Vibe Coding Expert: How to Build & Deploy a $1M App in the New Era of AI (Step-By-Step Guide)

1. 0:00 – 2:01

   INTRO

   1. MPMatt Palmer0:00

      Anytime you see a SaaS, anytime you see software period, there's a $100 million idea there. That's because it's so easy to create software now. To validate an idea, to build an MVP, to build a product now that maybe thousands of people use or that can make you tens of thousand dollars a month, you can just use Replit. A good example is like link in bio apps. You could build that on Replit in like five minutes. Now you can go straight from idea to application without the code.

   2. CJCalum Johnson0:24

      Okay, so today's guest is Matt Palmer. Matt is our Replit Insider, and before we dive in, let me tell you what inspired this conversation. A few weeks ago, we recorded an episode with Arlen Hamilton, where she showed us how she was using Replit to build landing pages for her companies in just a few hours. Since that moment, I started seeing more and more people online that were building profitable products in a weekend using AI tools like Replit, and so my curiosity started to grow. And I realized that there are so many people, maybe even you watching at home right now, that have ideas that you want to build but you feel excluded, blocked, because you're not a coder or software engineer. That was the position that I was in, and that's why we brought Matt on the show. He's the Vibe coding expert. He works at Replit, and in this episode, he shows us step by step all of the insider secrets to build your first successful product on Replit in just a few hours. It sounds even crazy for me to say that. But anyway, let's just get into the conversation. You are listening to "The Calum Johnson Show." In the preparation to this episode, I just started to get really excited, and there was really like two quotes or things that I heard that kind of sparked it, and so I wanted to read it back to you. The first one was actually from the founder of Replit, Amjad. He said, "This is actually one of those rare examples where building has been truly democratized.

2. 2:01 – 7:16

   The golden era for builders!

   1. CJCalum Johnson2:02

      Anyone with an idea can now create software." And so that was the first one. I, I then heard you say this in a podcast. You said, "Some of these apps I used to say, 'Oh, that'll take weeks to build.' Today, you can build an app in a morning or even an hour." And so when I heard both of those things, it slowly started to dawn on me that we're really in this like golden era for building things and for people that just have ideas and wanna build something meaningful with it. And so for that human being, they're not technical, they're not a coder, they're not a software engineer, what is going to be the value to that person of listening to the conversation you and I are about to have?

   2. MPMatt Palmer2:48

      That, that's spot on, man. And, uh, you know, I think it's one of the reasons that I get so excited about my job, helping people build with AI, helping them learn about what's possible today. Um, and so for those folks, like, that are just ambitious or really just love getting, uh, into the details of ideas or exploring what they can create, um, you don't need to know how to code anymore. And so watching this episode today, I think you're gonna be able to understand how to turn the ideas that you have in your head, all these, this ambition and these things that you come up with or that you're thinking about late at night, into something tangible, into something that you can use, that you can share with people, and that might one day be a business.

   3. CJCalum Johnson3:22

      Yeah. You know, and it, and it's cool because we were talking before this, like we're gonna get into kind of like the four-step framework for how someone can build their first product, their first app on Replit. But you just said something interesting. For that human being who's ambitious, maybe they're non-technical, but they've always had kind of ideas in their head that they want to build. The funny thing is, is actually that's your story. And so I just wanted to, to read this quote, and then we can get a bit more context on you. You said, "At one point I was punching numbers constantly into Excel sheets, and I'm like, 'A computer could do this.' Like, I should not be doing this. An app or a script should be doing this. And so I want to spend my time building and expressing my creativity. But when I brought that up about automating it to my bosses," you were talking about at the time, "that got completely shut down." You know, I kind of think it's a, it's a, it's a relatable moment for people. Can, can you just take us back there? Like, give us the context, give us the setting, uh, of that moment.

   4. MPMatt Palmer4:32

      Absolutely. Um, and, and this is super fun to talk about, but you know, like I studied physics and economics in college. I'd always been like a pretty motivated person. I never really knew what I wanted to do, and so, uh, I kind of just figured, "Hey, after school I'll go into finance. That'll, that'll be challenging," right? Like, everybody kind of told me this was like, um, something for people who, who wanted to accomplish things, right? That's kind of how I perceived it. Uh, but maybe a year or two in, I quickly found that I was doing these kind of boring, repetitive things, and that was just kind of accepted by the people around me, right? Like, I was looking at, uh, the group that I was a part of, the team that I was a part of, all really great people, but everyone was just okay with these processes, that I'd look around and think, "This-- I have better ways of spending my time," right? Like, "I am capable of more than this." Um, and I think it can be really challenging to sit with yourself and sit with that fact, right, that cognitive dissonance of, "This is the thing that I'm doing, but I shouldn't be doing it." Um, and you know, when I tried to think about, well, how do I, how do I move my life into a place where I am doing things that I'm capable, I, I am, I am pushing what's possible or getting better. That got completely shut down, right? And as a, a student that had recently transitioned into that kind of working life, so to speak, this is the first time I'd ever had a job. I was starting to wonder, "Well, is this life? Is this what life's about?" Right? Like, I wake up, maybe I have time to work out or like do something in the morning. I go to work. I work a lot, right? Like, I'm the type of person that enjoys working, but I go to this work, I don't really enjoy it, it doesn't feel like I'm pushing myself, and then I go home, right? Is that all that there is to life? And so I really had to sit with that and, and that discomfort.

   5. CJCalum Johnson6:07

      Mm-hmm. I'm curious because I think there's a lot of people that have had that moment, that realization of like, "Okay, this isn't really the thing that I want to do long term." Going from that moment to then actually making the decision to go and build something or go and learn the new skill that like takes you into the next thing is a completely different ballgame. And so I'm, I'm curious from your story, like what kind of pushed you over the edge where it was like, "Okay, I'm actually gonna take action right now. Like I can't be in this environment. I'm gonna move somewhere else, move in a different direction"?

   6. MPMatt Palmer6:43

      Yeah, for sure. And, and to set the context a little bit more, this was, this was five years ago, right? So AI didn't, didn't really exist. Um, and what that meant for-- was when I brought up like, "Hey, I wanna automate this thing." Well, I had to figure out how to like install languages on my laptop. I had to figure out how to write code. I had to figure out all these things and, and sitting there I was like, "This is gonna take months," right? This is gonna take a long time. Now, as we'll get into, there are tools that are available today. I'm gonna show you guys some of those. Um, but it just meant that it was, it was that much harder for me to finally decide I've had enough, right? And so I remember

3. 7:16 – 9:40

   The wake-up call: are you on the WRONG path?

   1. MPMatt Palmer7:16

      one day, um, we were doing, you know, like punching these numbers in and, and there's this process that I had to do that I basically was handed off to me. Every day at 4:00 PM, I'd have to go over to this computer, basically a special terminal, this was a Bloomberg terminal for those familiar, find this number, this rate for the day, um, take it back to, uh, an Excel spreadsheet, pop it in the spreadsheet, update a formula, do a bunch of manual stuff. I would mess it up sometimes. It would take 20 or 30 minutes, and every single day I knew at this time it would interrupt my flow. It would interrupt whatever I was doing. And I was just like, a hu- a human being should not be having to do this every day, and I'm the type of person that just, that drives me crazy. Um, and so I think it was at that moment kind of looking around realizing there are a bunch of processes like these. There are a bunch of people that had to do these manual things, and it took away from their ability to actually create things or make things that were unique or novel, right? Um, I said, "Okay, I'm buying a Python textbook. I'm buying a, a SQL textbook. I'm gonna sit down, I'm gonna learn these, these tools to hopefully automate, you know, some of the things that I'm doing." And I later learned that like, hey, this is what it means to work in tech. It's kind of what led me down the path that I'm in today.

   2. CJCalum Johnson8:24

      Yeah. Yeah, you know, you said something when we spoke earlier and it was like, it was such a relatable moment because I, I've even had it in my career, um, when I left my job in consulting. And you said, "At some point you wake up and you say, 'What is the outcome if I work here for the next five years?'" And then you look at the people who've been there for five years and you're like, "I don't want that." If I'm actually on a path where in five years I'm not doing what I want to do, that's the scariest thing possible to me. Can, can you, can you talk about that? That, that realization and it, it's-- I, I get what you mean because it's so vivid sometimes when you can like, you can see the person who's three levels up from you and like what their life is like.

   3. MPMatt Palmer9:10

      Yeah, for sure. And it's funny, I, I'm, I laugh because it's, I'm hearing you read these things back to me and it just kind of evokes that emotion, right? Like when I'm, when I'm thinking about these types of things. Um, and I think, you know, very early on, uh, I was talking with some friends and we were talking about like, "What's, what's the scariest thing to you? Like what are you most afraid of?" Right? This is like a very deep, visceral question. It's hard to ask, answer that honestly in front of people. And I sat and I thought about it and I think my response was, "My deepest fear is being the same as everybody

4. 9:40 – 12:26

   The fear of being average — escape NPC life

   1. MPMatt Palmer9:40

      else," right? Like I don't want to be an NPC, right? Like [chuckles] I have things about me, you have things about you. Everybody here listening has something unique about them that makes them completely different from any other human being on the planet. And I think about that in the context of what I'm doing, the way that I spend my time, showing up, doing something for 10 hours a day, right? Like that's a lot of time. That's really precious. And that compounds over a year, five years, 10 years. And if you don't, if you're not really attentive to what you're doing or where you're headed, right, you could go down the wrong path. And I'm really, I have this constant pressure that, hey, what if I just go down the wrong path or I just spend my time so focused on my work I don't think about where I'm going? And then I wake up in five years and I look back and I think, "I could be somewhere completely different. I could be accomplishing something that I didn't accomplish. I could be realizing these things about myself, um, that are unique and using those to provide value to people or like move the needle on something bigger than myself." Um, and so that's kind of where that quote came from and how I think about things. Uh, you know, is it good to be like worried all the time about this stuff? No. But I think some healthy concern for, um, who you are, what makes you different, what you're capable of, the person that you could be if you apply intentional effort every day, I think these are all really important and I think a lot of people, probably a lot of your viewers, are thinking the same thing.

   2. CJCalum Johnson11:01

      Hmm. I agree. I agree. You know what, you know, it's funny even as I, I listen to you because you talk about not, not wanting to be like an NPC and, uh, for you at that time, if we, if we went back five years ago, it's like you're punching numbers into a spreadsheet. For me five years ago, I'm, because I was a consultant, I'm like aligning columns on like slide decks and like that was my version of the, of the NPC. But you know what? You, you said something interesting. You said at that time it was like you had to go and learn Python or like a coding language in order to go and build something. Now it's different. Can you kind of talk about even your experience of like finding Replit, Vibe coding, just you even kind of learning about that whole scene?

   3. MPMatt Palmer11:49

      Absolutely. So kind of what you're saying, right, back in the day, five years ago, even a year ago, the thing about that kind of grunt work is that someone had to do it, right? Like we weren't fully there where it could be automated. Like somebody had to align the columns in the, the PowerPoint, and if it wasn't you, it was gonna be the person that was hired next after you. That's kind of one of the things that I realized is like if I'm not punching the numbers in the spreadsheet, actually the next junior person that's gets hired after me, I'm gonna tell them to do that. I don't even feel good about that because I think it's nonsense, right? But today You can do it with AI, right? Like there are tools to be able to automate these things even if you're not a developer. And so honestly,

5. 12:26 – 14:40

   Vibe-coding & the rise of Replit

   1. MPMatt Palmer12:26

      my journey to Replit started, um, about a year and a half ago. I joined just because I was, you know, in development and Replit was a tool at that time that made it really easy to, to like deploy apps or, or to build apps. You still had to know how to code, right? A lot of the tools we're gonna talk about today are brand new. Uh, Replit has only been doing what we're doing for like about a year at this point. Um, but it was just a really simple, easy place to go from kind of idea to code to application. Now, with the release of a lot of these newer AI models, um, we've dropped AI on top of the Replit workspace. I'll show you how that works today. And now you can go straight from idea to application without the code. And so it was really this, uh, sort of, um, fortuitous position that I was in working on a great platform. All of these AI models come out and now I'm like, "We need to show people how to build with AI on the platform." Uh, and you know, our leadership at Replit kind of pivoted into the AI, uh, the AI wave, and we built out a lot of really robust functionality on how to do just that.

