Vibe Coding Your Own Productivity Tools

Hello and welcome to another episode of App Stories. I'm John Voorhees and I've got Federico Vettici with me. Hey, Federico. Hello, hello, hello, John. Hi. uh hey hi i've got a snowman looking over my shoulder you see that you have a snowman it's a holiday spirit right there we did a little decorating of the house over the weekend how about you uh we put our christmas tree up today

same same because we were doing some work in the living room especially for a particular gadget that i will talk about in app stories plus i have a gadget or two to talk about myself and plus because yeah we are you know what we're doing this is like This is like the pre-show to the experiments that we do over the holidays. This is the...

kind of the unwinding research. We are setting the stage for us when we go on break for a few weeks, all the things that we're going to do. We'll talk about that later, but yes. Before that, so I had this fun idea.

We've been talking a bunch about AI this year. We never fully talked about vibe coding, and I kind of wanted to talk about that with you. What is vibe coding? Do we consider ourselves, to an extent, vibe coders? Uh-huh. And I kind of want to talk about everything about it, the silliness of the name, you know, people taking it to an extreme. So I'll just get into that. If that's okay with you. Yeah, no, definitely kick it off because I think it's important with something like this.

especially given the kinds of things I've seen online to kind of contextualize it a little bit and help people understand how we view it as compared to what they might have seen somewhere else online. Yeah, so the whole idea, first of all, the term, vibe coding. So you're coding with vibes. So this originates, if I'm not mistaken, from a post on X by Andrej Karpathy. I think that's true. He's a former, I want to...

say like maybe co-founder or one of the original co-founders of OpenAI like one of the leading AI researchers and one of the you know global experts of LLMs and a few months ago maybe last year, he came up with this idea of like, I like talking to these chatbots and building this like...

personal small projects with code where I'm not really paying attention to the code I'm just going with the vibes and I'm you know sending feedback to the LLM and I test something and it doesn't work and I send another message and I'm just going with pure vibes i don't even look at the code i use the llm to build me simple personal projects and that that's where the the term of vibe coding comes from and it kind of took off from there and what's funny about it is that uh on the

I mean, it sounds ridiculous, right? Vibe coding. Why would you call it vibe coding? But there's that idea of... These LLMs, especially the modern ones, your Claude Sonnet 4.5 or Opus 4.5 or GPT 5.1 or to an extent, some of these Chinese models even, like they are pretty good at writing code.

because they scraped the entire web and GitHub and Stack Overflow for these billions, trillions, I don't know, of lines of code. They're pretty good at spitting out code that works. And they're pretty good for building... personal projects, like personal, like little things like scripts or personal web apps that you can run on your Mac, or I don't know, automations with Python or bash scripts in the terminal, like these things.

These LLMs are pretty good at making those up and making them work. Now, I think the negative connotation of vibe coding comes from some people that have taken it to an extreme. as always online. And they consider themselves like professional vibe coders. And I mean, if you don't see the irony there, the oxymoron of that, like a professional vibe coder.

professional viber it's like saying i'm a professional amateur runner like are you a professional or are you an amateur like right but but you know there's and there's What happened with VibeCoding is that we've seen this entire cottage industry that has sort of popped up in the past year of these services and apps that have become sort of...

A middle layer for the middle layer. Like these services that promise to let you vibe code apps and even commercial software into existence just by talking to an LLM.

And I think that comes with a lot of negative connotation because, I mean, personally, I feel like you shouldn't use, you shouldn't rely on vibe coding for... commercial software for software that you're going to sell to people that you're going to charge people money for because you're not a programmer because you don't actually know the code because you're going with the vibes and because there's an

It's not just about copying and pasting some code and having the code work. When you're making software for a living, there's an entire... scaffold on top of that, which is technical support, customer support, bug fixes, improvements, infrastructure, security, database, marketing. I mean, all kinds of stuff.

It's not just by virtue of copying and pasting some code that you vibe coded. It doesn't mean that you are a software developer. So that's sort of the first line that I will draw from my perspective is if you want to make software for a living. You cannot rely on vibe coding. And we've seen so many examples of people saying, oh, I vibe coded an app. I launched it on the app store. And then a month later, the person is like, ah.

