321 | How to Build an AI App Without Coding: From Idea to App Store with Replit & Claude by Bryce Rattner Keithley

0:00Hello and welcome to the Leveraging AI podcast, the podcast that shares practical, ethical ways to leverage AI to improve efficiency, grow your business, and advance your career. This is Isar Maitis, your host, and I got a really fun, interesting, and helpful episode for you today. Almost everyone have an idea of something that can help them, either help them in their life, help other people, help their business, and most of us just keep it as a good idea and don't actually do anything about it. Now, usually we don't do anything about it because you feel that you either don't have the knowledge or the time or the budget or some other reason or some kind of combination of the above of why not take your idea and turn it into life.

0:39But these days are over because AI enables you to take your ideas and turn them into whatever they need to turn into, including full applications that can do really amazing things for you and for other people. Now, to do this, again, this used to take having a product team and a development team and a research team and a deployment team and IT people and DevOps and all these kind of things that mean nothing to most people. But in today's episode, we're going to show you that literally anyone can do this with zero technical knowledge. Our guest today, Bryce Ratner, is a busy mom. She has other stuff to do other than developing applications, and she has no technical background whatsoever.

1:21And yet, right now, she has an app in the App Store that people around the world are downloading and using. And in her case, it's just to help other people. That was the whole purpose. But the goal of today's session is to inspire you to take the idea that you have in your head and show you the steps, how to go from an idea in your head to a working application that you and other people can use. And like I said, first of all, this particular use case is really, really cool and really fun the way it actually works. And I will let Bryce share with you exactly what it is and how it is. But the idea is not that.

1:53The idea is to show you, again, that with no skills, no technical background, you can actually build applications that can help you, can help other people either in your personal life or in your business. Because I think this will inspire many of you to actually take action and do these kind of things, I am truly excited to welcome Bryce to the show. Bryce, welcome to Leveraging AI.

2:15Hi, Ishar. Thanks for having me. This is so cool. You and I chatted twice already about this thing. And I think your story is really, again, I use that word already, inspiring, right? Because you really don't have any technical background. So I think what would be cool is to, first of all, take people back to how this whole idea started. And then we can dive into the actual technical details of what you built and then how you build it. Absolutely. As Ishar mentioned, while I have no hard technical skills, I've spent most of my career in tech. I've worked in recruiting and talent strategy for some well-known companies, and I've worked alongside extraordinarily talented technologists and builders and seen incredible products come to life.

2:58And there is an interesting duality to that exposure of interacting with some of the top paid, most sought after magicians of technology and having it fully abstracted to me how they do what they do. And on the other side, spending time either bartering or going to head to head negotiating with these people, you also know that they're kind of blue blooded, imperfect humans like the rest of us. And so the genesis of this idea, As Isar mentioned, I'm a mom. My kids were really young when the pandemic started. And as a true survival mechanism, I started doing what I now know is called micro workouts, which was like doing 10 pushups here or 10 squats here while waiting for coffee or next to, you know, bath time or whatever.

3:46And I had this idea that if someone just told me what the exercise was every day, I would do 100 of them. There was a 100 pushup a day challenge for like a hot second. And so as some of the buildable apps came to be last fall, I was like, I should really play around with this technology and see what's going on. What's something lightweight? And I thought I could probably push an idea to myself, a workout to myself every day. And why not start there? So that was the beginning of it. And like Isar said, I just kind of slippery slope. It's now in the app store. very cool so so it really came initially as your personal need right you weren't thinking about building an application for other people you definitely were not thinking putting it in the app store what did you do very very first step you're like okay i want something to send me a message every single day with an idea for an exercise and hopefully how to do the exercise because some of them have weird names you don't really understand totally and before even leaning on technology, I tried to just beg my friends who know more about fitness than I do.

4:47Can you just send me? And like, that's not reliable to do that. But it was sort of thinking, oh, this seems like a thing that this AI LLM could do. Meaning to your point of the ideas, okay, I could probably generate a list of 30 body weight exercises to start with. So I don't have to do that thought exercise. I can show you, I actually built the MVP, the V001 of it in parallel in lovable and replet at the same time, just to kind of see what would happen. And it was, it was in that. And I actually think to your point of me lowering the bar and making this accessible, it was having a beginner's mindset that asking a really high level greenfield question and seeing how much came, even if not perfect, but how much came in terms of progress lit a spark that really warmed the waters and made it interesting and friendly and approachable to continue.