   2. CJCalum Johnson13:21

      Yeah. You know, okay, so 'cause we're gonna, we're gonna get into it, and I, I even read the quote out, uh, from Amjad, the founder of Replit, right in the beginning of like, "Anyone with an idea can now build software," which is like a crazy thing. And so before we get into the steps in actually building, helping someone build their first products on Replit, I did some, like I was doing some research and I was even watching like podcasts before this one, and some of the stats about like Replit and the growth that you guys have experienced in the last six months is like mind-boggling. Just if I read this out. So the company-- So, so Amjad actually came up with the idea like a long time ago. The company was formally founded in 2016. At the end of 2024, the company made $10 million in ARR. And then you release Replit Agent. Since then, the business has gone from $10 million to over $100 million in five and a half months. From, from, from your vantage point, when you see something just take off in that way, right, what, what do you think is going on? Like why has Replit Agent and what you guys have built, why has it caught on with people in the, to the degree that it has?

6. 14:40 – 17:59

   An $100M company in 5.5 months

   1. CJCalum Johnson14:40

      Okay, quick pause. If you're looking to build something, this segment is for you. I remember when I started my first business. I spent more time figuring out how to start than actually doing the thing, and I was using all of these different website builders and plug-ins, payment tools. It felt like I needed five platforms just to sell one thing. Then Shopify, the sponsor of today's episode, changed the game. Shopify is a powerful set of tools that helps you start and manage your business, whether you're selling T-shirts, candles, coaching services, whatever the thing is, it gives you everything in one place. One cool feature they've recently added is Sidekick. It's your own personalized AI assistant, and you can actually ask it to help you set up your store, write product descriptions, tweak your homepage, even run marketing campaigns. And the cool thing is, is that it doesn't just follow commands. It reasons through multi-step tasks. It's literally like having an AI business partner, which is so helpful in the early stages of building something. And so if you've got an idea in your head right now and you're ready to make it real, go to shopify.com/calumjohnson to get started today. That's shopify.com/calumjohnson. Thank me later, and let's get back to the episode.

   2. MPMatt Palmer16:01

      Absolutely. And you know, I think this is gonna go back to, to largely what we discuss, um, in sort of the tutorial kind of phase of this video, which is that anybody with agency, anybody with an idea can see that idea realized in like 15 or 30 minutes. And it might not be perfect. There might be, you know, some things you have to tweak, some things you have to polish up, but you can take something that's in your head and then have a working prototype of that in like a morning. This is something that, number one, you had to teach yourself code. You had to learn how to build in that language, and then you'd have to spend like a week, two weeks, a month building out something just to even understand if your idea would work. And so that like dopamine hit of like, "Wow, I had this idea and now it's real. Now it's like on my computer screen in front of me and I can play with it," man, it's super powerful. And I know personally, um, I mean, it's changed, changed my life. It's changed the lives of many other people, whether they were able to, you know, create an app for their family, create an app for their business, all of these things. Um, and everybody sees that, and that's why we've seen the growth that we have.

   3. CJCalum Johnson16:59

      Yeah, you know, I just wanna get into it because I think a big part of the reason that I get really excited about these conversations is, to your point, the, the first kind of like barrier of entry when it came to building software was like you had to be able to know how to code or you had to have a lot of capital that you could pay someone who knew how to code. And so just that used to prevent so many people from being able to build things, even if they were creative, even if they had great ideas. But like with tools like Replit and some of the other things that we're seeing with AI, it's like anyone can do it. And so before we get into it, I'm curious just like, are there just everyday examples of like people that you've seen just building things on Replit where it's like if we even f- rewound like a few years, this person never would've been able to build this, but because of this agent, they've now been able to build something meaningful in their life? Like maybe you can just, just show us some of that.

7. 17:59 – 20:45

   Step 0: IDEA MINING (simple notebook exercise)

   1. MPMatt Palmer17:59

      Yeah, for sure. Um, so there are a ton of examples. Uh, I think Billy Howell is a great one. Um, he's an entrepreneur. He builds his ideas for clients, so he'll just go out and pitch a bunch of clients like ideas for applications, and they'll be like, "Yeah, I need that," right? Or, "That would really help our business." And then he'll go to Replit, he'll build it super quick, he'll sell it to them. He has a business basically selling these applications. So he's a great example Um, Amjad was also recently just on Joe Rogan. He discusses a lot of those examples on that podcast. Just today, uh, somebody posted a tweet. I'll actually share my screen real quick. We're gonna do, uh, a lot of this today, but, um, uh, just passed 10K MRR using only Replit Developer app. It took one month to create an MVP to sell to customers, and it's been 3.5 months since we started. Our growth tripled from last month and we are a team of two. Thank you for your service. Um, obviously the thank you is for the whole team, but, uh, you know, these are examples of someone who has an idea. They're on a small team. They don't have the development resources, they don't have the capital, like you mentioned, to get started, and they can use Replit to leverage, right, it's technology. They're using technology to leverage their ideas. And it's now much more about the ideas that you have, your ability to actualize those ideas than writing the code. And something I talk about every now and then is software development used to be really capital-intensive, right? Like, hey, this is why you need to be venture-backed. You need to raise a bunch of money and go hire a bunch of software engineers, and you have to pay them a bunch of money to then go create something. And I think at the, the end of the spectrum, that's still true, right? Like if you want like Uber, if you want an app that serves hundreds of millions of people, you need really good software engineering talent to build that. But to validate an idea, to build an MVP, to build a product now that maybe thousands of people use that can make you tens of thousand dollars a month, you can just use Replit or you can use another AI tool. And that means that the cost of investment to build a working application has fallen to almost just time and really good ideas.

   2. CJCalum Johnson19:51

      I just wanna get into it. If we were gonna start like right in the beginning, for that person that they have an idea in their mind, or maybe they don't even have an idea in their mind, let's start even earlier, and they know that they want to build something. They're like inspired by this conversation and other things that they're hearing. They wanna build something this week, this weekend. What is step one?

   3. MPMatt Palmer20:18

      For sure. So what I'll do here is we're gonna walk through a four-step process, but before we do that, I'm gonna take a step back for the person that doesn't have ideas, right? 'Cause I think if you get put on the spot, right, like if you ask me right now what my favorite movie is, I don't know. I would kind of, kind of freeze up, right? So I get it. Um, and ideas can be hard, but, you know, going back to the sco- the story I described earlier, and it seems like you're, you were in a similar position. For those people that feel like they're capable of more, that feel like they're always

8. 20:45 – 25:22

   Find your $100M idea

   1. MPMatt Palmer20:45

      thinking of new and clever ways of doing things, like you don't have to push too much. The ideas are going to come, I promise. And the most simple way that I find to do that is to just go through life and be observant. Like go to the store, you know, just go do your normal life, go to the gym. And along the way, you're gonna notice things that are inefficient or that you would probably do differently. Like maybe, you know, you're tracking your workouts in like a Google spreadsheet and you're like, "Ah, you know, I don't know. This-- I feel like I could do a better job. This doesn't like describe my workout exactly the way I want it to." Or you're writing your workout down in like a notes app and that's, that's non-optimal. Or your to-do app, it doesn't store your groceries the way you want it to. Um, or you wish you got notifications when a certain thing happened, right? And that forms the basis of product, right? This is good product thinking. It's like just observe the things around you and then think about how they could be better. And you know, I think this is old Steve Jobs quote. Uh, Steve Jobs said, you know, "Your perspective on life," I'm paraphrasing here, "it really changes once you realize that life is this thing that you can kind of like poke and something comes out the other end," right? Like life is not stagnant. You can change the things about your life that you don't like. And that starts with product and it starts with software because you don't actually have to build anything like physical with software. You have a computer, you have Replit, as we'll get through today. Um, you can just create ideas. So start thinking about life as this thing that is malleable, that you can change, and then observe all the different characteristics of your life that you would change. Because if you have an opinion, if you think something's different, should be different, I bet other people do too. And if you're gonna go out there like I did and make the claim of, "Hey, I'm different. I'm not an NPC. I'm, you know, I have all these creative ideas," well, you need to be able to back that up with actual ideas. So [chuckles] this is kind of, you know, more meta, but like as you're thinking about the things that you're doing, think about what you would do differently and then take those ideas, just write them down somewhere, and then on the weekend, you know, "Hey, I have 15 minutes, I have an hour," or after work you're excited, "I have an hour, two hours," sit down in Replit and prototype them out. That's exactly what I do. Um, so that would be my, my idea for like coming up with ideas. I'm curious though if you have any ideas for like ideas for how you come up with ideas for things you want to build.

   2. CJCalum Johnson22:48

      Yeah. Just, just to, 'cause I think that's such a good point. Just to kind of, um, build on what you were saying, 'cause I actually, I heard this from, uh, Dan Martell, and I actually spent a week doing this, which is you just get a notebook, um, and you just write down like how you spend your time. And you do this, like you have like these check-ins like once or twice an hour. Like how are you spending your time? And what you'll start to notice, and it, to your point, it's like you, we have to force ourselves to be observant because we, we go through our lives and we kind of like take things for granted. Like there's so many things that we do that we're not even fully aware of. We're just on autopilot. And so you do this process of like writing down just how you're spending your time. What very quickly starts to become apparent is actually where you're like, you're wasting time or like you had a pain point or there was friction.

   3. MPMatt Palmer23:38

      Yep.

   4. CJCalum Johnson23:38

      And so it's like in those gaps, it's like the ideas start to come, like they start to come around. And it was funny actually, we did, um, on the point of like pain points, we did an episode with, um, we did an episode with Greg Isenberg a few, a few weeks ago, and, um, he has this thing where he's like, "Anytime you're using a piece of software and you see an export button, there's like a multimillion-dollar idea that's just associated with like that export button." Um, the thing he said, he said, "I have this thesis that when you're in SaaS software and you see an export button, that every time you see it, there's a $100 million a year AI business that could replace that export button." So I guess those would be like my two. Uh, yeah.

   5. MPMatt Palmer24:32

      No, I love that. Um, and I think it's important to understand what you do and to reflect on your life because especially for people like me, probably for people like yourself, we're all busy, right? And when you're super busy, you got a lot of stuff to do, you're excited about what you're doing, you don't always take a step back and like detach and evaluate why you're doing what you're doing. And so it can be easy to kind of just get into this flow where you're not thinking about things or you're just executing. And so I actually think it's hard, but just sitting kind of with nothing or sitting with silence or just going on a walk and just not even trying to think about anything, the things that'll pop into your mind. I think it's a combination of what you're saying, which is like reflecting, working hard, being involved in the prod- problems, but then also detaching from time to time and really giving yourself the space to create. I would say even more than what Greg said, though, anytime you see a SaaS, anytime you see software period, there's a $100 million idea there.

9. 25:22 – 33:18

   The 4-Step Framework: PLAN → PROMPT → PERFECT → PUBLISH

   1. MPMatt Palmer25:22

      And I think that that's because it's so easy to create software now. You know, I think like a good example is like link in bio apps or whatever, right? That's just a website. You can build that on Replit in like five minutes. I'm gonna show you an example today of building an app on Replit, um, in a little bit more than five minutes, but it's gonna be because we're having fun, not because we can't do it in five minutes. Um, and so when you expand that and you combine it with the idea that the world is malleable, you can just do things, and there's all this existing software out there that was built by people that frankly are no better than the people watching this podcast. These are just normal people that built everything that exists around us. You can build great things too, and that's the mindset that you have to have going in. It's like, "I can do this." We'll talk a little bit today about what that means and what failure means, um, because that's important as well. Uh, but it really starts from the mindset of I can create this thing and even if I run into a little speed bump, I'm just gonna keep trying. I'm gonna perverse- persevere and it-- we're gonna get something done.

