Build An App with a Backend Using Ai in 45 min (Cursor Ai, Replit, Firebase, Wispr Flow)

We're going to be using Cursor Composer, which is the greatest feature, I think in the history of the world. And I'm not exaggerating.

You can actually make really good looking, really useful software that actually works, that you could actually give other people access to.

You could just tell it to design. And I'm not a designer. Like, this looks good too. Like this is like a legit, I might make this into a site.

It's wild to see you doing this.

Right.

Hey, welcome to the Next Wave podcast. I'm Matt Wolf. I'm here with Nathan Land and today we've got an absolutely mind blowing episode for you. We're going to build some code on this episode from scratch. We don't even have an idea for a product yet. We're going to come up with the idea for the product, build the product and then actually demo using the product all in the course of this episode. And today we brought on a special guest to help us do that. We brought on Riley Brown who has absolutely exploded on TikTok and Instagram and all the social media platforms with his training on how to code using AI. So he's going to break down his methods for us and he's even going to give us his templates. So we're not starting from scratch when building these apps. So if you want the templates that he's already done most of the hard work for you, those will be linked up in the description, wherever you're watching or listening to this video. But it is a really, really cool process that we can now create something that used to take two weeks and do it in a matter of 20 or 30 minutes. And we're going to show you that we can do that right on this episode with Riley Brown. So let's dig in. Today's guest is Riley Brown and Riley didn't know how to write any code, went down the rabbit hole of developing software using AI and today he's going to break down some of those methods with us on screen and we're going to build an app completely from scratch by the end of this episode. So welcome to the show, Riley.

Super, super hyped. Yeah, I love building software with AI. When you're first getting started, I would just use cursor. I, I, I use replit because it's very easy to deploy. But I would recommend starting with cursor and just sticking with that until there's like a clear narrative change, that there's a better tool. I like, in my opinion, like cursor. I tried GitHub copilot and I regretted even trying because cursor's like that much better. And so like, if, if you actually want to get to the point where you're creating stuff, you need to go deep before you like try a bunch of tools or else you're just going to get lost in like tutorial hell. And yeah, that's, that's kind of the way I see it now.

Don't cursor and repl it work together? Can't you like write something in cursor and have it pushed to repl it or something?

Yeah. Okay, so let's just let me talk about the template and I think this is a good intro to the what we're going to build. So this is just a code base and it is set up via ssh. I'm not going to lie. I don't know what SSH stands for, but it is a setting on replit that you, you basically just go up to SSH and then you basically go to cursor and you generate an SSH key which basically connects the code bases so they mirror each other. Any change I make in repl it will mirror to cursor and vice versa. And so when you're beginning to code, like when you start any project, the beginning is the most annoying part, which is why projects took two hours. My now business partner Anj and I basically went through and we did the plumbing, that's what he calls it, where we did all the annoying dirty work at the beginning. To give you the base template. This is the base template that before we started this episode, it takes about to a new person just starting, maybe 15 minutes to set up. For me it takes like two or three minutes to just like quickly go through and like set it up. And so this is like the base app where you basically just have simple sign in features. It's cooked up to Firebase so you can sign in so you guys could sign into this app. You would have your own profile and you can, it's just like a very simple like social media app just to make sure that like your database is working and you can just like post and this way it gives the AI kind of context. Be okay, this is what you're starting out with now. Your first prompt starts out with like, okay, you're starting out with this template. Now let's create basically whatever you want to create. And so that gives us just a good starting point. And we're going to be using Cursor Composer, which is the greatest feature I think in the history of the world, and I'm not exaggerating.

Which is this place where you get to describe what you want Cursor to code. And then when we hit save, we will see the changes as soon as we hit save in the repl or the, the web view on replit.

And so, yeah, and so Cursor just. You could correct me if I'm wrong.

You've.

You've used it a lot more than I have. But basically, if I'm trying to write code directly inside of something like Claude, it's basically looking at like the one file I'm trying to fix every time. Right. Like, let's say I'm just trying to fix the index dot HTML file or whatever. I can paste that code into Anthropic, tell it to make some changes for me. It will fix just that code. Right. But what ends up happening as you build out these files is you start to get all these JavaScript files and you start to get all these HTML files and CSS files and all these different files to make the site work together. Right. And after a while, Claude struggles to pull in the context from every single file that you're building. And from my understanding, Cursor makes it so that all of the files that are in Cursor, the AI is seeing with every single update. So it, it knows the entire code base every time it makes a tweak for you.

Yes.

Where Anthropic, it will sort of lose the thread after a little while.