5:45Do you want to see how I started? And lured you in, right? Lured me in. So while you're opening the thing, yes, I want to see it. While you're opening it, I'll say a few things for the people who are not from the tech world. So MVP stands for minimal viable product, which basically means what is the smallest thing you can build that still provides value? It's a big deal in the software world where you want to build something and you don't want to build for six months. You want to build for the shortest amount of time to see that it actually provides value and works. So that's kind of like the first thing that you said.

6:14The two applications that you mentioned are Lavable and Replit. So there's a pretty wide range today of applications that allow you to develop software while not really knowing how to write code or create anything. They fall into different categories, and there's today more and more overlap between these categories. But on a very high level, there is the more entry-level, user-friendly, I don't need to know anything applications, which on the top of that list are lovable in Replit and probably Base44 now. So these three applications allow you to basically come and know nothing about how is it stored and where the code goes and what the hell is the database and how it's deployed and all these things that software people need to know are all built into it.

6:54And you literally just need to ask things in English. So that's kind of letting you know. And then you get to the more advanced ones like Cursor and GitHub Copilot and stuff like that, that are built more for developers, even though day-to-day people like us can also use it as well. And then there's the stuff that's kind of in the middle, like just the raw cloud code and stuff like that. So this is kind of like putting things in perspective. So you kind of understand what we're talking about. Now I'll hand it back to Bryce that is already sharing her screen. This is what the app looks like today.

7:24So this is where we're working with today. And if you want to do standing hip circles, that's August 20th challenge. So before you move, because a lot of people are going to be listening to this. So what we're looking at is a really stylish page that has a cat in a gym, a humanoid cat, right? It's not a cat, but it looks like a human that looks like a cat. And it's doing an exercise in a well-lit gym, explaining or demonstrating, I should say, how the exercise looks like. And you can pick different things and variables, I guess, and timers and stuff like that underneath that. So it looks very professional.

7:58The video looks really, really cool. And so this is the current version, right? This is the current version. This is Daily One. Is there a different animal for every exercise? Not for every... There's like maybe a dozen, like 12 to 15. There's a fox, there's a rabbit, there's a turtle, there's some anthropomorphic mythological creatures as well. There's a genie and an angel. There's a lot. We've had, let my kids choose what we make for some of them as well. This is so funny. But I started in a really basic way. So apparently my replet history doesn't go back this far, but I'm fairly certain I copied and pasted the same prompt.

8:33And this is where I started. Create a text interface app that will prompt users on a daily 100, a different exercise each day, selected on rotation from a list of about 30, and to complete 100 reps throughout the day. Users will reply back with numbers completed in each set, and the app will log and track the count and celebrate with emojis when hitting 100 each day. App will also track with a red to yellow to green color continuum the streak the user has mained out of 100 each day. That was where I started. YOLO. Just, this is my idea. And as I scroll, you'll see I actually never replied. I just wrote this.

9:14and it came up with like basically that. Issa, I don't know if you want to narrate and talk a little about what you're seeing, but that was my magic. It was in seconds, minutes. Yeah, so again, going from an idea to version zero that does the very basic thing you requested requires literally just that, just asking any one of these apps. And yes, the user interface is not as slick at the current one, but it still looks very reasonable and modern. It gives you the exercise you need to do without any video and demonstration because that was not requested. It tells you what's today's challenge, how many reps you've completed, allows you to log, put in a number and hit log set and it adds it to the counter.

9:53And that's basically it and shows you a streak. So - Oh, it didn't get the colors, right? But yeah, it lets you add some. There is a slide bar that gets added as you add more reps, which again, for what you needed in step one, it is completely a workable solution. Exactly. And so I did this A-B tested, if you will. I did the same prompt in Lovable as I did in Replit. It was a relatively similar V0 that came up in both. My advanced decision-making process was that one of these talented technologists that I've known in my career is a director of engineering at Replit. And I decided it might be valuable to be able to phone a friend in one of these companies.