   2. CJCalum Johnson26:16

      Yeah. Okay, you know, I lo- I love it 'cause and, and here's my, here's my little prediction. So if we'd say that step zero is just becoming more aware, becoming more aware and like opening yourself up to actually having ideas through becoming more aware, my prediction of what's gonna happen is, is that if the people listening at home, if they just spent a few days doing that or maybe even a week, they're gonna have like a list of potential ideas. How do you-- and I'm curious for you personally, when you have that list of ideas, what is almost the way that you're thinking about it of like, "Okay, there's 20 ideas here. This is the one that I'm gonna action on and actually build." Like, what, what is that transition for you?

   3. MPMatt Palmer27:01

      Yeah. This is a great question and I would say, right, I think it's... I use, I'm a, I'm a perfectionist. We'll take a step back. I, I want to do everything. I want to do it perfectly and it has to be excellent. That is the type of person that I am. And so to tell me, right, "Hey, step one is actually just becoming more aware," I would say, "That's ridiculous. I need to do something." But [chuckles] right? Like, but I think the counter to that is consistency and realism, right? So if you're working a full-time job, if you have a family, if you have all these things in your life, that m- step one might be perfect. And over everything that we discuss, consistency trumps everything. So if you work an hour a day for 10 years, that's a lot, gonna be a lot more impactful in compounding than working 10 hours one day and then flaming out 'cause it was just too much and like not getting anything done, falling off, and then you just drop it, right? So everything that I say is gonna kind of be couched under that kind of consistency, uh, consistency mindset, um, and deliberate kind of practice mindset. Um, so say I have a ton of ideas. I always, I have a ton of ideas all the time. I have a to-do list with like 30 things on there. I'm not gonna get everything done today, but that, I'm gonna recognize that. I'm gonna say, "I'm gonna work really hard today and I'm gonna pick up on it tomorrow." That's the first thing. And then second, I think we just bias action over everything. So it doesn't really even matter which idea. Say I wanna learn how to use Replit. Well, which one of the ideas makes me the most excited? I'm gonna pick that one and I'm gonna sit down on a Saturday and I'm gonna see if I can build it. And then if I can build it, I'm gonna share it with people. And if I can share it with people, I'm gonna see if I can turn it into a business, right? I would say start over everything and it's really more about consistency than it is picking the perfect idea.

   4. CJCalum Johnson28:37

      Yeah. You, you know what's funny? In a way, like that is the most important thing. Like if you were just to take one thing from this conversation, I, I hope you take a lot more. But if you were just to take one thing from this conversation, that bias to action. And it, and it's funny actually because I was having, um, I was having dinner with some friends over the weekend and, and one of my friends, he's been, uh, he has this craft. He's been working on it, like music. He's been working on it for years, uh, like producing. And one of my other friends was like, "Have you released any of the like songs or anything that you've made?" And he's like, "No, no, no. I'm just, I'm just waiting just to get a little bit better. Like I'm, I'm working on this, this style right now." And it's like it's been years and it, it's one of those things I even had to reflect on it on my journey, which is like a lot of people's dreams, it dies in like this perfection, perfectionism, like analysis. Like just getting a little bit better, but like in complete private. Like no one's heard it, it's not been shared, it's not been deployed. Um, you know what? Let's get, um... I wanna, I wanna get into it. I know that you-- Let's, let's start whiteboarding. Like let's start, let's start building some stuff.

   5. MPMatt Palmer29:53

      Let's do it, man

   6. CJCalum Johnson29:54

      Okay, quick break. So it's been four months since I started using the Eight Sleep Pod, and I'm not gonna lie to you guys, I love this thing. We're in the thick of a humid New York summer, and usually that means I'm tossing, turning, sweating through the night. But with Eight Sleep, the pod actively cools my bed down during the night so I get better, calmer sleep. And they've just dropped their latest model, the Pod 5. It's a smart mattress cover, and it uses precision temperature regulation to control each side of the bed separately. It can be cold for me and warm for someone else. But it's way more than just a temperature pad. It tracks your heart rate, breathing, and sleep stages without wearables. It will detect snoring and even subtly elevate the bed to stop it. It even has built-in speakers for white noise and guided meditations. Every morning, I wake up to a personalized report on my sleep. It's literally like having a personalized sleep coach, a thermostat, and a health monitor all in one. And their new autopilot health check will even flag trends in your heart rate variability and breathing disruptions so that you can spot issues early. The Pod is actually clinically proven to increase sleep quality by up to an hour each night. And so if you're ready to upgrade your sleep, go to the link in the description to get $350 off your very own Pod 5. You get 30 days to try it at home. For any reason, if you're not happy or if you wanna send it back, you can. Okay, quick pause. I wanna take you behind the scenes for a second. Twenty-four months ago, my workflow was in chaos. Docs are in Google Drive, notes in my phone, random PDFs just buried somewhere on my laptop. I'd waste so much time trying to find stuff, and it was killing my focus. Then another entrepreneur that I knew told me how he ran his business on Notion, the sponsor of today's episode, and honestly, it's changed things. Everything I do, from planning this show, to managing my team, to building projects, it runs through Notion, and even that's been taken to a whole 'nother level with Notion AI. And that's their all-in-one AI tool that's powered by your work and puts everything in one place. It automatically captures meeting notes, instantly finds the exact content you need, drafts documents for you. And it's so convenient because it has all of the best AI models like ChatGPT or Claude built right into it. Lately, I've started using their AI meeting notes, and so whether it's a team call, a guest interview, or even just a solo brainstorm session, it grabs every idea, every decision, every next step. It's not a bot. It's no hassle, just clean, organized notes that I can take action on immediately. And so if you wanna stop wasting time and actually get ahead at work, then head to notion.com/calumjohnson. That's all lowercase. So again, that's notion.com/calumjohnson, and explore the most powerful AI tool for your work. Thank me later, and let's get back to the show.

   7. MPMatt Palmer33:06

      The way I like to do this if, you know, this, this tutorial assumes, right, the folks on this call are a little new to building with AI. Maybe you haven't built anything before. That's totally fine. I think about it in these four steps. First,

10. 33:18 – 38:57

    LIVE BUILD starts: planning the Zen-Timer app

    1. MPMatt Palmer33:18

       we're gonna plan out what we want to build. This is really important, right? 'Cause maybe we don't have that idea. We need to figure out what we want to build, and I can guarantee you, when you sit down and you write out what you want to build, it will be more complicated than you think. And a lot of people fail at building with AI because they're like: Oh, well, I just want, you know, uh, an app that, you know, tracks my workouts. And then they type in the prompt, "Build me an app that tracks my workouts." AI is like, you know, it's like you're explaining instruct- instructions to someone that knows nothing about what you want. You're not gonna get what you want if you just type that in, and it's quite likely that you'll get something that breaks. So the very first thing I like to do, especially for new folks, is plan. We're gonna do that together. Then I like to prompt, right? So we're gonna think about our plan and translate that into English. We don't have to get into how AI models work. We can talk about that a little bit, but the important thing is that you take your ideas and you're incredibly specific about what you want and how you want to achieve it. From there, right, we're gonna perfect or polish what we prompted. This is also important. We'll talk a little bit about this, um, later during our tutorial, but AI is not perfect. We have a ways to go before it is perfect. Despite what, you know, the internet says or, you know, people on the internet, there are errors. You will fail. You will run into issues. The funny thing is that that's exactly what engineering is. Maybe engineers don't tell you this stuff, um, but most of doing hard things is just dealing with problems and solving problems. And I bet everyone here listening, if you have a job, if you have a family, something in your life is probably imperfect, and at some point, you have to do hard things to get around that thing. That-- Engineering is no different from that. Building is no different from that. Building with AI is no different from that. And what that means is that you have to have a little bit of resolve. You can't just give up as soon as a hard thing happened. I bet everybody watching this show has overcome hard things at some point in their life. Maybe they sit around and think, "How can I do more hard things to, to practice that? How can I get better? How can I be more resilient or have more resolve?" Building software, creating applications is a great way to practice that because there are problems, and you have to sit with the problems and think, "Okay, I need to overcome this thing." Then finally, we're gonna publish our app. So Replit makes this pretty easy, pretty straightforward. This is actually the simplest part, taking what we built and putting it on the internet so that other people can access it. Now, there are some follow-up steps to that, making sure things are secure, et cetera, um, that we'll discuss, uh, but that are a little bit more time-consuming, so that's important as well. But we're gonna follow this st- this, these four steps, and if you're someone thinking, "Oh man, I don't even know where to get started," and you wanna do this on your own after the video, follow these steps. Follow along with what we're doing. I think you'll kind of pick up ... what we're putting down. So I'm gonna share an idea. Um, I was brainstorming this one, uh, earlier. It's actually a thing about meditation, and I bet a lot of your viewers are interested in maybe meditation or meditation tracking. I was thinking about, like, an app that has maybe three pages, so I drew this out. So we have, like, our Zen meditation app. Maybe there's some guided exercises, maybe box breathing, which is just a meditation practice, four-seven-eight, just a timed session, maybe some other things that I don't even know about. That might flow into a timer where we can add maybe some, uh, background noise or interval bells, um, and then maybe we get some sort of usage statistics or tracking out of that, right? Um, and I bet a lot of you have used a similar app before. Maybe you've used an interval timer, and you probably looked at it and you thought, "I could make this better. I could do something that would make this closer to what I want, so what I want to build." Now, this is what I was talking about before, right, in planning. What, what do we want? The essence of, of all of the things we're talking about, AI, it's making it easy to express your ideas. But if you don't understand what your ideas are in the first place, how are you going to translate that to AI? Um, and so what are we building? Well, we're building a meditation web app with customizable sounds, interval bells, and session tracking. We want to be able to save sessions, timers, and goals. Now, I'm gonna talk through what this means really quick here, and then I'm gonna pause just to make sure, uh, we're all aligned. But you might look at that and you might say, "Great, I'm gonna type that in the prompt box." I think this is a really important step in planning. We have to break down what this means, right? Like, think about it like, you know, like LEGOs, right? We're assembling all the pieces. We need to figure out what the pieces are in order to build this thing. And, and I talk about, uh, the difficulty in building with AI a lot. It's actually like it's more than just being a developer, 'cause you're planning out the entire product. You're being a director. You're like the CEO and you're the developer and you're the marketer and you're all of these things. You're running your own little team here. Well, for our app, if we want to save sessions and access them later, right, like our meditation sessions, we need somewhere to put that data, which means you kind of, something called a database. You don't really have to worry about that. We'll talk about it later. Um, but you need somewhere to store the data, and a lot of tools, they don't, don't actually have a way to do that. You need a way for people to save those sessions in the first place, so like a f- a front-end kind of, um, interaction, so like a web page, a way for users to log in. That's actually a complicated problem in and of itself. You need sounds, right? Like, we need our little meditation bell or our background music. Where are we gonna get that? Um, and then maybe we want it to be mobile-friendly, and there, there's something called a, a progressive web app. You don't really have to know about that either, but that's what allows sort of some really good mobile friendliness on the iPhone. Um, other things, you might want tools to make sure your app's secure or a way to sell your app to other people, especially if you want to make money. These are all things that set Replit apart from some of the other tools out there. Now, why do I bring this up? I bring it up because people hear terms like database. They hear terms like maybe pr- progressive web app. They hear

11. 38:57 – 45:03

    PROMPT MAGIC — AI Agent writes the entire code

    1. MPMatt Palmer38:57

       terms like user authentication, session management, et cetera, and they get overwhelmed. They say, "Hey, I'm not a software developer." This is very common. However, those things don't matter. It doesn't matter because the whole essence of building with AI is just un- understanding the puzzle pieces that you have to fit together. It's not really about, like LEGOs, right? It's not about understanding the nuances and intricacies of how these things, uh, actually would be implemented. It's understanding maybe just the concept or the idea of, like, what it is, right? Um, if you notice something's wrong with your car and you just know enough about cars to tell, "Oh, my battery's probably dead," it'll save you a lot of money just to go buy a battery and switch it out, which is, like, maybe not the hardest thing to do, versus going to a dealership and having them analyze your whole car and, like, all of these other issues, right? So it's about understanding concepts and things at a high enough level that then you can debug that on your own. Does that make sense?