Yeah. You know, Claude projects you can actually upload all of your code base. I used to do this at the beginning. I would upload the entire replit code base, plug it into Claude and then ask it to generate code. And it will generate multiple files at once. Right. It'll generate the full, the code. But what Cursor did, and like the genius of Cursor, is the fact that they have this apply feature, which is actually a separate step. Not only does it generate the code, it then it has your existing code base and it has the code that Claude generates. And then there's this apply step where they use an AI model to take the instructions or the new code and then apply that code to your existing code base. It looks for the difference and it generates new files based on that. So it's not, it's not doing it in one step, it's actually two. And that, I didn't realize that was as hard of a step as, as it was because it's not deterministic. It's like it's an Actual train. They train their own AI model to, like, apply it. And that's where the magic sauce is from, from what I can tell, at least.

Right, right. But if you upload like all of the files initially into Claude, CLAUDE is just basically seeing the starting point as the context. Anytime you prompt something. Right. So, like, as stuff gets updated, you either have to go and like, remove the old file from the context and update it with the new one, or it's going to be constantly sort of looking at the old code base, I would imagine.

Super annoying process. That's mostly it. Yeah, super annoying. Highly recommend using Cursor.

Let's get into it. I think, I think we need to, like, come up with an idea first.

What can we do? What could we do? We could do something fun. We could do, like, what about a daily AI tool.

Skill tracker where you put in. You have a user and you put in the skill that you learned with the tool that you used, and then you can upload like a link to a video that you created it or something like that. Like, you can track your skills or.

Like, like a knowledge management system kind of thing.

Yeah.

And it could, you could have, like.

Be held accountable to, like, learning a new skill. Let's. Let's just. Let's just try it. Let's try it.

Okay, so I like that idea.

Yeah.

Yeah. So let's say, let's say an hour a day of learning a new AI tool. I think that's manageable for people. They can squeeze. And it could, you could have, like.

Be held accountable to, like, learning a new skill. Let's. Let's just, let's just try it. Let's try it.

Okay, so I like that idea.

Yeah.

Yeah. So let's say, let's say an hour a day of learning a new AI tool. I think that's manageable for people. They can squeeze it in there. And then at the top we can have, like favorite tools, maybe a few links to tools or something like that. And maybe it's social, like you can see other people's status and then you guys can sign in to this app once we finish. Um, so. Yeah, let's do it. So cool. Okay, so I want to create an app that allows me to track my AI tool usage and skill development, mostly skill development. And I want to be able to every day log what tool I used and then what specific thing I created with that tool. And so over time I can track how I got better at learning with AI tools. And so I want to be able to sign in this is a template. And we've already set up Google Firebase in the template, so make sure to use that within your code generation. And I want you to allow me to obviously sign in and create a profile automatically and log that every single day, come up with a unique way to show that information visually. Like each day the progress, maybe a month view. So you can switch between the months and see which days I succeeded and which days I failed.

Okay. And then when in doubt, type at @Code Base. This process is a lot better if you know what files you want to change. It's faster and it's more precise. But like when you're just getting started, just make sure to just do ODE base, which means it'll look at the entire code base instead of the file that we're on currently. So we're just going to run this and see what happens. We could get an error, but let's.

See, look, we've all been there buried in data, trying to figure out what it all means for your next business move. Well, AI can make that process a lot less painful.

I know.

I've used AI to analyze YouTube video performance and optimize for growth. Well, to help you out, I've got a list here of more than 30 AI tools designed for data analysis. I actually got this list from HubSpot's how to Use AI for Data Analysis guide. It's going to teach you not only how to integrate AI into your data pulling process, but also the benefits of it. You can save yourself so much time. Plus there's that really cool list of AI tools that I mentioned. This free guide will completely change how you use data and make your life easier. So if, if you want it, you can grab it at the link in the description below, come back to the show. I like that. You could just sort of kind of ramble into the microphone with just the sort of random thoughts and ideas you have about the app, but it just kind of collects them all for you.

Yes, yes, I, I operate the best with voice for sure and it makes it way more fun. And so you can see here that it's making changes to one, two and three different files and we just basically wait for it to finish. And you'll notice as I was talking about earlier, it's generating all these. And watch this right here, we're going to see it apply once they're all done generating. See it says applying right there and it's generating, applying, done, generating. So this is the fifth file it's working on applying and it's just Generating.

All of these new files too. You can see over on the left sidebar as new files are created.

Yes, I think app TSX was, it edited that file, but the rest I believe. Oh, and index TS was like part of the template, but the rest are generated. And so we can hit save all and don't be alarmed if there's an error. Nope, there we go. It created this right here. So we have this progress tracker. We're getting some errors, which is fine. Okay. This error is a little bit weird. I don't even really want to explain it because I'm not gonna lie, I don't fully understand it, but I know what to do when I get it.