10:31So I stayed with Replit and that's where I built 0.1, 0.2, et cetera. My history doesn't go back that far, but I do think there's some interesting things to see. You can, if you're watching this and if you're looking at the experience, the interface for the app that you see in the preview panel is what the app looked like when I first launched it in the app store. So let's call it version one. It had a dark mode as the primary. It has a circular interface where when you add the reps, it fills it out around this sort of in a clock rotation. And it does become a gradient along the way. But I built it in here.

11:09The top prompt from five months ago, which is as far back as it goes, starts with me saying, is the agent down? Can we resume building? Because this is generally how I approach things with the robots. My preview panel went away, but perhaps it will come back. Any questions? So far, I can talk you through a couple more examples from here if it's useful. Yeah. So before you do that, I just want to say the obvious for those of you who didn't get it. Like all it requires is talking in English and explaining what you wanted to do. And when you're not sure what to do, you do exactly what Bryce said.

11:39I said, oh, is this done? Like, how do we continue where we start? And he will tell you how to continue or what's going on or what's the next thing to work on and all these kind of things. It is most of these work in a very similar way. There is a chat interface where it's just like all the AI chats that we know. there's a preview pane that shows you the code as it's running. So what it's going to look like when you actually deploy it. And then the slightly more advanced one, like Replit, also has files and folders on the side where you can upload documents and see the code, which is completely not necessary if you don't want to.

12:12But if you do, then it's helpful. One big unlock, which I really think actually leaned into my posture that a beginner's mindset can be a competitive advantage when building with AI was learning the concept of plan mode, which in Replit is this small checkbox down at the bottom of the screen, right in the chat interface, where it took trial and error to be like, oh, wait, I didn't actually want you to make that change. But when you click the plan mode, I think it's cheaper on tokens too. You can zoom out or press pause on development. And my approach is like, hey, robot, what are we gonna, this is what I think I wanna do.

12:49and talk about it. I'll often ask the robot to ask me, here's what I told you. What else do you want to know? What questions do you have for me? What am I missing? When it will tell me its strategy, I may ask it to evaluate its level of confidence in that strategy. What other strategies it might have considered, not knowing any of the pros and cons of these strategies myself, mind you, just putting it out there, asking it, are there risks to this strategy or vulnerabilities that might be created? Again, no horses in these races, no knowledge about it, but like kind of asking itself to pressure test a little bit.

13:28And so the plan mode can be really nice as like a de-risking a low risk situation even further. I love it. I want to say a few things about what you said. One, as far as the plan mode, what it basically does, it allows you to A, brainstorm, and B, plan. So the brainstorming part is, okay, here's an idea. How do we develop it? What are the pros and cons? But the design is, okay, if we're going to do this, what do we actually need? Oh, we will need three new pages for the application. One will be this. Excuse me. The second will be this. The third will be that. We need these things in the backend.

14:01It needs to connect to this kind of API, blah, blah, blah, blah, blah. And again, when you don't know, like, okay, what is this API going to do? What is an API? Like, why do we need it? So you can go back and forth. So this is one thing about the plan mode. And then once you're ready, you get it out of plan mode and then it can actually write code and make changes to the application. So this is one and it's available on all these tools in a very similar concept. The second thing when it comes to planning, the way I now develop really complex applications is I actually start in ChatGPT or Claude in my case to do all the brainstorming and planning.

14:34And I actually took the concept that Bryce mentioned of ask me questions to the next level. So I actually have what I call a PRD process. PRD stands for product requirements document, where the AI will ask me, I don't know, 100 questions on who's the target audience, what technology it's going to be running on, why do we need this? What does it need to connect to? Like a lot of questions. I answer them to the best of my knowledge. And sometimes I say, I don't know, go and research this and come back with three options and then we can figure it out together. And then at the end of that, it generates a requirements document that is between 20 to 70 pages long that now has a very clear definition of the broader vision of what we want to develop.