    2. CJCalum Johnson39:51

       Yeah. And, and you, you know what? I'm gonna, um, just, uh, context for you, Matt. I'm gonna jump in even as you're kind of, like, going through this stuff, and I'm gonna be kind of like that complete beginner that's, like, asking the question. And so one of the things actually as you were going through it, you, you used a few terms. You said, um, like being the director almost of building your product and understanding all of these different components and how they fit together, and you even kind of outlined some of them. And it's interesting because I have no, uh, background as, like, a product manager or even, like, what it takes to, to build something, and a lot of the things that you mentioned I would've taken completely for granted. Like, I wouldn't have even thought about it. And so I'm curious, even as you're like-- and I'm thinking about that person who is the, you know, up until this point in their life, they've been the consumer of apps. They haven't been the person actually behind the scenes building them. How are you actually figuring out, "Okay, this is exactly what I have to plan before I build," versus, you know, "I can just kind of, you know, deploy the thing first and then iterate later"? Like, what is almost, like, the minimum that someone needs to have in place before they can start building?

    3. MPMatt Palmer41:16

       Yeah. So I mean, that's great, right? That's a great question because-- and also I appreciate, please do interrupt because otherwise I'll just be talking and that's boring, uh, and, and we need some live commentary. So, you know, what I'm, the exercise I'm going through is j- through is just to conceptualize my idea, and I, I wouldn't get too hung up on the specific technologies because Replit does handle a lot of that for you. But really the exercise is just to understand if this is possible, right? So for example, here's an example, uh, that I get a lot. Somebody's like, "Hey, how do I build an iOS app on Replit?" That's actually possible, and I have tutorials on how to build a prototype for an iOS app, but it's hard. Why is it hard? It's really hard because to get an app on the App Store, you actually have to pay Apple $100, and you have to go through all these hoops, and then your app has to get reviewed, and there's, like, a lot of back and forth. And so, right, if you understand at the get-go, okay, well, like, that's actually really hard You know? Like maybe you can refine that down into an MVP or something simpler, where you just try and build out, again, the, the MVP, you said the minimum lovable product, I would call it a minimum viable product, um, so that, right, you have 30 projects. You don't pick one and then you get 5% of the way there, but it's such a gigantic project that you kind of fall off. So when I think about this planning stage, what I'm doing here is I'm thinking: how do I think, how do I express the simplest possible version of this app? And then how do I kind of like break it down to understand if this is worth, you know, like how long this is gonna take me, right? So like what we're talking about here, when I'm saying MVP, maybe I'm like, "Hey, you know what? Actually, you know, this is actually a little too hard right now. Let's-- We, we only have 30 minutes. Let's focus on just a way for users to log in and some sounds and mobile friendliness," right? So when I'm thinking about this, I'm really just thinking what is that quick win that I can get that's gonna get me excited so that I can really nerd out. Because then, then when I really nerd out, I'm gonna want to go the, the, the distance or I'm going to want to sink more hours into this rather than saying, "Hey, actually, day one, you know what? Actually, we have to build out the entire product." It goes back to the consistency piece.

    4. CJCalum Johnson43:20

       Yeah. You know, you know what actually as I was just thinking, and it-- and I'm interested to hear what you think about this, Matt, is... 'Cause I, I think it kind of, um, one of the things that I like doing is almost getting like a frame of reference, like having like a, another similar product that you can then kind of focus on and pick apart. And so almost I kind of like that idea of like stripping a lot of the features back and like focusing on one thing. And because there's a ton of meditation apps out there, right? And so you could focus on like one key feature or one key improvement that you're like, "Okay, I just want that to be present in my app." And to your point, that's like your first win, like figuring out how to express that to the Replit agent so that that can be built. What, what, what do you think about that? Am I com- completely missing the mark?

    5. MPMatt Palmer44:09

       E- exactly. Exactly. That's exactly what I was trying to get at, and I think that that, that was going to be the next step that we're gonna go into, actually, is like, okay, well, like we have this, right? Well, how do I just turn that into a little prompt? And so this is like prompt building 101. So we're, we're planning, and now we're going into prompting, right? And so to kind of tie it all together, maybe to your point, you know what a meditation app, agent probably knows what a meditation app is. How do we make this different? How do we set it apart, right? And like maybe, hey, there wasn't an app with box breathing. I like to work out. Maybe it's an interval timer. I've had a really hard time of finding interval timers that have customizable intervals that match exactly what I like to do. So maybe you have a special meditation practice that you're like, "Hey, other people need to know about this thing. I'm gonna build an app that's like every other meditation app, and then I'm gonna remix that. I'm gonna riff on it." That's actually the powerful-- one of the most powerful things for coming up with an idea is that we didn't talk about that, is like pick something,

12. 45:03 – 53:50

    Debugging 101 (You will Believe you can build ANYTHING!)

    1. MPMatt Palmer45:03

       copy it, and then do it better. [chuckles] Like, I think now is a really great time to be able to do things like that. And if, if you're really like, "Man, I'm stumped," I'll, I will say you can just use other AI tools. So like go to Claude, say, "Hey, help me think through how I'd build a meditation app." Claude's gonna write something down for you. We don't wanna just copy and paste this verbatim. I think this is important with AI. We'll get into this a little bit more. You shouldn't just be mindlessly dumping stuff into AI and then copying and pasting it. You should be really thinking about what you want. But like, I'd approach building this app by defining the core user experience. What did we just do? We defined the core user experience. Then building out the technical foundation and content strategy. And so like this is how you can start to learn how to think th- from first principles, right? Describe from first principles, principles what I'd have to do. There's no excuse anymore because, like this is the free plan of Claude. [chuckles] You can go to Claude and have it walk you through how to think about all of these things. And so, you know, I'm sure listeners of your podcast are getting a ton of information already just by, by, uh, viewing. But like any problem that you have in your life, you can go and say, "Help, help me approach this thing logically. Help me break this down," 'cause I'm, you know, it's a really big goal. Like, I wanna build an app that I can quit my job and like focus on building this app first, first full time. That's a big goal. And so, you know, we see a lot of people successful with Replit, but I don't want people to think it happens overnight. It doesn't happen overnight. It happens from doing the work and learning how to do these things, and it can happen a lot faster thanks to AI today, but it's still about learning the skills. However, you can use AI to learn skills.

    2. CJCalum Johnson46:45

       Yeah. You know what? I, I actually love the point of going to other AI tools and almost like asking, whether it's ChatGPT or Claude or Gemini or like whatever the thing is, and asking it questions and actually helping it, like letting it help you guide you through the process. Because I think a lot of the times, and I've felt this, when you're building something, especially when it's for the first time, you, you, you know, obviously come across a point where you're like, "I don't really know what to do next." And it's not really that you don't know what to do next. It's like there's some questions that if it was asked to you, you probably have an answer, and then you just pursue that answer. And so it's actually this very helpful process of like just have the AI ask you questions or like guide you through the process. That's gonna clarify your thinking, and then you can take that clear thinking to Replit.

    3. MPMatt Palmer47:41

       Exactly. And, and I wanna encourage that. There are so many tools out there. There's so many things. Use all those tools to build up a plan, and then we're gonna walk through how to build something in Replit. Um, but people get stuck and they're like, "Matt, you know, like I don't even know where to start." And I'm like, "Just go search Google. Ask ChatGPT, ask Gemini, ask Claude. Do whatever you have to do to get the information." These are things like typically you used to have to read a book to learn or you have to go to a library or do a Google search. You can do them in 15 seconds today. We're gonna talk more about that in the context of building this app But this is like the prompt that I sat down from our plan and I said, came up with. I was like, "Well, how do I instruct AI?" We'll break this down. "Help me create a meditation app. It should feature different meditation presets like box breathing or timed practices." I'm giving an example. Hey, I want different presets. What does a preset look like? Well, I know box breathing. Um, and then the next sentence: "Do research on popular meditation techniques to incorporate them into our app." Now, this is a, a feature unique to Replit. It can do web search, so it's actually gonna go out and do the research for us. So now we're in a point where, right, you can actually start to rely on AI for some of these concepts, right? I, I talked about breaking down our ideas, but you can also ask Replit, "Go find those ideas for me." Um, then the next part of this prompt: "The timer should be the core piece of functionality, so do research on techniques for adding a timer to a browser," right? Like, I'm Matt, I don't know how this stuff works. I just want you to figure out the best way to do it, um, and I want you to do it in a way such that the user can engage with it over longer periods of time, right? 'Cause if we have a timer running in our browser for 50 minutes, I don't want the timer to, like, stop while I'm meditating. N- you know, I'm just too zen. I'll meditate for two hours and be like, "What happened?" You know? I need it, [chuckles] I need it to continue working and to pull me back in. And I'm saying: Hey, make the app hyper-minimalist to be distraction-free and zen. So we're being very descriptive about what we want to build here. Um, and from that description, now we're gonna get into Replit, right? So now I'm sharing Replit.

    4. CJCalum Johnson49:38

       Just super quick. N- no, 'cause I, I do wanna go into, uh, Replit and actually, like, making the magic happen. Um, but first I had to ask this, and I had to ask this because of, like, I was that overthinker and that person that would get, like, stuck in the planning phase of a business, and so I just want to address it head-on. Um, for that human being, is there, like, a point in the planning process where you know that you're actually ready to take the idea to Replit? Because h- here's my concern, is that people will just keep planning and keep talking with Claude or ChatGPT, and it's like, or doing research or, like, whatever the procrastination is-

    5. MPMatt Palmer50:23

       Yeah

    6. CJCalum Johnson50:23

       ... versus just building. And so, like, what to you is the sign of like, no, y- you're ready to actually just go and build the thing?

    7. MPMatt Palmer50:30

       Yeah. No, I mean, that's a great question, right? Um, I have a concrete idea. I, I know something that I want, and I know the pieces of that idea, and I can write them down in, like, what is this, like, 10 sentences? One, two, three, four, five sentences. You can, if you can write it down in five sentences, I just say send it. And you know what? The cost of compute is so cheap. The cost of AI today, you know, you have to pay for this stuff. It's not free, unfortunately. But it's so little. Like, to validate this idea, it probably would've took me, like, a, probably a morning, at least four hours or something, to spin up an app that might do this. Today it can take 15 minutes. So if you think an idea is even remotely ready, just go to Replit and click start, and just, like, see what happens, see what comes out of it. I do that all the time. Sometimes I'll spin up a couple tabs and fire off a couple different prompts just to see what I get. And because AI's doing the work and, like, you know, we have Replit on the desktop. We have it on your phone as well. You can do everything that we'll do today in the browser on your phone. Like, don't worry about it. I definitely bias action. I'm doing the planning so you can understand, like, how to basically get a good output most of the time. I can break that down more conceptually. I have some, uh, a little slide prepared. Um, but don't sit around and plan too much, right? If you're watching this video, first night you do this, you can follow along with me. By the end of the video, you'll have an app. If you're thinking, like, "I have an hour," you should have an MVP. Maybe it works 10% of the time. Maybe you learn what doesn't work.

    8. CJCalum Johnson52:00

       Mm.

    9. MPMatt Palmer52:00

       And the way I learn is by doing this stuff, and figuring out what doesn't work sometimes is a lot harder than figuring out what does work, and the only way you learn that is by trying.

    10. CJCalum Johnson52:08

        Yeah. Y- you know, you, you said it so good, and it, and it reminded me actually of something, uh, Riley said when we did the episode with him, Riley Brown. And, um, when he said it like this, it actually, it kind of like clicked for me. He was like, "The process I'm gonna show you, it's not like, um, it's steps, but it's not like a linear step-by-step process where you do step one, step two, step three, step four. It's like you might get to step three and then all of a sudden go back to step one." Like, you might go and deploy the thing and be like, "Actually, no," and then go all the way back to the planning phase, and then go back to dep- like, back to deployment. And it's kind of this, it's this mindset switch with actually using AI, where it's like moving between these different steps is actually much more fluid. And so we've done step one, which is plan. Step two, prompt. And then step three is perfect it. I guess let, let's, maybe we can, we can hit fire on this and see, see what we get.