How did you know that there was an error? I noticed that you opened up the inspector and then you saw the error in the inspector, but how did you know it was throwing up an error before opening the inspector?

Thank you. Thank you. I, I, I went too fast there. You're right, you're right. Okay, so when you're on replit, and so this is replit, this is cursor, and I like to keep them just, it's fun to have them side by side, but when I'm working on one monitor, it's easier to just like switch back and forth. Right. And whenever you have the web view open, which is like what you're working on being rendered in real time, right. You have this little dev tools up here. I can zoom in a little bit as well. Maybe we'll go in one more. You have this dev tools and it pops up here and you just see that it has like these errors within here. You can actually just like copy this whole thing. Oh, yeah. Oops. In this specific case, what you need to do is you actually need to open up a new tab and then you need to go out here and you need to hit inspect and then you need to go to Console, because Firebase actually sets this up for you. It says, aaron, Snapchat requires an index. So if you just click on this and sign in with the same email you use to sign sign up for Firebase, it will actually build the index for you right there. And this is just a, this is just a fast way for the database to find the information that it needs to display, basically from what I understand what an index is. And so that actually takes a few minutes to build. But we're just going to keep going. So let's see what it created. We're going to go back to this app right here. So as you can see, after this, it created we can go to the next month and we're still getting that index error, I believe, so we'll just wait for that to finish. But yeah, so we can go to the next months and previous months, recent entries. I don't see. Oh, log AI tool usage. Okay, so we can type in Claude. What did I create? Actually, let's do cursor. What did I create? I created a web app that logs AI tool usage and we can. Ooh, one to five. I like that. I didn't even ask for that. It just added the. How confident were you? One to five. I like it. All right, we'll keep it.

I guess over time you can sort of track your confidence levels on a specific topic. That's really cool.

I love.

I mean, all with one prompt. That's cool.

Yes.

It's actually kind of cool. Like this, a real thing where you're. You're actually just showing experiments of things you created using AI would actually be kind of a cool. A cool site.

Yeah, just keep it. A log of it all. Yeah, already. Already pivoting our first time.

No, that's actually like a wheel. A real cool website that you could actually log.

That's a legit, simple, actionable, like, cool website. I agree. Let's see what happens when we log the entry. Okay. When we log the entry, what Error. So it said it requires an index. So the index, for whatever reason on Firebase takes a few minutes to build. Let's just come back and we'll see if that fail. Let's make another change here. Let's. This is a really fun one also for those of you who are doing this, avoid. This is just a very important thing. After each one, just hit Save all. Hit Save all. Not Accept all. Accept all basically sets whatever you're doing in stone where you. When you hit save all, you're basically like saving the changes that cursors made, but you're not, like, committing to them. And so you. The last thing you want is to, like, commit to code that actually broke your app.

Or that when you say commit, are you talking about, like, get committing or like, what do you.

What does that mean? Very similar. So kind of I will show you how to git commit. So I like these changes, so I don't want to change them. So now I'm going to press Accept. Accept all. When you're using ssh, I use the git features within replit and they're the best. I love it. All you have to do is hit Initialize Git repository and we're Done. So it's. It shows that we did an initial commit and then after we make another change, it'll show us all the changes and we can basically just put whatever we did in that section of changes and it will save it right there. Replit allows you to, like, go back and load any of them with like one click, which is so many people ask me, like, why do you use replit? Why don't you just do local hosts? Because it does all of this little stuff as well as deploying it on the Internet. Like, I can deploy this and upload my own custom domain immediately. So that's why I use replit. But let's go ahead and let's go back. Let's take a look and see what this looks. Right. This is boring, right? It's kind of lame. Um, you know. Oh, okay, so this worked. So this all worked. This logged in the database after we're still on prompt number one. So if we were to like go to Firebase and we would go to Firestore database and hit data, we can see the AI entries were entered into the database, which is pretty cool. After one single prompt. Oh, wait, did we see it on the calendar? Did I put. Oh, invalid date. Did I put a date in? Let me see if I. Oh, there's no option to put a date in. So that's what we should fix. Right? So it said invalid date.

So maybe I'm at like pre select pre select today's date and if you don't, and then allow you to change the date, but by default it's today's date kind of thing.

I'm just gonna mirror exactly what you said into the microphone. Okay. Okay, so what I need you to do is I need you to allow me to add a date when I'm adding an AI tool usage. By default it should be today's date and then it should allow me to change it. And I also want you to. I want you to think as a designer here and I want you to make the site. I don't want you to go crazy with super loud colors, but I want you to make something that looks clean. Act as a designer. I don't want you to change the functionality. I like the functionality so far. Just make it look a little bit cooler. Right now it's white. It's boring. I don't like it. Just bring out the sauce a little bit.

