one long sentence. It's, like, i- it's what you, like, it, it is, it is possible, it is very possible to do all of these things that you're probably envisioning or that you don't think you can. It is, it, it is more than possible. And then the second thing, which I'll append as a sentence, is, um, don't give up because it compounds very quickly, very quickly. Um, it's one ... And now, let, let me explain. So this is not part of the sentence, but I'll explain what I mean by that, and I think my past self will understand, is, um, people don't understand that when you're building and when you're learning all of these skills, the beginning is really the hardest part. It exponentially gets easier, I think, and you exponentially pick up these skills where you get faster and faster and faster, and you start picking things up. Like, I didn't do the waitlist page until my second app because I, I learned my lesson. I didn't start monetizing two years late, but after I learned that lesson, I was like, "I'm never le- making that mistake again." Immediately the, the payment stuff is already integrated day one. Like, you pick up all these things that you don't, you know, you s- you're gonna start not even realizing you're doing these things. Like, coming up with a logo, second nature to me. Coming up with a name, second nature to me. Like, I can do this ... I did all of this stuff in a week in preparation [laughs] for the podcast, which probably would've taken most people and my past self three or four months to do. But as you do this more and you practice, it gets so much easier, and more importantly, you get better at it. So it does compound. Um, being able to acquire users, you'll do it faster. Being able to build insanely good products, you will start doing it faster. Do not give up. Keep, let, like, let that compounding happen. So it is possible and it compounds very heavily. Those are the two ... Those are the, that's the one compounded sentence I would, I would give myself.

    8. CJCalum Johnson1:51:50

       Yeah, I like that. And I, and I think it's also something I, I would've told myself as well, is also just have fun in that beginning stage. Like, it's crazy that we can even do this, and I think sometimes we put so much pressure on having this, like, perfect output right from the beginning that we're, like, trying to rush the process. We're trying to rush getting to that point, and just having fun in that, like, that messy kind of all-over-the-place beginning stage. Um, Chris, thank you so much for coming on the show, man. It's been a pleasure.

    9. CRChris Raroque1:52:21

       Yeah, thanks so much for having me, Calum. Really appreciate it.

    10. CJCalum Johnson1:52:24

        Okay, so if you enjoyed this episode, you're gonna absolutely love what we did with this one right here. Click here.

Episode duration: 1:52:32