    11. MPMatt Palmer53:13

        Amazing. Um, so I have some ideas. We're gonna, we're gonna fire this off. I don't want people to wait too much longer, and then we're gonna jump back into some conceptual things that tie into what you just said, 'cause it's really important. It is not linear. You're, you fail. You, maybe the best course of action is starting over. Um, and we'll talk about that as little, as well. So I had this prompt. I kind of crafted it. We didn't spend too much time waiting around. The first thing Replit Agent's gonna do is going to make a plan. So we send it. This is the first screen that you'll see, and it's gonna, agent's saying, "Hey, I'm gonna help you create a plan for this app." We planned. [chuckles] Now we're giving a plan to agent. This is the implementation details of our application. And now what you'll see down here

13. 53:50 – 1:02:45

    Perfecting: log-in & data in minutes

    1. MPMatt Palmer53:50

       is it's searching the web. So agent's gonna search the web. It's gonna come up with a plan for us, and then it's gonna show us that plan. That typically takes one or two minutes. So while this is happening, we're gonna come back and we'll see everything that's happening. I'm not gonna skip any steps. I wanna talk about what you just said. Everything, most of the things you've done in your life, right? Maybe you go to school, you go to a job, and there are steps that you do in order. And if you do those steps, you get an outcome. So, like, maybe you press buttons and then something happens. And you kind of learn, "Hey, if I do this thing, if I do X, then I get Y." AI is not like that, right? So if we have pe- people here that took statistics or know what, like, a distribution is or what it means for things to be random, AI's like a little random. That's actually kind of crazy, right? And so what I'm explaining here is like, hey, think about the outputs quality. There's some likelihood that you get an app of a given quality, and so we can mark this up a little bit, right? Maybe most of the apps that you get are here, right? Like the, there's a high likelihood that the quality is like average-

    2. CJCalum Johnson54:51

       Mm

    3. MPMatt Palmer54:52

       ... from a given prompt. But then there's some likelihood because AI is like a little random. Obviously at Replit, we're trying to make these results as good as possible. Some of your apps might be really great, some of your apps might not be really great, and that just might be because AI makes errors sometimes. And I've done this live on podcasts before. I try to build something and it breaks, and it's really embarrassing. So what I'm trying to, to emphasize is the same prompt is gonna get you different results. Some of them might not work, some of them might work really well, and there are things we can do to increase quality and likelihood of success, and that's what I'm explaining to you in this video. But that doesn't mean you won't fail sometimes, right? This is like sports. If you ever played a sport, if you played baseball. I played baseball. You bat 300. That just means you literally get out 70% of the time and you're a great baseball player. Think about it like that, right? Like not everything is going to work. It's not gonna work all the time. I'm not saying you're gonna fail more than you succeed, but what I am saying is that you might fail. And the implication of that is that we can't be afraid to fail, right? We can't be afraid to start over. And the secret of being an engineer is actually failing a lot. And so, like, there's this concept of variance, which just means like there's a likelihood associated with these probabilities. Um, and that's the mindset I want to bring into building with Replit, is that we're gonna fail and things might, might look different sometimes because AI is just random, right? Like the prompt I gave to Claude, um, might be different different times I send it.

    4. CJCalum Johnson56:12

       I, I was just gonna quickly say, it, it's actually funny that you mention that. Um, I was using ChatGPT this morning and I was having that exact thought of like, sometimes I, I'll wake up and I'll use, uh, ChatGPT or other AI tools and it's like the answers it's giving me are so perfect and so great. And then other times I'll wake up and I'll be using it and it's like completely off the mark. And to your point, it's like we have to kind of get our minds around, and it was even an adjustment for me. These, there's not like a perfect way to use these tools. There's not like a perfect prompt, and they're unpredictable. So sometimes it's like it's firing, sometimes it's not. And so the actually, the important thing is that can you correct it? Like, can you almost like correct, uh, the ship-

    5. MPMatt Palmer57:04

       Mm-hmm

    6. CJCalum Johnson57:04

       ... uh, to your point? And, and obviously we're gonna get into to how to, people can do that.

    7. MPMatt Palmer57:09

       Yep. And, uh, that's perfect. So if there are people here who use ChatGPT, it's all about context. When I say context, I mean the information you give to that. In our case, it was a prompt, right? Like this is the context that the AI has. Sometimes it could be a link, as we'll see later on in this episode. Sometimes it could be an image. It's about the context that we give to AI. So I want to break down exactly what's happening here, and then we're gonna jump into actually, like we're gonna jump into the building. So Agent said, "I'm gonna help, help you create this thing," and then it searched the web for meditation techniques, right? So now it's like learning about meditation and it's gonna integrate those into our app. And based on the research, it gathered information about the app and then it created a visual preview of what our app might look like. So this isn't actually a working, uh, working application, but it is a preview and it tells us exactly what it's gonna include in the application: a robust timer system, meditation presets, um, customizable timer durations, visual breathing guides, pause/resume, session completion, progress tracking. That's perfect. That's a, a lot of what I wanted, and this looks good. So when I click Approve Plan, we're gonna get dropped in the workspace and Agent's gonna start working. See this little message down here. And what it's gonna do basically is build out this visual preview. Because this is just a preview, it's not gonna look exactly like this, but Agent's using this basically. And when I say Agent, that's just our automated developer. You can think about it like a junior software developer or something that can build apps. It's gonna use this preview for what it's building. And so conceptually, I want to talk about what's going on because this can be confusing to a lot of people. I notice that a lot of people kind of don't understand just how cool this is or why it might be different. I'm on replit.com here. I didn't install anything on my computer. I didn't have to install Python. I didn't have to, you know, learn what Python is. I didn't have to install JavaScript or maybe some of these other things you hear about. And I just created a workspace and now I have this autonomous developer that is writing files to build a thing that looks like what's on my right. And so this is a pretty user-friendly display. I would like to think it's pretty user-friendly. You know, it can always be better. But under the hood, if I click open files, there is code being written here.

    8. CJCalum Johnson59:18

       Mm.

    9. MPMatt Palmer59:18

       And that's important for a few reasons because at some point, say you have a wildly successful app, maybe you hire a team of developers. Well, you need to give them the code. Or maybe you want to like learn more about how to write the code. You can ask Agent or Assistant, another tool in the workspace, "Explain to me what this is," right? Um, "Explain to me how this app works. Explain to me what you're building." And it can break it down. That's actually, I talked about being a software developer. I learned Python back-end engineering. It's very different from front-end engineering, building these apps. But I learned that through AI. Through like, okay, I'm gonna build this thing, and then I'm gonna open this up and I'm gonna go in and I'm just gonna look at files. [chuckles] And like after the course of months and years, you're gonna understand what they're doing and you're gonna write that code on your own. So if you want to learn, if you want to go down that rabbit hole, this is actually really important for learning. Um, if you don't want to learn and you just want to build cool stuff, that's the whole point of this episode is that you can just build cool stuff. But just like I talked about learning how a car works, starting to understand how things are organized can be really helpful, and that's gonna be a major theme of this video as well.

    10. CJCalum Johnson1:00:20

        Yeah. Y- you know what? And, and this is why I was actually really excited to bring you on is because we've actually, we've had, uh, guests come on the show and they take us through Replit and they show certain things. Um, the cool thing with you is like You actually work at Replit and are like part of building this stuff, and so just your knowledge of the agent and of the platform is like on another level. And so I have this thing, I think what kind of separates like a beginner from an expert is the expert has so much experience that they're like just aware of certain things and features that like for a beginner, like someone starting out like myself or someone listening, it's like we wouldn't even think about. And so what are even some of like the, the features or like the aspects of Replit that a lot of the times you see people entirely ignore, but like if they took it into account, it would be tremendously valuable even when they're, even if they're just building their first product?

    11. MPMatt Palmer1:01:22

        100%. This thing, this environment that we're in right now, it's like a computer. Think about it like a computer. It's running in the cloud. We're accessing it through our browser, but this is like if you opened up your laptop and you installed a bunch of stuff. What that means is that this is gonna get a little technical, right? But there is a, a shell. Like if I wanted to go over to here and type in shell, like this is the command for listing files. I have a file system. So under the hood, this platform has access to everything a developer could do. So what I'm showing you right now are things a developer would do. Maybe they wanna install languages, maybe they wanna configure that environment. Our agent is doing all of those things. And people take this for granted, right? If you're building, like if you build something with like Claude, you can build like Artifact or whatever, or some of, some other tools. What you don't realize is that's not actually an environment. It's not how a developer would build an app, and that's why it's really hard to get that thing running. Like maybe you have to, if you tried to build anything before, you have to figure out how to set it all up on your own computer. You have to figure out how to add a data source. I'm gonna show you exactly how to add a data source. You have to figure out how to do all of these things, all of these things that Replit offers kind of for free because it's built on the platform, and the platform is the thing that's really powerful. And so I think a lot of people get confused when they come in here and they see Replit, uh, because they don't understand, like actually what we did was we built all these tools a developer would use. We built all of this

14. 1:02:45 – 1:06:40

    PUBLISH your AI app in one click — deploy to the web

    1. MPMatt Palmer1:02:45

       really complicated infrastructure, and then we just gave AI access to it. And so the other platforms out there don't have access to all the different tools, right? And, and that's a really big differentiator. Like what it means to deploy something, what it means to add a database. These are all things that sometimes people take for granted. What it means to run a, a project in Replit is the fact that I could take this and open up my phone. I mean, we'll see if this works right now. I'm pretty sure I'm logged in. Um, and if I go to my apps on my phone, refresh it, uh, and then open up what we're building, you can probably, maybe you can see this. This is the same workspace actually-

    2. CJCalum Johnson1:03:20

       Mm

    3. MPMatt Palmer1:03:21

       ... that's running on my phone. So like in real time, I can see what I'm building because it's in the cloud, right? And so then if I say I had a Lyft coming, I just take my phone with me and I can keep prompting and keep building, keep seeing what things are doing. And so like real time, you can build in a completely isolated environment. You can take it with you on the go, and it has access to all the same tools a developer would have access to, which as we build out this app, you're gonna see enables some pretty powerful things.

    4. CJCalum Johnson1:03:47

       Yeah. Okay, you know what? 'Cause I, I, um, I wanna continue for people and, and go a bit deeper. If someone-- So it gives you kind of like the plan of what it's gonna build. As you're like assessing the plan, 'cause I'm just even thinking about the person sitting at home doing this right now, and you're like reviewing the plan. You're like, "Okay, this sounds good." Like at what point are you actually hitting go on it? Like what, what is kind of giving you pause? Like what is making you go back to the drawing board versus just like, "Okay, let me just run this and see what it comes out with"?

    5. MPMatt Palmer1:04:20

       Yeah, no, I mean, I basically, I just send it most of the time. I just do a quick glance to make sure it's not like do something crazy, right? Like I just wanna make sure there's nothing in the plan that, that is too complicated. I'm actually thinking from simplicity. I'm like, "Is this plan simple enough?" Because I want something that works on the first run. And so I'm really thinking, is this simple enough? I'm, I'm basically looking for things that would, I would want to remove, not add necessarily. So just send. Most of the time I'm just sending the plan, to be honest with you.

    6. CJCalum Johnson1:04:50

       Hmm. What is an example of something like if you did see it, it would be like a red flag? Like you would actually be like, "Okay, I actually want to remove that"?

    7. MPMatt Palmer1:04:57

       Yep. I think about complexity. I, I think an overarching theme hopefully that you get from this, uh, from this example is simplicity over everything. I talked about consistency already. S- simplicity and consistency over everything. If I see that it's trying to build out, right, like if I'm trying to build Uber and I'm like, "How do I get a million people to use my app?" That, that's gonna be really complicated. I just want a timer that works here. That's what we're gonna get to. So that would be like my example, right? Like if you start to see really complicated architectures, things that like you're not understanding or that are unrelated to your vision, start with an MVP. Boil it down to the simplest possible thing that you can create, and then start marching towards that step by step because, um, again, the goal here is to get to something that works. The, the, the way I think about it is like to tinker, it's a lot easy to tinker if you have something that works, so we wanna get to the simplest possible thing that works and then iterate to build something really cool. Um, that's like, hey look, if I wanna build, if I wanna understand how a clock works, I can just take a clock off the wall, you know, like anal-- back in the day, old school analog clocks, right? It's easier to open up something, open up a piece of electronics, open up like a mechanical watch and look at it and see like, oh, I can see how those things fit together and I can think about what I would wanna add, versus spending all this time planning, spending all this time building from scratch, building a, a completely new clock because you're just into being really complex. You know, I think if we can just start from the simple pieces, that's, that's the best bet.