And we'll see how it does with that.

And. Yeah. Okay. Did it show us the.

Web logs? AI tool usage confidence. 4 out of 5. What? I just Realized, you know, in Excel how you can do the, the heat maps where it's like brighter green or red based on the, like, you could do a confidence level. We can do that same thing. Um, that'd be really cool.

So while we're waiting on this, the tools that you're using, right, I just want to kind of go over them real quick. So you're using Cursor as the main sort of area where the code's being written. All of your files are organized. Firebase is like the database that is storing the records that you're creating, essentially. And it also, I believe, created the login. Right? It created the sort of Google login to be able to log in. And then Replit itself is where the sort of front end is being hosted. Right. So Firebase is the backend database. Replit is basically storing the front end online so people can view it. Right? That's.

It's storing all the code. Like you can have a backend or front end. It normally you would have to host it locally. And I found Replit's just a lot easier to deploy and also do git, so it's a lot easier to save your code because like Replit and Cursor literally have the same exact files. It's just Cursor right now is better at generating code. It is the best in the world. And so that's why I'm using Cursor. If I didn't have to use Cursor, I wouldn't. But, like, just right now, it's the best. And I'm. I, I like to use the best tools for things. Like, I want to use the best tool that makes it the easiest for me to deploy. I. It's really hard to deploy apps on other apps. I've tried a lot of them and so that's why we use Cursor. But yeah, you've all the other things you said. You nailed it. Okay, let's try this. So as soon as we hit save, all this is what it looks like. Now as soon as I hit save, it's going to up, it's going to update. So three, two, one. Boom. Let's see. Okay, you see the. Okay, so it changed this to like these dots here. It added like these clean bubbles. Let's say, like.

Can you add a little bit of color and make the background of the entire site a nice clean gradient? Like have it be like a, A, a dark gray to really dark gray gradient. And I want you to keep the cards the, the calendar view that you just created. The same. I just want to change the background behind it. Let's also check before we make any changes here. Let's go ahead and let's try and add another one. Okay. So by default it has the date. So Matt, your change is there. We could say mid journey. What did we create? New editing feature. I don't know. New editing feature. And we could put confidence level and it was kind of shaky.

Can you change the date as well?

We could have changed the date. It's still set. Invalid date, which is a little bit weird. But we'll figure that, we'll figure that out. It's good to go to get some problems in there. Maybe let's try doing tomorrow. And then let's put chat. Chat GPT. Let's just put test log entry. Okay. We're getting an invalid date no matter what. And so I'm actually going to hold off on more styling because we got to solve some of this stuff for whatever reason. It says invalid date no matter what date I put in there. Uh, can you please fix that? And before you fix it, just give me a few sentences on why you think that's happening.

It's wild to see you doing this, Riley. Because, I mean I've, I've, I've done like several tech startups where you're like, you'll work with engineers and you'll like give them tasks like this and then you get the result like the next day, right?

Yeah, yeah, yeah.

Versus you're just sitting there talking to it by yourself. That's amazing.

Well, it's super smart the way you approach it too, where like pretty much whenever you have it fix stuff, you're trying to learn yourself how to prevent that in the future so you can speed up your next project.

Yeah, like people have seen me do this and they're like, oh, does that increase its pro, like increase your. Its successfulness? I'm like, no, I just do it cuz like I'm curious on what's like on what happened. And it is, it's like that's if you just want to shut your brain off. Like I love learning about what's going wrong. And so if you like fall in love with that process of like figuring it out, you're going to get better so much faster the more like the more projects you take on. So the issue is likely happening because the mismatch between handling dates and firebase in our components. In the ad entry component, we're using timestamp, front end. Okay. Okay, this makes sense to me. I think this Will work. There we go. Okay. Okay. Now what we can do, actually. Ooh. I actually now that there's some color on here, I don't know if the background needs to be changed, because.

I like that. So now what I want you to do is when I hover my mouse over the day, I want it to show all of the entries for that day in a super clean, organized pop up that should just pop up while I'm hovering over them and show me all the entries for that specific day.

When you're talking with this, this tool, what was the tool called again? The one that you're speaking with?

Whisperer Flow.

So are you like pressing and holding a button and then when you release the button, it knows you're done?

Correct? Yes. It's good. It's really good. I've tested all of them because I'm a huge voice. I'm a huge yapper.

Well, I was just curious because if you use something like the new ChatGPT voice assistant and you pause for a second, it's going to interrupt you and try to, you know, talk over you for a moment. But I noticed this one isn't doing that. So I was wondering if you were like holding down a button and then it knew you were done when you release, which makes sense.