I got to pull it from the App Store because I don't know how to fix this bug. Yeah, and that's a real problem. And look, I mean, there's always... developers who are new to development might who might run into a similar problem even if they're not vibe coding coding but

I think with vibe coding, the risk of that sort of problem is a lot higher. I think, too, it's worth keeping in mind. I think when I think about vibe coding, I think of it as an extension of something that I learned very early on. And I know you've done in the past, too. which is you know it's it's an acceleration of what you and i used to do where we would go to stack overflow or through blogs for tutorials and piece together bits of code in order to create something whether

It was in Pythonista maybe, or it was me with my original iOS apps. And sometimes when you do that, you find something and, oh, you know what? It finally works. But you're not entirely certain why it works. And that, in a way, is just a slower version, in my mind, of what VibeCoders are doing today. And very similar to what I'm doing when I'm writing Python scripts now. I'm using Claude and other tools.

do that and i don't always fully understand the python that's being generated but one difference that i really like a lot is that if you're motivated to try to learn some of this along the way is that you can use the llms to help teach you.

what it is that's happening so one of the last and we'll get into what we've been working on later but one of the last steps i always take when i create a script is i ask claude to create a comprehensive user guide and explanation in plain english as to how it works and that way i can kind of go through side by side through a markdown document and the script and i can see what's what's going on how it works and i know enough about code that i can look at things and say oh this

Looks like it's a little messy here. You know, I can see where the problems might lie a little bit enough to go through and then fix things myself, or at least learn a little bit more so that next time I'm vibe coding some sort of script. I know a better starting place than I did for the last one, which oftentimes it's those initial instructions that are the most important to getting yourself on the right foot when you're trying to create something.

Yeah, absolutely. And I think an important distinction that should be made is that not all LLM-assisted coding is vibe coding. There are so many examples of actual professional software developers who now rely on LLM-assisted coding tools for their jobs. And those people, they know what they're doing. I've talked to a bunch of them, and they'll use it for a very significant chunk of the code they're writing every single day. And that's...

very different from vibe coding. That's very different from a non-developer like me or you taking up on an LLM to vibe code something into existence. LLM-assisted programming. In the hands of those people is something completely different. And I mean, you can take a look at Simon Willison's work, for example, or I will actually recommend checking out the blog of Harper Reid. If you don't know Harper Reid, a brilliant programmer, you know, a very...

famously worked on one of the, I believe, on the original Obama campaign from 2008. Harper Reid has published many, many blog posts about his agentic LLM-assisted programming workflow. And it's... It's a thing to behold, really, the documentation that these folks... produce with LLMs the workflows that they produce and how they're still you know sort of they are still the humans in the loop

in this entire workflow, but they know what they're doing. And for them, LLM-assisted programming has been an incredible accelerator because they know the code, they can take a look at the code, and they can rely on these agents.

to speed up work and thousands of lines of code that they don't have to write themselves anymore. And I've heard from so many people, you know... developers by trade were saying our work is now shifting from having to write every single character, every single line of code, to being actually the software orchestrators.

of this team of agents but that's a very different thing thing from bytecoding it is this whole thing is quite a spectrum and it's that's what i think is probably the key takeaway to this first part of the show is because you know On the one end of the spectrum, you have people who are arguably, I mean, a lot of people, you'll look at it and you're like, that's just lazy coding. And maybe in some instances, it actually is.

But I also think that part of where some of these negative connotations come from is that there's a feeling of perhaps this vibe coding is somehow a threat to... existing developers and i think that that's not i don't see that really being the case but you know true professional vibe coders

talk a good game and make it sound like they're coming for everybody the reality is i don't think they are because it's the people who are able to to understand the code and be the orchestrators of the agents like you said that really make a difference I don't think that this is really a big threat to any developer, whether they're working at a big company or they're an indie developer working solo, at least not in 2025, 2026.

yeah yeah and like you said there are so many people there are so many people who you know they're sort of i don't want to say selling snake oil but it's kind of similar to that pretty close where they're just hyping up this thing so much and they're making it sound like, oh, the era of, you know, if you're a programmer, you're going to be out of a job in six months. And I don't think that's the case, first of all, because somebody...