    8. CJCalum Johnson1:06:18

       Yeah. It's almo- it's almost like, okay, what is the core feature and then making sure that that core works. And like once, once-

    9. MPMatt Palmer1:06:27

       100%

    10. CJCalum Johnson1:06:27

        ... you kind of have confidence of that, it's like we're ready to go. Okay, so, so what i- if we, if we hit deploy, well not deploy, if we hit go and then we can kind of talk about like the step three, which is perfect

15. 1:06:40 – 1:14:50

    Design tweaks with screenshots & Figma import

    1. MPMatt Palmer1:06:40

       Of course. Yeah, no, let's, let's dig into it, right? So, you know, Agent ran this. It took about-- We were talking the whole time, um, took a couple minutes, I would say. Um, you know, like the first thing is it's a timer, so if I hit play, the timer, number one, this doesn't appear to work, so [chuckles] you're gonna see in real time, right? Like we're building something here. This is, uh-

    2. CJCalum Johnson1:06:58

       This is what we want

    3. MPMatt Palmer1:06:59

       ... this is different.

    4. CJCalum Johnson1:06:59

       Yeah.

    5. MPMatt Palmer1:07:00

       [chuckles] Right. So we have the, the layout. It's telling us to inhale and hold, it looks like every four seconds, exhale, right? That's box breathing. So if you don't know box breathing, it's inhale four, hold four, exhale four, hold four. That's kind of the pattern, right? So th- that's actually working. So it's keeping track of the time, but it doesn't look like it's actually updating the state. This might seem pedantic. This is important, right? Because what we're gonna talk about now is actually the process of debugging, and that's like a really important skill in Vibe coding, uh, in building your ideas. Um, and so I'm gonna switch over to a different tab. What is debugging? What do I mean when I say... Sorry.

    6. CJCalum Johnson1:07:38

       And ju- just quickly, 'cause you, you said, you had such a great quote on this. I just wanted to read it out before you go deeper. You said, uh, "You try something, it breaks, but now the AI helps you fix it. That moment when you debug with AI is the moment you start believing you can build anything."

    7. MPMatt Palmer1:07:59

       Totally agreed. Totally agreed. You know, and so actually more than, before we-- I'll, I'll switch back over to, to Agent, um, and I'll say like, "The timer isn't working." Uh, that's actually not descriptive enough. "The timer isn't counting down, but I see the box breathing, um, like assistant moving." I'm being very descriptive about what's going on, and, uh, "Can you fix it?" So the thing about Agent is that it has access to this entire environment. So behind the scenes, what's going on is there's, like you can look at the console and actually see like the console's just a place that shows you what the, the server of the, of the database or of the application does, right? Um, then there's a front end kind of console as well. And what you'll see is that Ag-Agent is saying, "I need to fix the timer to- countdown issue. I can see there's an LSP error in the storage layer." I don't even know what the heck that is. I'm not too worried about it, but [chuckles] it's gonna look at it and start fixing things. Now it's saying, "I need to examine that timer logic a bit cl- more closely. I think it might be this thing or that thing." So you can learn about the implementation, what's going on. Again, this is all kind of complicated. You don't have to worry about it too much. But now you can see that Agent is going through these things. It's testing them to make sure that it works properly, seeing some React errors, and it, it's actually doing the debugging work for me, right? It's doing a lot of the, the nitty-gritty for me. How do I tie this back to what I was gonna say before? Debugging is a process of observation. You know, we're finding bugs in the app and then we're removing them. And at the simplest, right, maybe you have a lamp that doesn't work. The first thing you ask, the question you ask is, "Is the lamp plugged in?" Now, with AI, uh, without, let's talk about without AI. You'd have to go look if the lamp is plugged in. Now, with AI, you can say, "Is the lamp plugged in?" AI's gonna figure it out for you because Agent has access to that environment, right? So I sent that command. It's like going through and basically debugging the application for me. But the thing I had to do, we talked about context, was provide context. I had to visually inspect my app and say, "What isn't working?" And I had to break that down logically. "Hey, this part of the application isn't working, but this other part is." And I'm gonna give that information to AI to kind of get to the root of the issue, right, and tell the LLM, tell the, the system what's wrong. And that kind of, that flows into context, right? Like context can be the prompt that we provide to a- to AI, but it can also be images, documentation, errors, other details about the application that AI doesn't know about. So all we're doing is giving context to AI to help it debug. And, you know, we're building this app. Th- this is another great example because I built this app [chuckles] before this podcast. Obviously, I wanna make sure that it works, and it worked on the first time. And now, of course, I do it live and it doesn't. But that's like part of it, right? So we're having fun together. I talked about the countdown timer not working. I click it, and now it works, right?

    8. CJCalum Johnson1:10:57

       Yeah.

    9. MPMatt Palmer1:10:58

       And this is simple. This is like maybe you're saying, "Hey, Matt, that should've just worked on the first time anyway." You're probably right. I'll take this back to the team. But these are the foundations of what I wanna talk about, 'cause you're gonna build apps and you're gonna be like, "Why doesn't this thing work? Why does this other thing work? I don't know." Right? Like, does the pause button work? Yes. Does the stop button work? It does. It resets. And when I press start, now I'm getting this nice little countdown. Did we break anything else, right? You might be asking, can I go through and start box breathing? The box breathing timer works. Hold, two, three, exhale, right? So now this is working. Now we have a little timer that's going according to our rhythm. Perfect. So I really like this. It looks like for the most part it's working. The hard part, right, is how do we track these sessions? How do we make it so that anybody can log in? And this is one of those things you wanted to know what sets Replit apart, why should be- people be using Replit. We have all those tools available. So what I'm gonna say here is add, um, user login, also known as authentication, you know, some people might call it that, uh, with Replit Auth, um, and store user session data, um, so that I can visualize it for my account. I'm gonna send that.

    10. CJCalum Johnson1:12:13

        So this is almost like if someone, say that I had an app and instead of, you know, every time you go in, it's a fresh s- session, like someone's actually able to build an account where it like logs their data and like tracks it over time. That's essentially-

    11. MPMatt Palmer1:12:27

        Yep

    12. CJCalum Johnson1:12:27

        ... what you're building right now?

    13. MPMatt Palmer1:12:29

        Yeah. So, right, if I log into Replit, I have my Replit account, and I can see my apps. If you have an account, if you have an app that you want to share with people, well, they need to be able to customize that app. They need to be able to save information That's actually called authentication, um, or, or user login, right? Basically, you log in with your Replit account, or your users would create a Replit account to log in. It looks like they're logging into your app.

    14. CJCalum Johnson1:12:52

        Yeah. You know, it's, uh, it's really cool and it's like, I think it's actually a really cool part of the experience is kind of the thing that I said earlier is for 99% of us, we're only used to using apps as a consumer. And I think it goes back to that point of why it's so important to start and, like, move quickly and deploy and publish is as you go through this process, it's actually kind of teaching you some of, like, the foundational elements of actually building something with software.

    15. MPMatt Palmer1:13:23

        I think that's exactly right, and that there are things, um, I forget sometimes, uh, or that I've learned just by building. And so I want to call out a few things. First, if you're looking at what I'm building and you're saying, "How did Matt know how to do that?" Well, it's actually, like, pattern matching, because as you get good at building things, you will be able to match, like, "Oh, I s- I've seen this thing before. I know how to solve this problem." As a human being, you'll, you'll be able to do that, I promise. Um, and then two, knowing about what exists. And you also learn that by building. And so, you know, the way I talk about it a lot is I talk about frameworks. And so this is like React is just a framework for writing a certain type of code. You don't have to worry about that. The important thing that you do have to worry about is that you don't know what you don't know, right? So exactly what you just said. If you think, "Oh, it's gonna be a walk in the park for me to add a user login," and then you go down a two-hour rabbit hole of trying to do that, you're gonna learn a lesson and you're gonna waste some time, right? So my favorite part about the job is showing people what they can do and how to get over all the stuff that I struggled with. Don't do that. Do it the simple way, right? How do I do the thing I want to? What frameworks allow me to accomplish this thing? What frameworks did somebody else build that I can just plug into my app, or do I need to build out? What works best with LLMs? If you don't know, you can just ask LLMs, or to your point, you can just start building. And so super important, super important lesson, right, is to, like, learn to know what you don't know. That's a super hard, uh, part of really life. So, um, I don't want to trivialize that. Um,

16. 1:14:50 – 1:23:20

    How To Market Your New Business! ($1M Apps do this!)

    1. MPMatt Palmer1:14:50

       it's difficult. Uh, and you know, it's something that we're gonna need to work through.

    2. CJCalum Johnson1:14:54

       Yeah. And you know what? I think, um, I even, I, I wanted to read it again when you said, "That moment when you debug with AI is the moment you start believing you can build anything." And I, like, the, it really just stuck out to me because I think that a lot of people get discouraged. And I can even imagine for myself, I would get discouraged if you watch a video like this, you rush out there, you start trying to build the thing, and it comes back and it's like, it seems like the thing that should be simple is broken. And it's actually that descriptive process that you kind of outlined where you're kind of explaining, like, what went wrong back to the agent and then it fixes it, it's actually giving you momentum. Like, it's actually back to that thing you said, you'll start to believe that you can build anything. It's actually when things get broken, in a way it's this reframe of, like, you should look forward to it, because it's like you're actually going through the process. Like, that's how it should be. Um, yeah, I, I wanted to make that point for people.

    3. MPMatt Palmer1:15:58

       Thank you. I love that. Um, and th- I think about this every day, you know, because separate from Matt at Replit, I'm Matt, a human being too. I think we all see these things in life and I, I want to tie this a little bit back to life, right? Because, um, you know, I'm sure our viewers out there, everybody has a life. And I'm sure, well, I'm not-- I, I know it for a fact. Everybody at some point is going through something that's kind of sucks or is that, that's hard or that's baggage or that's, you know, sets you back or that you feel like is unfair, that, you know, you get resentful at the world for this thing that's happening to you, right? And you might want to feel like a victim, like, "Oh, woe is me," like, "There's nothing I can do about it." You can get momentum by doing hard things and by not even doing hard things, just building stuff. Maybe it's hard, maybe it's easy. You get momentum by executing, by waking up and saying, "I'm gonna go to the gym and then I'm gonna go to work, and then I'm gonna do good work, and then I'm gonna go home and I'm gonna, like, do the right stuff for my family," right? And that's all, all this stuff is very important, right? So I don't want to tie that back to, hey, you know, now you can use Replit. But building these things, Replit is a platform for you to build your ideas. And right v- off the bat, we saw a problem, we saw an error. Could've just gave up, and I probably would've went, [chuckles] I would've been like, "Dang, I got on this podcast and I failed." Sometimes you're gonna fail. That's, that's part of life. But just one prompt, observing things and oh, now, now it feels a little easier, right? Now the pressure's off. Now I think, well, if I come to a problem, I actually can solve it just by being clever or by, like, working hard or trying things out. So to tie that back, that back to my, my larger point, right? Like, you build momentum, exactly what you're saying, right, into your life by seeing challenges and facing them down and, you know, becoming [chuckles] basically a better person. And, and that's really important. And, and you can do those things through creating with AI, and that's the point I'm trying to drive home.

    4. CJCalum Johnson1:17:47

       Hmm. Yeah, no, I think, I think it's so brilliant. Okay, you know what? If we just, if we wanted to fast-forward, let's assume that we were like, we were ready to deploy. So we've, like, we've moved through the, the perfecting it phase. Like, we've gone through the steps, we've debugged, we've, like, made sure that it's working, we're ready to deploy. I think one of the cool things with, with Replit when I saw it is, like, it's so easy to actually deploy the thing. Could you demonstrate that?