I use the old version of the voice model all the time because I like to be able to just. Or actually they. They stop letting me do it where I can pause. Like the, the old interface for ChatGPT's voice assistant was perfect. Like you could just hold it down for as long as you want. Perplexities is really good too. Yeah, I don't like that it interrupts me. I don't. It's good sometimes, but it's. It's a double edged sword, in my opinion. All right, let's hit save. Let's take a look at this. I forget what we changed. Oh, yeah. Boom. Look at that. Now we have this pop up. We're like, that's cool.

That looks clean too.

It actually looks pretty good.

What else could we do?

Are we gonna make it social at all or no?

Yeah, I mean, okay. Yeah, I guess. Okay, we can make it social.

Maybe. Maybe a feature to edit recent entries. Can you edit one of those entries once it's there?

Great. I like that.

You also could post when you recorded one or something like that. Could like post on X or something, share it.

That would take a little bit longer and I don't want to deal with the X API.

It would be all the APIs you can deal with X will probably be the worst API you could possibly deal with because.

Yeah. And I. So like, in the app that I'm. We're. We're building, Matt, I think it's. I think it would be interesting to talk you. You're a huge fan of my mind, right, Matt? Like the app, my mind.

Mm.

We're building. We're building the basically voice notes plus my mind. Like we wanna build both of those into one. And so like whenever you create a voice note, it takes all the things that you saved and matches it with what your voice note is about. Because I'm a huge fan of my mind. Except they don't allow any social features, which is one of the things that annoys me. But I think. Okay, what was the idea? The idea was to. Oh, edit.

Yeah. Let's start with just being able to edit one of the entries once it's there. You know, maybe you accidentally put the wrong date or you put a typo or something like that. You wanna go back and it.

Okay, I like it. Okay. I want to be able to edit.

Actually. I want to clear cursor composer. Just like. Just like Claude. You don't want to let the context get too big and so you want to reset the composer. And so let's go ahead. This is what it looks like when you save. So I'm just going to put like added.

Pop up when hover. And. And what else did we do on this?

Good.

We did fixed dates issue, slight styling.

Yep.

Yeah. And so we can just hit stage and commit all changes and we're done at any moment. We can like click on this initial commit and then it gives you the option to like go back to that version. And so if you just do a good job saving, you can always go back, which you will do, not get stuck in a death loop. I think, Matt, we were talking about how you got stuck in a. An error loop of hell, which is not a good place to be.

Which seems to be more common if you're just trying to code straight inside of ChatGPT or Anthropic, one of those tools. Because what can end up happening is. So basically you write code and you commit the code and then it starts working and then you make one little change and. And then you save the. Let's say I'm just like storing this locally. We're not using replit, we're not committing any past saves, anything like that. You can change the code, save it on your local computer, run it, it all works. You make one little tweak. Run it again. It doesn't work anymore. But you can't figure out which exact tweak that was made is what broke it. And you have no way of going back to like the previous version before it was broken. So what you're explaining there with the commits is like insurance for that happening. You break one little thing, you can't figure out what you broke. Okay, Go back to your previous save and move on from there.

It's exactly like Super Mario. Like you want checkpoints. It would be annoying if you were on level 30 and you died and you had to go back to level six or like. Yeah, it's, it's just, it's like saving and you want to do it as frequently as possible. Okay, so let's see what happens here. So we now.

Oh, wow. Okay, you see that? It put the edit on here. I was expecting for it to get put down here. I'm just gonna ask for it.

Can you please put them on the entries that are down below the calendar instead of on the calendar directly? Actually you can keep the ones that are on the calendar pop ups, that's okay. But add them to the entries that are down below the calendar and instead of making the entry of a new AI skill on a separate page, please make it a pop up center peak from the main page and it should be right above the calendar to add it. I don't want to have to go to a new page to do that. Yeah, because I can't even get to it because it like. Oh, there we go.

Oh yeah. You just got to kind of keep the mouse in the exact right sort of path.

And we're getting a ton of errors. That's fine. We're gonna, we can get out of this. It's not difficult. This is a lot of changes actually. Like we'll see if a, if AI gets this. This will be pretty impressive. We're adding, we're adding an edit feature and then we're moving the, the entry feature and actually it shouldn't be too bad, but we'll see.

Now I don't know if you're the best person to ask about this or not, but out of curiosity, do you know how long what we have here would have taken traditionally, like without AI?

I don't, I have no. Yeah, I have no idea. I, I like people. People give me a bunch of different numbers.

Because like the really good developers that I talk to who are like excited for me, like are all use AI and they code 10 times faster than I do now because they just like go so much faster because they're not just using Composer, they're using the chat feature. So they're going in and asking AI what to change. They're just doing it right there. Composer takes longer. And so like when you, you, you start to, you have to watch a really good developer code with AI and you realize like the added benefit and efficiency that they're getting out of it. It's insane. It's actually nuts.