will have to program the AI anyway. So at the very least, you know, until, I don't know, until we reach AGI or ASI for that matter, and maybe the AI will program itself, but that's still... ways off. And I don't think programming is going away. I think programming is changing. But we're talking about something different. We're talking about the idea of can anyone

Here's, I think, the core of this conversation. Are we really right now in December 2025 in a place where anyone, like Sylvia, can pick up her iPhone and... write in natural language or your wife can can she pick up her iphone and write in natural language an app that will be on the app store and no my answer is no like it we're not there now

we are seeing the rise of these utilities that sort of sit in between you and the LLM. So the LLM is already like a translation layer in the middle. So there's another... Of these utilities, like VibeCode app is a popular one. There's another really popular up-and-coming one called Wabi. That's W-A-B-I. BitRig is another one, which might still be in beta.

Beatrix is another wabi-wabi, I don't know how to pronounce it, is in closed alpha. And there's a lot of hype around it. And there's sort of... And I think Google is working on something with AI Studio. I saw some teasers of Google testing an application called Build Anything. So there are companies sort of trying to... act as an intermediate step between users and the app stores, plural. And maybe that's a trend that will continue to grow in 2026. I feel like it's kind of like...

Canva or something like that, where Canva is for designers, a simpler tool than maybe Photoshop and the other Adobe products. You know, they're trying to create something where like you and I could go into Canva and create a flyer. or a web page in a way that maybe we couldn't if we were using Creative Suite. I think that that's the goal of these products. I have yet to see one that I think is very good. I haven't tried all the ones that you mentioned, but I have tried some.

And I think that they fall down pretty quickly beyond just the very basic UI elements like buttons and lists and things like that. And also like the moment that you start iterating on these things. And I know because, as I'm going to talk about in a few minutes, I've built three days ago a Mac app myself. I vibe-coded it into existence. But I'm not a programmer.

But I had some underlying technical knowledge to know what was going wrong. And I could tell in technical terms, the LLM, what I thought was happening. And I was able to fix it and to make it work. You're absolutely right. That's a great point. Absolutely right. But there's still a lot of friction to these tools. And I think in 2026, we will see some of that friction go away.

Yeah, I think we've already seen a lot of it go away in the sense that I feel like the kind of stuff that you and I have been doing is actually easier than creating shortcuts. Because I think that... tools like python scripting tools like python are more robust and reliable once you get to the final product whereas a lot of times shortcuts just

break for so many different reasons and i think that there's a friction there despite the fact that shortcuts was founded on the idea of making coding more frictionless it still has enough of a learning curve that I found myself gravitating more and more towards building my automations with Cloud Code.

and python scripts and you know various other combination combinations of things because i find myself fighting the tools less and getting to a result faster and something that a lot of times is more complex than i could ever have created in shortcuts to begin with yeah yeah and i read something that i think is maybe not necessarily true today but

It's something that I will want to talk about next year. I read this idea of we're reaching the point where it's going to be easier to create something than to distribute it.

on the App Store. And I think there's something to that. There's something to that idea. And this is why we're seeing the rise of these vibe coding apps, these utilities that sort of act as... galleries for people to discover other sort of like a new generation of app stores that try and and some of them even promise to give you the ability to publish to the app store and it's still convoluted because you gotta have a mac you gotta have a developer certificate you know

the kind of stuff that most people will just not bother. But there's something to that idea. And I think Apple is paying attention. I think Google is paying attention for sure. There's something to the idea of people may soon have the tools to just use plain English to create... a widget, a utility, something, but then how do you get it?

onto the app store well that's why in part why i think apple revised the mini app rules just as you know past month is that they see coming what started with kind of chat apps and things having mini apps inside them now Now, we talked about this recently, OpenAI is going to have many apps within ChatGPT, and they're not going to be the first one. They're not going to be the only one either.

Yeah. So let's talk about the things we've built. And first of all, I think we have the same workflow, judging by our notes. So personally, I always start in the regular chatbot UI. Yeah, I almost always do too. In the Claude app, or I used to be ChadGPT, now it's just Claude. And I just brainstorm an idea.