    5. MPMatt Palmer1:18:14

       Absolutely. Um, so assuming we have one minute here, let's just see if this worked because we tried to implement our login here. So if I click Get Started, let's see what happens. Um, we're gonna log in. So now I get a pop-up. You probably can't see this, but I'm just logging in with my Replit account. And now we got our app. Top right, that's my profile picture, so something worked. Our app still works, hopefully, uh, like we expected. So I can start my medica- meditation countdown, but now I'm logged into my app and I should be tracking sessions. So if I pause this or, like, say stop it, let's see what happened. Um, we'll refresh to see if anything updates. Um- So we didn't get our session. Just wanna kind of recap what we got to. But we have a database, and if I go click see my data, I have meditation sessions, I have users. If I clicked user, I should see me, like my email will probably pop up, so we won't do that just now. But we have all of the sort of basics of being able to log in, create user accounts, right? Uh, we have all the basics there, and now it would just be like quite literally one follow-up prompt of saying like, um, you know, "Implement, uh, session tracking for my user." And then we'd have kind of exactly what we described at the outset. Now, I think that's astounding. To go back to what you said, what, what does, what does deploying mean, right? Everything that we just built is in its own little environment. So I said think about this like a computer. Well, what does that mean? Well, that means that if it's running on this computer and we just took a snapshot of that thing and deployed it, well, then it's gonna be running in the cloud, and that's the amazing part about Replit. So if I click deploy, you get some configuration settings. This can be confusing. We're not gonna worry about that right now. Most of the time you can just click appl- approve. We're gonna call this Zen Timer. Um, we'll call it, yeah, zentimer, uh, .replit.app. Um, and we're gonna go down here and click deploy. Some other notes, you could run a security scan to make sure that this is secure and that there aren't any gaps. You could, um, link a custom deployment, so your own or a custom URL, so like matt, mattpalmer.com or something, right?

    6. CJCalum Johnson1:20:16

       Mm-hmm.

    7. MPMatt Palmer1:20:17

       Um, to be able to go and, to, to an actual website. But what we're doing is we're taking a snapshot of this environment and pushing it up to the internet. Now, that snapshot piece is really important because a lot of times people build stuff and it breaks. So one other thing I wanna call out is that when we're building this, this application, oop, it's getting a little sticky there, um, Agent is taking snapshots or making checkpoints of our apps, just like a video game. So here we have a checkpoint that was made 24 minutes ago while we were talking. It's after we built the core UI. And, at any time I can preview what that looks like. I can roll back to that checkpoint. So we can roll back the files, the agent, actually our entire database, all of our data as well, and we can know that we're completely safe. And the amazing thing about Replit is that you don't have to set any of this up. You don't have to, to worry about losing your work. It also means you don't need to worry about failing, right? 'Cause if I went kind of crazy and started implementing a bunch of experimental stuff like we're doing right now and it broke, it's okay, because I can just roll back to the last checkpoint and continue building.

    8. CJCalum Johnson1:21:16

       Mm-hmm.

    9. MPMatt Palmer1:21:16

       And so what we were talking about earlier, right, this mindset of not being afraid to fail, of like randomness, of all this stuff, well, your work's actually always saved, so you can be more, um, adventurous. You can be, you can take risks and not worry about messing anything up. And, and so to tie that back in, right, I talked, I spent a couple minutes talking, a couple minutes yapping. We have our app deployed. Uh, I'm gonna switch to that app. We get exactly what we had before. I'm gonna click get started, and this is the login screen I was talking about before. Gonna log in, so now I have my timer. It's, uh, working exactly like we'd like. You can't see this, but it's at zen-timer.replit.app. I can leave this up for your viewers. They can play around with it. Um, now we got a countdown. I just asked it to implement our session tracking, so maybe if I pause or stop, see. Does anything work or are we gonna have to-- would this be a future-- That might be a future endeavor. So session tracking, that can be pretty difficult. Um, but the bare bones of our app, we built it, we took an idea, we turned it into reality, and we deployed it live to its own website.

    10. CJCalum Johnson1:22:17

        Nice. So this is actually-- So people can actually go and see this. Like this is actually live on the-

    11. MPMatt Palmer1:22:23

        Yep

    12. CJCalum Johnson1:22:23

        ... internet right now.

    13. MPMatt Palmer1:22:24

        This is Zen Timer, zen-timer, we'll go back over here, uh,.replit.app. So if they're watching, um, oh, I'm on the wrong tab. So if they're watching, uh, and they wanted to just like, you know, scan this QR code, we could throw that up. I'll leave this app up, um, for your viewers to check out.

    14. CJCalum Johnson1:22:40

        Yeah. And, and so just quickly, 'cause I, I wanna actually, I wanna spend a few minutes talking about, um, I think the thing that like in a way builders sometimes dread, which is like marketing, actually getting people to, to use, like or getting your first few users. Um, but before, before we get into all of that, in terms of like, 'cause I can imagine it for me, like if I, if I launched it and it's now live on the internet, the first thing that I would think of is actually like almost the design, like how it looks. Is there a way to, like say that there's certain websites

17. 1:23:20 – 1:31:40

    How to get your FIRST users (Matt’s playbook)

    1. CJCalum Johnson1:23:20

       that I really like the look of, I could actually go back into Replit and like be like, "Okay, I really love the way that like the Apple page is designed," or Nike or Warby Parker, like whatever the brand is, Airbnb. Is there a way that it can actually almost design this page so it looks a bit more like aesthetic, so it looks a bit more similar to something that I might see from other-

    2. MPMatt Palmer1:23:46

       Absolutely

    3. CJCalum Johnson1:23:47

       ... bigger brands?

    4. MPMatt Palmer1:23:48

       Absolutely. And, um, you know, I, I think we coulda got, I would've loved to get into this a little bit earlier. There were some other features I kind of planned to build out. But one thing I wanna drive home is that you can give all kinds of context to Agent. So, you know, say we, we see a timer that, that we really like. Um, oh, well, like, hey, here's one, right? Maybe I want my timer to look a bit more like this. I just took a screenshot of that page, now I'm gonna go back to Replit and I'm gonna paste that screenshot in, and I'm gonna say, "Make it look like this." Or even one, one step further, right? Um, what if we just have the URL, matstimer.com? You type in the, the URL and you can click take screenshot, and then you get a screenshot of the application. So you can take screenshots, provide them as context. You can just type in a URL and, uh, and provide them as context. Um, some, some apps, right, like if, if the app doesn't ac- if the URL doesn't exist, right, that would fail. Um, you can do a couple other things, right? So, uh, I was thinking, hey, maybe I wanna add white noise to this app. Well, I found a site with white noise on it Uh, download the audio and add it to our app. And I can paste in that URL, click Get Text Content. It's gonna scrape that page. It's gonna pull in the MP3, and it'll add white noise to our app, just because I found that. One other thing f- specifically for people who like to design, we also have a Figma import. I have separate videos on how to cover that. You can design something in Figma, pull it in. You can build in some of your other tools, pull it in, keep building with Agent. Um, these are all really great, uh, really great ways to get started building.

    5. CJCalum Johnson1:25:26

       You know, you know what it comes to mind just even as you were explaining it? It's almost as if, um, it's almost like having a teammate. Like if you had like a... If I was running a business and I had like a web designer on my team, and I spotted something on the website that they deployed, I'm actually like, "I actually want it to look more like this." It's like the way that you were talking to the agent was almost as if it was a teammate and you're just directing it what you want.

    6. MPMatt Palmer1:25:53

       Exactly. Exactly. So think about it like you have a junior developer. Think about it like you have a team and you're the, you're just the director of all your different teammates, and you're helping them kind of figure out what they want to build, helping them, right? Like if I talk to... Say I'm baking a cake, like I'm, I'm making a cake for my friend's, uh, for my friend's birthday. Actually, I'm hiring someone to make a cake, right? So you're getting somebody on your team, and they say, "Well, what cake do you want me to build?" Well, you have to think about what your friend likes. You have to think about all of these things. How do I make this perfect? And then how do I explain that in plain English to an employee to, to help them create what I want? And, and what we're doing with AI is we're using it like an employee or like a coworker. "Hey man, I really want you to build this thing." And I have to be really descriptive to get what I want, and I have to be really precise and understand English and understand from first principles what's actually happening in my app. And these are, these are exactly the skills, um, that I'm using to build this app. And then I'll just call out one other thing, so that actually our session tracking did work. We just had to wait for the session to end. So kinda soup to nuts, we, we actually did get a lot more done, I think, here than, than even I expected with, uh, being able to log sessions and save accounts.

    7. CJCalum Johnson1:27:03

       Nice. Some of that, uh, some of the things that you put in with like even the screenshot, could we actually see it apply that? Like I, I dunno, I'm just curious to just see like what it actually ends up, how closely-

    8. MPMatt Palmer1:27:15

       Yeah, yeah

    9. CJCalum Johnson1:27:15

       ... it's able to replicate, um-

    10. MPMatt Palmer1:27:18

        So that's, that's another thing. We'll talk, we'll talk about this a little bit. When you think about like just giving AI a screenshot, what does that mean? Like AI is going to see an image. What are the limitations? Like we're gonna think from first principles here. What are the limitations to give, giving AI an image? Well, it probably won't know the font, right? It won't know exactly what text it used unless, you know, maybe in a year models are super good. Um, it'll probably match the colors. It can match colors pretty well. Um, it won't have image assets. It won't have all these other things. And so I'm giving you reasons why this might not work, um, because like doing it live and then rebuilding the entire app often doesn't come up with super great, [chuckles] great results.

    11. CJCalum Johnson1:27:58

        Oh, yeah.

    12. MPMatt Palmer1:27:58

        We can, we can do that. Uh, but if you think about from an, from a perspective of, "Well, can I figure out what font it's using? Can I figure out where those images are?" And if you can find the assets, right, if you can find all these different things, you're gonna be able to get a lot closer. Or if you figure out how to build it in Figma or if you find, right, like clever ways of getting AI the things that it needs, right? If I said to you, um, right, "Hey, go clone this website," and I just gave you an image of it, what would you do? You would go out and try and figure out what font was on the website. You'd go and try and figure out where you could find the images. You'd go out and you'd try and figure out how to use the text. And these are all things that AI struggles with. And so from my perspective, if you're trying to clone something, that's still one of those things. I don't want people to think, "I'm just gonna send it a screenshot of Nike and have my own shoe store."

    13. CJCalum Johnson1:28:47

        Mm.

    14. MPMatt Palmer1:28:48

        There's actually, you know... It'll, it'll be able to do web research. So if you said, "Do web research to like understand what fonts there were. Do web research to understand similar images and these things," AI might be able to get us closer there. Um, but along with the screenshot, I'm doing exactly what I just said to you, the work of a human. So if I was gonna add the screenshot, I'd say like, um, "I want my app to look more like this. Uh, do research to understand where you can find a nice..." Now, here's where some of that background knowledge comes in. There, in this, this app, there's what's called a serif font. How do, Matt, how do you know what a serif font is? 'Cause I'm a huge nerd, that's how I know. Um, but these are like the kind of terms you don't know if you don't know, right? A nice serif font, and AI knows what that is. Um, make the app dark mode and find similar free assets online that we can use for our timer. So we could send, we could s- sorry.

    15. CJCalum Johnson1:29:52

        Basically, the more detailed you can go, the more closely it's gonna be able to replicate, uh, the thing. I guess if you just want it to be in like the ballpark of it looking like the, uh, the app or the website that you had in mind, like, yeah, I guess it's a starting point to share the screenshot. But if you really want it to closely resemble, it's like you need to give it as much detail as possible.

    16. MPMatt Palmer1:30:16

        100%. And so this is, but this is where we get into our levels of fidelity and how we think about just quickly cloning something or kind of going the, the distance and, and saying, "Okay, well now I know what I want," right? And what I've found in building with AI, hot take here, it's gonna be really hard for me to drastically change the functionality of this app. If I was like rebuild this entire app and I sent that in a prompt to agent, that might take a long time. It's really hard for AI. It gets confused when projects start getting larger and it's building out the structure. That might be a point. I do this a lot. I'll build something and then I'll be like, "Okay, this is cool," right? Like we have our app. It says, you know, like we have like ocean sounds or something that-

    17. CJCalum Johnson1:30:54

        Oh, okay.