If they know all the jargon and technical terms, you're going to be able to better steer the AI.

Yes. Okay, this is a really good example. So we just got a full blown error and this is our first like big error, I would say. And so like I said, we come up to dev tools. I'm just going to press command a, copy the whole thing. We're going to go back to repl it and I'm just going to say fix this now. Row.

Does that help? Saying calling it a bro.

No calling it bro. You get more. You get efficiency gains.

Doing it for the. Doing it for the entertainment. Let's see here. Okay, so let's refresh. Um, so the edit feature was the first feature that it like struggled with. Um, we can actually go. This is working now. I think I was just. Okay, so that just works. We're able to get to it.

I don't see where the edit is though. Is it popping up in a new tab or something?

We can't. That does not work. I was just saying the pop up was. It was letting me select it. That was what was working. Let's see if we actually get the option now. Let's see if it fixed that error. There we go. Okay, cursor bro. Let's save this. So now.

Cursor bro. So the edit worked. It works. And I changed this. I didn't test this last time to be like a center peak. So we'll see if this works. There we go. I didn't like the fact that it was like off. Oh, I have another good feature. If I double click on a specific date, I want to be able to add a skill that I learned on that day. So you basically act as if I were to press new skill learned except have that date pre filled as the date and I'll add in some of that information and it will be on that specific date.

Also. Also I want to have another page that logs all of the tools in graph in table format along with all of the days where we use that tool and other relevant stats.

We're just having fun now. All Right. So let's do this. Hit save. I don't even remember what we added. Oh, yeah. Ah. So it added this table down here. Look at that. Look at these tool statistics. It already added this. Average confidence. First use, last use.

Oh, that's so.

I mean, are you kidding me? Like, this looks good too. Like, this is like a legit. I might make this into a site. I'm not even. I'm not gonna lie.

If you click into one of them. If you click into, like, one of the. So go back to the last screen that you were just on.

Yeah, yeah, yeah. And I. I'm not gonna make this. This bottom bar. We're gonna change that next.

What would be cool is if you clicked into one of them, it shows you all of the dates you practiced it or something like that, you know? So, like, if you clicked on the cursor, bro, it would say, well, you played with this on this date, this date, this date, and this date. Right. So it's sort of like a log of how many times you've used that specific tool.

Can you please. In that table, when I click on any given row, it should be a dropdown and it should show me each individual day that I use that tool and then show me the entry for that day.

Something like that.

Yeah. And what's really cool about. About this too, is because you're building this with Firebase, every person who goes and uses this tool, like, let's assume we made this public and other people can go use it. Everybody who uses it, they see their own individual database. Like a. Like a real legit SaaS. Right. It's not like they're seeing all of the tools that we just added in. Everybody gets their own database, essentially.

Yes. And like, yeah, you could see your own, and then there could be like a public calendar. Like, every time you log it, it logs it into a feedback. So, like, there's like a feed, and then you have your profile, which is this right here. Progress tacker shows all your tools and then you can make it, like, not designed for anything you would want to keep private at. The purpose of it is like a public profile.

Right.

And like, it's like, group can learn and then you can log what other people. Like, this is actually good. Yeah.

Should we show, like, an example of what you did there? Like if you created something, should we do like a screenshot or something? Because it's kind of more tangible versus just being text that I. Oh, just wrote.

Whatever. Okay, let's. Let's do this one. Okay. Oh, I Thought I entered this one also for each entry. I also want to be able to upload an image. I want to be able to upload an image and make sure to look at the full code base because we've already set it up to Firebase, so you should store those images in the storage and.

And it should be visible in the popup that when you hover over each day, if you upload a screenshot.

When you. When you start dealing with images, it gets a little bit more difficult. But, like, I. I think it should be fine. This is like, unironically, like, kind of a cool tool, I'm not gonna lie. Yeah. Just to, like, see some.

Some other places where you could take it. Right. Like, you can have like a public feed and start seeing other tools. And then it might be a way for people to discover tools they'd never heard of because they're seeing the names of tools in the public feed and.

Yeah, yeah, yeah, yeah, yeah. Like, I. Because, like, social media is. I think it's changing a lot. It's like becoming more community oriented. And so you might want to be able to just build one of these sites for maybe like maximum 50,000 people, but, like, at any given time only, you know, maybe 500 people might be on it. That's manageable for Firebase. It shouldn't be too expensive. It's when you start getting up to, like, hundreds of thousands of users when it starts getting expensive. But this is pretty. Like, this is. This is doable, actually, I'm like, I'm definitely thinking about it.