15:13Then I go to the next step and say, okay, now let's develop a development plan. Kind of like what Bryce is saying she's done on a small footsteps. You can now do big, like plan the whole thing out. And only then I would go into a replete, lovable, cursor, cloud code, whatever the case may be to actually do the development because I have a much broader are planning, but the concept is the same concept. Allow AI to interview you, to ask you questions about what you're trying to develop. Question AI's decisions. It's like, oh, why are we doing this? Give me other options. Between one and five, how would you rank this compared to the other one?

15:46All these kinds of things. So even if you don't know anything about this, you still have judgment. And once you get the information, your judgment kicks in and you can make better decisions, helping the AI guide the overall process in the right direction. So awesome information. Thank you. Okay. So, so now you had, you said, let's go through a few of these problems. Let's just see some of it. So again, people see that just a conversation in English. Just a conversation in English. I had just told it like, if something seems wrong with you, friend, like, are you back online? Can we get back to work?

16:19And it's figuring things out. And then it says we can get back to work. And I just start nitpicking. We can remove the words like and dislike from the upvote downvote buttons. I was letting people rank if they liked or vote if they liked an exercise. Simple fix. Thank you, Replit. Done. DDD. And then I said the thumb looks weird when filled in. What are our other options here? This might have been in plan mode. It looks like it gives me some options. I said, great, let's go with option one. it's just it's to Isar's point I don't need to know anything about the how I can just talk about my wishes and hopes and dreams I a guest on a podcast I listened to describe I think he was also non-technical described his role as just being the pickiest customer and so sometimes I think of myself as that it's like no I just don't like that like let's try something new let's fiddle with it until we find something that's more palatable or delightful for me.

17:18And you can just, I had this whole thing with these two-sided exercises, think about like a lunge or a bicycle crunch where you're not sure does one side count as one, or do you need one of each side to count as one? And what were the semantics going to be about how to tell people they had to do a hundred on both sides? So I could fiddle with that however I wanted to. There was a problem. I had a, I was playing around with metrics and I was realizing that at some point, I think it was probably because I have a daily double, meaning you can add a second exercise if you want to be an overachiever.

17:46And the metric was showing someone having done eight out of seven days, which I'm like, well, that doesn't work. So that's a good example. That's a super overachiever. It's a super, exactly. You found extra time in your week, but it was a good example of like, you know, it's like kind of the robots are stupid. Like sometimes you can't assume judgment or that like they have common sense for the things humans think about. But as always, it's just natural language interactions. Awesome. I want to ask a few, two interesting follow-up questions. One is how long from the moment you started developing it to the point you're like, oh, I can actually share this with other people, not still on the app store, but just like you felt comfortable that this is a real app that people can actually use.

18:29That's question number one. And I'll hint on question number two, but we can go to it afterwards, is when did you start adding all the cool bells and whistles of like videos and animals and stuff like that? But let's start with the first question of how long when you said, oh, this I can actually give to other people? Or maybe it was the other way around and people saw you using it and asked you for like, I don't know, what was the process? So that's a great question. It was pretty quick. It was pretty astounding. And again, I've built my career working with technologists. And so part of me who felt like a child was show and tell them, like, look at what I did.

18:59I built a technology product. And I did integrate Twilio to set up like SMS notifications. And I mean, that was a whole Everest. So there were a lot of fun of friends in that process. So there was something necessity of sharing. I also, like I said, I'm a mom, my kids are a little bit older now, but like the journey of fitness and feeling strong again has been a thing. And I I've gotten questions about getting back in shape and now my answer is micro workouts. And so it was both, it was both me showing off the app and then also people asking about my exercise regimen and pointing them to it. So I think the chat it showed on lovable was October 21st of 2025.

19:38And I think I was probably having people look at it the first or second week of November. So maybe two weeks later. And I was just working on this intermittently. I was pretty quickly, I'm a fairly generalist recruiter, but if I've done anything the most, it's design recruiting. So for better or for worse, I'm ruined with UI and UX. And it really irks me when things are poorly designed. So please, there's lots of notes. There's lots of room still for improvement, but it was pretty quick that I was like, oh, dear, I can't look at this every day. I'm going to need to fine tune this and improve it.