    18. MPMatt Palmer1:30:54

        I, I don't know if you heard that.

    19. CJCalum Johnson1:30:56

        Yeah, I did.

    20. MPMatt Palmer1:30:56

        It's white noise. It's not an o- ocean Okay, we got, we got close. Um, but I want to change the entire structure. And along the way, along the, the way the process of building, I learned what the structure should look like. This is huge, right? Because if we're taking that MVP approach, that prototyping approach, by building, by being observant, you're going to build and then you're gonna say, "Man, I wish I would've done this differently. I wish I would've done that differently. I wish I hadn't made this mistake. I wish I would've just changed my prompt at the beginning to do this other thing." And you can do that now, right? So maybe instead of saying change the entire, uh, application, what I might do, I might go to replit.com, uh, and I might say, "Hey, you know, like I have this app and I want a meditation app that looks like this one." And so don't be afraid to start again,

18. 1:31:40 – 1:43:41

    Consistency • Simplicity • EXECUTION

    1. MPMatt Palmer1:31:40

       start over with new information, right? Don't be afraid to kind of like say, "Hey, I learned this from building with this app and now I want to take it into a new thing." I've done that like four or five times where I built something, it was like kind of janky, but then what I built, I said, "I'm gonna extract all the knowledge that I learned from this and take it into my next application."

    2. CJCalum Johnson1:31:57

       Yeah. I love that. And, and it goes back to that point we were saying in the beginning of like, it's not a linear process. Like sometimes you'll go-

    3. MPMatt Palmer1:32:03

       Mm-hmm

    4. CJCalum Johnson1:32:03

       ... all the way to the end just to go back to step one, and that's actually how it should be. You know what? Before, before we get out of here, um, [lip smacks] I wanted to talk, 'cause you've seen so many products, like there's hundreds of thousands of apps that have been built on Replit. Some are incredibly successful and have like raised money. Some, you know, don't go anywhere. If you were to kind of take those two extreme use cases, the people that tend to get like users and actual like adoption and their thing kind of goes and there's traction, what have you found-- Like when it comes to getting users, what have you found is beneficial? Especially when it's like those first few people, right? 'Cause those are the ones that are kind of tough to get. It's like your thing isn't fully validated yet. You don't have a ton of money to build it. Like h- how, what have you seen work for people when it comes to getting those first few users?

    5. MPMatt Palmer1:33:02

       Yeah. Um, you know, it's hard, so I don't want to say it's easy. Uh, it's hard to build something that, that makes people's lives better. It's hard to build something that people care about. A lot of times you're gonna get that rejection. I think the worst kind of rejection is just being ignored, right? Like building and nobody's even telling you, "Hey Matt, this looks bad." They just ignore what you're building. That, man, that's, that's hard to deal with, right? And so building and shipping is like, it's consistency. It's a number-- it's simplicity. It's thinking about what you're doing and doing it over and over again, right? Like I think some people see creative acts. They see like the act of creating a company is, is creative by nature, right? Like the idea that you have, you have to pick an idea that resonates with people and then execute on that idea. And just like I talked about, the results, they might be luck, right? Like, hey, maybe you had the perfect idea and you fail, and then somebody else does it next year and they succeed. Or like Amjad, right? He had an idea for building AI into Replit. He kind of tried to implement that in like 2020, 2021-ish, kind of flopped, but then two years later it worked, right? Like what does that mean? Well, it means that there is some luck that's involved. It means that there, there's definitely skill and there's definitely creativity, but it means that you might try something and fail and then try it again and it might work. And so I think the things that I talked about, right? Consistency, simplicity, sharing things with other people, these are things that if you build those muscles, you're gonna get really good at. And, you know, you might not get users immediately, but you will eventually. You know, I'm, you're, you're YouTube famous. You're much more YouTube famous than I am, right? Like the, I'm sure the first few videos you made, getting users is no different from getting like views on a video, right? You have to get people's attention. This is where we get, product gets into marketing, right? I have to solve a problem that matters to you, and the solution has to matter enough that you say, "That way that I've been doing things for 10 years, I'm done with that. That way that I've been doing things my entire life, I want to try out this new product." And that's like, that's really hard to get people to do because most people, they just want to do the things the same way. Your idea has to, and execution has to make people say, "I want to try this thing, and this thing is better than the way I did things before," right? And it's the same thing with like a YouTube video in a very micro context. This video is worth my time, right? And that's like, you know, most people, they, they care about their time. Um, and so I'm sure like the first videos I made, nobody watched, nobody said anything about them. But you do it over and over again, and you look at it and you refine your craft and you make the product better and you make your process better, and you make how you share it better with people, and you make it more attention-grabbing and you make it more informative and more educational. And then eventually you wake up and you're like, "Wow, people want to view my stuff. They want to use my product." So it's, it's all consistency, simplicity, execution.

    6. CJCalum Johnson1:35:44

       You know what? You, you, you said something, you, you mentioned, you've mentioned it several times. It's been like a reoccurring theme of, um, not just building but actually sharing it with others. And it's really interesting because I've been paying attention to a lot of the people that have been like Vibe coding and building products and like been successful at doing it with AI. And there's a part of their marketing strategy which I love, which is they kind of build in public. And so like as they're building the thing and they're coming across bugs or they're coming across issues or like they have a two, three-hour session of using like a Replit and they get to a certain sticking point, it's like they're documenting the whole thing and then they'll share like updates with people on Twitter. And it's like this really effective way of doing this like very subtle marketing, where it's like as you're building the rocket ship, it's like you're telling people like, "Yeah, this thing just fell off. Like the, the rocket exploded, like the engine didn't work." And it's like gradually building that base and then once you launch it, there's actually like a base of people that have been following the entire journey and are like ready and willing to be those first few users that you need. It's like this very organic, just authentic approach to marketing.

    7. MPMatt Palmer1:37:08

       Yeah. And, and you know, if we're talking about how that ties into marketing, right, it's building a brand because we are all different human beings. And like the people ask me, they'll say, "Oh, you know, like I had this idea for content and I just didn't start." This go back, goes back to the perfectionist mindset. "Oh, well, like I had this idea for a conversation or a video." One really good friend of mine, she was like, "Hey, let's just go have a conversation." And we had the conversation. It was an amazing conversation. And she was like, "Yeah, you know, I've always wanted to start a podcast." And I was like, "Just get people in front of you and do what you just did with me now because that's excellent," right? And so when you think about building a personal brand, you just go out there and you-- like I just, I am me on Twitter. If you go follow me on Twitter, on YouTube, like hopefully it comes across this is just who I am. And as it would turn out, people like people a lot more than they like faceless, nameless brands, right? And so if you can be yourself, you're gonna find other people. Maybe it's not like, you know, tech nerdy stuff, that's who I am, but maybe you really love sports or you love basketball or you love, I don't know, going outside on a hike, whatever. If you can express that, you're going to attract other people who like the same thing and that want to be around you because you give off good energy. And so that's my approach to like marketing. You know, there are all these things written about how to market to different types of people, but at the end of the day, it's just being authentic. And building in public is really powerful because you go out and you show people and you fail. Now the important part, right, is you learn from your mistakes. Every time I fail as-- that's actually an opportunity for content because I'm like, "Oh, my app broke. I'm gonna show people how they can avoid this same process." And then this is like I like to do how do I do one thing that does like a lot of things. Well, I'm gonna go build something. I'm gonna fail. Along the way I'm failing, but I'm also learning. And then I'm not only learning, but I'm creating something to help other people. And so you're creating stuff, you're learning, and you're sharing it with other people, which then helps them build a personal brand and is quote unquote like marketing technically. Man, that's pretty cool.

    8. CJCalum Johnson1:39:01

       Yeah. It's like kind of like this irony to like doing content and marketing is like in the beginning, and, and this is what it was for me, you come in and you think it's all about being polished. And the reason you think that is because-

    9. MPMatt Palmer1:39:13

       Yeah

    10. CJCalum Johnson1:39:13

        ... you're so used to watching like the other people who are huge content creators-

    11. MPMatt Palmer1:39:18

        Yeah

    12. CJCalum Johnson1:39:18

        ... who have been doing it for like, you know, MrBeast has been doing it since he was 12 years old. And so you see-

    13. MPMatt Palmer1:39:22

        For decades, yeah

    14. CJCalum Johnson1:39:23

        ... you see people who are like polished in the craft and you're like, "Okay, I'm gonna try and be that," and it gets-

    15. MPMatt Palmer1:39:29

        Mm-hmm

    16. CJCalum Johnson1:39:29

        ... no traction. And then y- if you keep doing it long enough, you'll get to the point where like you're so comfortable, whether it's on camera, you've just done so many repetitions, you-- it's like you're comfortable enough to just be yourself. And in that moment where you're comfortable enough to just be yourself is typically, from what I've seen in my journey and then also for others, that's when your stuff starts taking off because people can see themselves in it, and it's like this irony of doing content of like once you stop trying to be perfect and show the perfect path and everything worked and my story was just this perfect linear story, once you stop trying to do that is actually once it works. Um, [lips smack] you know what? Just quickly before we get out of here, I think back to like, uh, what you shared in the beginning and like you're working in that finance job where you're just punching numbers into a spreadsheet. And I, and I just wanna do a little thought experiment with you of like if the Matt that is sitting before me today almost like went back in time and he had to go and speak with the Matt that's there punching numbers into the spreadsheet at his desk at his finance job, and let's say that like Replit and all the AI tools, like we're in 2025, everything is available. If you were to give that Matt maybe just pause him for a second on punching those numbers in. If you were to give him just a challenge, like something to do this weekend that like, okay, if you start doing this thing, you're not gonna have an incredible business overnight, but you're at least gonna have like some momentum, like something-- you're gonna have a win under your belt, something that you've built. What challenge would you give that version of Matt?

    17. MPMatt Palmer1:41:22

        Absolutely. Um, yeah, no, I, I think this one's simple. I'd say go build an app on Replit. Go take your idea and turn it into something and prove that you can change the way that things exist in the world, right? And, you know, an, an a- app might not seem like you're doing that, but that's what you're doing. You're creating something. So go take that idea, do what we just did, plan it out, break it down, create it, and then what's the next step after you create it? Well, we just talked about building in public. It's to share it. So I'd say go on social media, go on Twitter, and share what you built with somebody. Um, and you know, I think we'll share some links, uh, to, to make it easier to do just that. But practice the habit, the muscle of idea, something that works, share it with people. I think that's the first step.

    18. CJCalum Johnson1:42:08

        Yeah. I love it, I love it. And you know what? L- just, just to build on that. So I think if-- I'll put both of our Twitter handles in the description. If people at us, uh, and have the link for whatever they build, uh, we can retweet some of them and maybe even, I don't know. We'll, we'll, we'll-- I'll have to check this with you, Matt, but maybe we can even have the Replit account retweet some of them as well.

    19. MPMatt Palmer1:42:33

        I m- I might be, I might have access to that account. I won't, I won't, you know, say anything for sure. And you know, what's bigger than getting the Replit account to retweet something is getting Amjad to retweet it and, you know, like he retweets really good apps. So if you build something really cool, you tag me, you tag, uh, you tag this show, right? And then we can get it retweeted by Replit. Maybe Amjad sees it. That would be pretty huge.

    20. CJCalum Johnson1:42:54

        Let's do it. I love it. Matt, so thank you so much, man. It's been a pleasure.

    21. MPMatt Palmer1:42:59

        Uh, honestly, the pleasure was all mine. I love talking about this types of, this type of stuff, not just building, but like the stuff we get down to that like experience of being human, especially of being like someone that's frustrated that thinks that they're capable of more, that, that wakes up every day and says, "I can be better," but they have to deal with like reality, which says at the moment, "What have you done that proves that you are better?" And now you can go out and do things that prove that to yourself, that help you take the next step, and that eventually gets you to where you want to be. So that all makes me super excited. Thanks so much for having me on.

    22. CJCalum Johnson1:43:30

        It's been awesome, man. Nice, man. This was good. Okay, so if you enjoyed this episode, you're gonna absolutely love what we did with this one right here. Click here.

Episode duration: 1:43:42