And I start asking, like, do you think this is possible? And if it's possible, how? And I start asking follow-up questions. And then when I think I have a complete idea, I usually ask, okay, put together all the things we've discussed and put together a... documentation, like a document and a plan for a future version of you, the LLM, to start building this, like a roadmap, features, you know, milestones, like put together a comprehensive document.

Then I take that document and usually like my tool of choice now, especially that it's available on iOS and on the web is Cloud Code. I set up a private GitHub repo. And I point Cloud Code to it and I give it the documentation and I'm like, okay, start this project. This is something that I've brainstormed with the previous version of Cloud. I want you to start building it piece by piece. That's sort of how I approach this. And then I go from there.

Yeah, I do something very similar. I oftentimes will use, I think you and I both use a project in Claude that... that generates prompts just kind of formats them with the XML and things. I often will run it through that as one of the last steps. Once I have my instructions ready to go, because that's just a project that. is basically has the prompting documentation that anthropic has on their website

baked into it so it can take it and tag it and do things to make it a little more efficient to use within the back and forth of the chatbot. So I'll do that. And then I use, I typically use clog code in the terminal, in the Mac terminal. which similar I haven't had over the Thanksgiving weekend. I didn't spend time on it, but I want to spend more time on cloud code within the actual cloud app itself and see how that.

works and look just looking at it briefly it looks very much very similar to what you see in the terminal as well so yeah that's that's that's how i get started as well and i also would i guess one of my tips would be try to It's easy when you're brainstorming ideas to spin them into something.

very complex right you go from a kernel of an idea to oh i'm going to have all these features it's a problem in software development in general i think it works best when you're working with these llms to start with core idea and then once the core idea is working then start maybe refining it and building off various features which is what i tend to do

That is excellent advice. And especially documenting everything. If you're working with an external API, like I don't know, not sure what to do, is to give the LLM the official URL of the documentation of that API because every so often the LLM will just...

hallucinate an API endpoint, for example. Give it all the documentation that you can find. Think through all the edge cases. I mean, you're basically abstracting in natural language, in plain English, all the things that a human developer typically...

goes through. And you're just brainstorming that. That I think is like, sure, you can open an LM and be like, hey, I want you to build this. And it's probably going to do it. But I think if you want to build even just a tool for yourself, because these are just tools for myself that I've built.

There are going to be edge cases with a future version of iOS, with a future version of macOS, you know, and try and think through everything before you actually produce any code with the LLM. Try and document all the possible things you can think of. Yeah, I mean, you can do things like you can say just build me a task manager or build me an app for converting images from one format to another. But you're not going to get necessarily what you want because that's very.

generic and general instructions if you what you really want is you always want to deal with pngs and jpegs you should specify that and get into the details that you get back what you need yeah so can i tell you about what i've built

yeah i want to hear i haven't heard everything i know okay so i i have to use on my mac a chromium browser I really have tried with Safari. There are so many web apps that don't work in Safari and other web apps that just have better performance in Chromium than they have in WebKit. Yeah.

And something that annoyed me about using any Chromium browser is the lack of shortcuts integration for running specific shortcuts with input from the browser. Things like, you know, even just... pinning specific shortcuts to the browser toolbar, or running a shortcut from the browser with the URL of the current web page's input, or the HTML selection.

of the current web page as input. That's something that only Safari can do by default and Chrome cannot. And I realized, well, maybe there's something that I can do. So I've built... Again, using the knowledge that I had, I knew, for example, that on macOS, apps like Raycast or BetterTouchTool

can look into the private shortcuts database and can load a list of all the shortcuts that you have on your machine, along with metadata for their names, their icons, and all those kinds of things. And so I realized, well... maybe there's something that I can do. And so I started brainstorming with Claude. I have built a host menu bar app for Mac that allows any Chromium browser

with a companion extension. So I've vibe coded both the Mac app that lives in the menu bar and the extension that you need to install manually in a Chromium browser that communicates with the shortcuts app. behind the scenes and the extension lets you run from any chromium browser any shortcut you want with any input from the current web page that you want.