Well, once you get up into the hundreds of thousands of users, I mean, you could. It starts to get fairly easy to figure out how to monetize. Right? Like, you could start throwing AdSense ads inside the dashboard and monetize through ads, or you can charge a small fee to use it, or you can have free and paid plans. So, like, to get.

You can migrate off of Firebase as well.

Yeah, yeah, yeah, totally, totally.

Yeah.

Um, okay, so now let's see if it's in there. Okay, you see that? Upload and drag a little file. Let's test this first. So building apps with AI.

Nice.

Okay, so let's.

So relevant, too.

Omg. No, wait, no. This is the tool used. Let's do cursor, bro. We want to stay consistent. And AI app, yo. And then let's make this. Let's make this. Well, I got you. If this succeeds, we can add it right away. Add entry.

I think it worked. Okay, so next month. No, where is it? Where Is it.

Did it actually do anything?

Okay, so this is exactly what you asked for last time. So, like, now these are dropdowns, so you can see it at this view. You can expand it. Now we have this nice editing feature, ChatGPT. Cool. But for whatever reason.

I uploaded one.

Yeah, Yeah.

I uploaded a skill and I put it on a date and it didn't show up on the calendar. And that's pretty annoying. Can you please fix that?

So the other idea that I was thinking is on the add new entry page, if you start typing one of the tools that's already in the database, have it autocomplete that tool. That way your naming conventions stay consistent.

Ooh, we could actually just have. As soon as you click, it'll just give you a dropdown of all of them, which would probably be easier.

So either give a dropdown of them or if it doesn't exist already, it would create the new. It would create a new tool. Right.

So could show you some stats too, about how many times that's been used or something.

Can you.

So many ideas. Can you. Can you please add in the add new skill section? I want to see all of the tools that I've added previously as a dropdown selection. Very simple. Please add that feature.

Okay, so we're going to save the previous changes. I actually don't recommend doing what we're doing. I don't know what. How hard the cutoff is for this podcast, so I'm just doing two changes at once. So what was the change before the last one? It was the image. Okay. The image wasn't working on the calendar. Okay, let's see. Let's go. Preview template. Let's do. Let's see. If it was a problem, it might could have been a problem with the fact that, like, it was a repeat tool, so it was it. That could have been a weird coding thing. So we could just put anthropic test and then let's just add entry. Let's see what happens here. Okay, I think that worked.

Oh, wow.

There we go.

That's cool.

Okay, so this worked. So maybe it's the fact that.

I don't know why this one's not showing up. This one is cursor bro. 1031, 1029. Okay. Cursor bro. Oh, they're both on there. Okay, so like, let's try and do. Oh, and then I added the double tap feature, I'm pretty sure. So we could go November 5th. Double tap. There we go. Nice. Wait, wait, wait. Before we enter that one, let's update the previous changes. Let's see if it. So now we'll do this again. Now it should do a drop down. Oh, no.

Wait, Maybe that's not on. Ooh, ooh, ooh, ooh. Auth is not defined. We got an error. Okay, let's just copy this. I think it'll fix it after this. I'm just gonna say code base.

I think we should be good after this. Honestly, this is actually. This is like. Yeah, one of the.

We could go ahead and call it a tool. I mean, we could continue to build it out after this, do something with it. But yeah, I think. I think we've proven the point on this episode of how quickly and easily you can build a software tool.

And I mean, no, we're not. We're not ending this podcast until we're at 10,000 mrr with this site.

So who owns this? We've got, you know, Riley, Matt, Nathan, HubSpot. Like, I mean, I think HubSpot owns it.

I think it's only fair.

But what's really cool about this is if you're like, trying to grow an email list or something. You know, traditionally you can do stuff like ebooks or videos or something. Well, now this makes it just as easy to like, make software something that you collect emails to grow a list off of. Right.

That's actually how HubSpot grew a lot, actually, was having free tools and stuff. Right? So, like.

Yeah, yeah, yeah, 100%, yeah. And just to finish it off, we did get that feature added, Matt. So now it does do the dropdown. You can select them super easy. Yeah, there you go.

Super cool. This, this has been awesome. I'm trying to think if there's, like, any gaps that we need to fill in for the audience that might either be listening or watching. So as far as costs of these tools, I'm pretty sure you can use Replit for free, right? Firebase. I know you can use free up to a point because I've never paid for it. I haven't really used cursor a lot, but when I was using it, I don't believe I was on, like, a paid plan. Are these all free tools or are they like.