20:09So I started sharing it. I'll tease one thing, which was as I shared it with people who, including my parents and my family, who are less technical than I, and they were getting URLs like.replit,.app. You know, I'm like, what do you mean? I can't find this in the app store. That sort of begged the question of when was it going to be a real app and how could I make it something that was more shareable? But it was also that audience where you can, oftentimes I was truly waking up to text messages from my parents saying, what's a bridge walkout? How do you do the bridge walkout? And bleary eyed being like, get into a glute bridge.

20:43And what are you doing with your body? How I describe this, this is not going to be the limiting factor of the app. And so I figured there had to be something that was more scalable to show my handful of users what to do every day. So I wasn't having to write out the instructions. And so that was what prompted building more things and adding some videos in. Did I miss anything in those questions? No, this was great. So I think the two critical things are that the road from having nothing to having something you can actually give to other people is very, very short. So this is not your full-time job.

21:17Like I said, you're a mom and work at the tech space in San Francisco, which means there is no clear hours of the day where the work ends. And yet in that timeframe between all the other things, just by chatting back and forth every now and then, you were able to get it to a point where other people can start using it. You can continuously improve it over time. And it's literally endless. Every time somebody else gives you an idea or every time you see something that you're like, oh, this could fit well. So every time you say, oh, I wish I had this and that, all you have to do is go back into whichever platform you're on, explain in English what you want the improvement or change to be.

21:55Look at the results, make whatever fine tuning, and then ship it off to wherever it needs to get shipped. Which leads me to the next question. How do people get access to this thing, right? So you work on it on Replit or Lovable, and now you want to give it to your parents or your friends. And before we even get to the App Store, how do you actually, quote unquote, deploy this so other people can benefit from the application you created? Replit made that so easy. There truly is just a publish in the upper right that, I mean, it uses a lengthy URL. Well, iPhones, non-iPhones allow you to save a web app instance directly to the home screen, which again was like a thing that I had to overcome and make an onboarding screen for to just make it more usable.

22:40But to me, again, like having grown up in technology and, you know, web one and web two, it really felt like the true accomplishment was getting this into the app store and just saying, yes, it is a true, it is a true app. So you could share the URL. Again, Replit made it really easy to just have it be a publicly accessible thing. It was a self-imposed milestone to get it published in the app store. I've not put it in Google Play yet, mostly because I don't have a testing device and nor the time, but it is in the app store. And that was just something that I felt would be a crowning achievement for my own whole fame.

23:13So there are many companies who spend many, many days figuring out how to get their stuff into the app store. How, as a non-techie busy mom, did you figure out how to go from, oh, this is running on a URL on Replit to I have an app running approved in the Apple App Store? I'm so glad you asked. I told Claude. I have a fairly advanced app that's built on Replit, and I'm interested in turning it into a real app that's ready for the Apple App Store. How can we go about maturing the app for submission? And it asked me, here are the key steps to mature your Replit app into an App Store ready iOS app.

23:52It says, choose your tech stack path, set up Apple developer infrastructure, harden the app itself, meet App Store review guidelines, UI UX polish, backend and data. It's a nine point test. And it says, what's your app build with? React, Python, Swift, et cetera. Knowing your current tech stack will let me give you a much more concrete migration path and tool recommendations. And my response was, I don't know. What can I send you from Replit to figure this out? This is brilliant. I still don't really know. So that's, again, a question. It's like every time the robots tell me to do something and I don't know exactly what it is, I just say you're going to have to explain that differently or try again or absolutely do not tell me two things to do at the same time or in sequence.

24:38We do one thing at a time because that's all I can handle. But I think that's the willingness to just truly ask dumb questions or just ask and clarify. It will do better. And you needn't be embarrassed because it doesn't judge you. It's like, oh, I get it. Let me do a better job. That was on me. They make it so friendly. I love this so much, like from so many different aspects. I think the one thing that I will add to this, that now all these tools have either built in or with a simple few clicks that it can walk you through, they have browser access, right? So back in the day, three months ago, you had to go to Replit and figure this out on your own.