No, no, no Cursor. Now you can't use Composer without paying for Pro. It's the easiest $20 you'll ever spend. I use about. I. I spend about $120 a month on cursor credits. So, like, if you use it all the time, you're not. It's going to be more than $20 or else they would be just losing so much money because each query costs a decent amount. Replit is free up to a certain point. And you're right. We. You could do this. You could. I have still not paid for Firebase yet and I've created like a hundred projects probably. Um, and so it's. It's when you start getting a lot of users and there's. It. There's a lot of demand on it. So, yeah, it's pretty cheap to do this, honestly.

Yeah, yeah.

With Firebase is very expensive when you scale, so it is worth knowing that.

Yeah, but that's why I would. If you actually have plans to scale it, honestly worry about that when it comes. Like. That's a good problem to have. Yeah, yeah. But, yeah, Firebase. We decided not to use Firebase for the app that, that I'm building because of the large number problem.

You're expecting scale on it.

Yeah, yeah, we. I mean, that's. That's the goal at least. And so we want to be able to scale it and there's going to be a lot of data and so there's ones that are just faster too. Firebase is the easiest to get started with because it's just so easy to use. The AI really understands it and so I recommend starting with that over Supabase. A lot of people have been saying, use Supabase. I have had major problems with it. I think AI understands Firebase a lot better.

Gotcha.

Cool.

Well, on that note, why don't you tell us a little bit about the platform that you're working on? Or is that something that is still too early? I know you've talked about it a little bit on Twitter and stuff. Is that something you want to talk about?

Yeah. So we released it last Friday. I made one video. It's an app that allows you to record voice notes at any given time. It's called Yap Thread. Yap Y A P T H R E A D. Based on the premise that I love to walk around. I'm a very active person and I don't like to write. And so I wanted an app that I could pull out at any time and just yap all of my thoughts. The difference between this app and a lot of other apps. A lot of other apps, it's just like one voice note and then you can create another one. You create another one, this one. After each voice note, it actually creates a thread. And so then within this thread, you can record as many voice notes as you want, like hours and hours and hours. And then at the end, you can use an, a custom AI prompt to turn it into any outline or any piece of like a format content that you want. And as these AI models get better, we're going to get a lot better at reformatting them into like really clean scripts in the style of your favorite creator or in the style of your previous videos. And then also while you're scrolling social media, you can save like all of the interesting things that you find to the app so that you actually have like ammo or ideas for when you start to write. And it can like, bring in your memory and like, for you, Matt, like some you. You do these like, AI news videos. And so like, you, from what I can tell, you use my mind.

Right.

And so it's a good way to iterate on that. And that's kind of the basic premise of the app.

Yeah, I know that that kind of lines up with Nathan's sort of thinking strategy. Right. Because you said that you like to use the advanced voice assistant and just go for walks and have conversations with it. Right?

Yeah, that's what I've been trying to do. I mean, I'm still, you know, like Riley was saying earlier, like, it interrupts you and there's like, you know, it's still kind of odd to you sometimes, but like, I can see like the potential long term of, of using it for everything I do, really. So I'm, I'm curious to try that out.

I agree with. I have. I had the same exact experience using it where I got annoyed that it, ideally it will remember everything that you've journaled before. And so that's what we're building long term. We're building AI chatbot. It's just not ready in my term. In my sense, it's too expensive to dive into that. And so I think this is a good place to start. But that's very much the goal is to build a chatbot that helps you write and be creative. That's the basic premise.

Is there, is there a cost to.

Use it right now you could use it for free for seven days and then it's 7.99per month. And yeah, we're adding a lot of really cool features that I think it's a lot of value for the cost. So I'll share more about that within the next two weeks. We did a soft launch on Twitter. I'll be doing a full announcement within the next week or so.

Very cool.

But it is on The App Store.

Ios, iOS right now. Are you gonna. You're gonna put it on Android eventually.

Oh, yeah, yeah, we're almost. We're almost done with that. Um, it was kind of a we use react native. I don't want to get into that. But yeah, we. We have it. It'll be on web as well, but it'll be on Android first and then it'll be on web within the next, like, two or three weeks as well. And that will be a bigger app that. It's gonna be fun.

Very cool. Well, this is. This has been amazing. I think we're. We're at a good stopping point on it for. For where we are today. Just to kind of prove the point that you can actually make really good looking, really useful software that actually works, that you could actually give other people access to.

You give me 10 more. 10 more prompts on this site right here, and it'll look completely different and have a couple more features that I think would tighten it up.

Awesome. Well, appreciate it. Thanks again. And if you're listening to this episode, make sure you're following Riley Brown over on X. And I've been saying Twitter, but it's actually X, as Nathan already corrected me. And if you enjoyed this podcast, make sure that you subscribe to us on YouTube for more podcasts like it. Or if you prefer audio, subscribe wherever you listen to podcasts. Thank you so much for tuning in. Really, really appreciate you spending the time with us today. And we'll see you in the next one. Hopefully.

One.