Oh, that's nice. The title of the page, the URL of the page, or the HTML selection that you currently have selected on the page. So is it kind of walking through the DOM and pulling out all the matters you made today? Yeah, so the Mac app lives in the menu bar, uses a bunch of private APIs and shortcuts events and all those kinds of things, reads the list of all the shortcuts that you have installed from the private SQLite database.

the shortcuts has on Mac OS somewhere in the file system. And to give you an example, because I knew that that was going to be tricky, I told Claude, I think you're going to have to build a Mac app. that is going to request full disk access permission right because it's got to get into those library support files because you gotta get into those files and claude was right it was like ah yes you're right you're absolutely right and so and look

I'm not going to distribute this anywhere, but it works. It totally works. And I can now use a Chromium browser with a shortcuts extension that nobody has made. And I... wield it into existence yeah no that that's amazing let me let me ask you this about it though where like what was the hardest part because i one thing that i'm i've been hesitant to kind of bridge the gap to yet, is moving from creating scripts in quad code to actually coding an app in Xcode. Was that tricky? I mean, what?

I'm not using Xcode. I signed into Xcode to authenticate my developer account. But then Claude put together just a folder with a... dot Xcode project file. And there's a build script. So the building happens in the terminal. I just point my terminal. Oh, it's all happening locally without having to involve Xcode. It's using the Xcode compiler in the terminal. So I'm not using the Xcode interface because I don't know Xcode. It's just so complicated to use. It's a lot, yeah.

I just point the build script to a folder on my computer, and it builds the app from there, copies the app into the applications folder, and it works. Excellent, excellent. Well, that's good to know. I'm going to have to check that out myself. That's very good. Yeah, and I told her like... build a modern SwiftUI app, read all the latest documentation APIs for macOS Tahoe, read about the manifest V3 that you're going to need for Google Chrome extensions.

make sure that everything is modern with the latest APIs. And so that idea of like, yes, you can vibe code by just saying, build me a task manager, but you cannot vibe code by saying, look, I want to build a menu bar app that looks into the shortcut database. You're going to need, for those kinds of projects, you're going to need slightly more technical knowledge. But still, is it vibe coding? Yes, because I am not a developer. Right. I am not going to sell this. It's just for me.

It's going to save me a lot of time on a daily basis whenever I'm using a Chromium browser. Well, your advantage is that you understand how shortcuts works. You have enough knowledge of those. I am technical enough to know what to do. You know what question to ask, basically. You know the prompt to make, basically, where someone who's not in your position might not be able to do that.

So let me mention a couple. One of the things I've been doing is just creating a lot of little... simple scripts like i think the one that i sent to you not too long ago splitting up csvs was a pretty uh was one of those things where i was running into a roadblock because i was trying to analyze a very large csv file of amazon deals and Claude can only take 30 megabyte.

And all of these different LLMs have some kind of threshold like that. And I was above all of them. I was above Gemini. I was above, you know, ChatGPT, all of them. And so I used Claude.

to figure out what the limits were for each of them. And Claude's is the smallest, but it can accept, I think, I don't know what the limit is, but I had 15 30 megabyte files that i uploaded and had analyzed so i created a csv splitter script which is really actually quite simple there's not a lot to it i just but Like you, I knew enough about CSV files that I knew that I could take them and break them apart. All I had to do was explain to Claude,

don't break up a row, like don't split a row in half. Uh, and it does it and you can, and I, you know, then I added a couple of features like to allow me to have the ability to. specify how big the files are that it's being split up into and that kind of thing.

The biggest project I've done so far is about 1,400 lines of Python, which is that Amazon deal analysis script that... that I wrote about for the last issue of Max Stories Weekly for club members, because what it does, and it was an interesting iteration process, which I think is what...

these tools are really good for where I started out saying, all right, here's nearly a million deals for Black Friday. Go through and... first i was like what let's create a scoring system let's prioritize things that are in the mac story setup page Let's prioritize things that we've reviewed before. Let's prioritize brands that we generally like at Mac Stories. You know, I created these lists and I said to Claude, let's build a scoring system on this.

And then I, you know, let's analyze how much is the discount, both in terms of percentage and absolute dollars. How much does the item cost? What is the category? I wanted to, you know. specify that it would be mostly electronics. So there's a whole bunch of things there. And when I first ran it, I got the scoring system in place. Then I got the script ready to go.

that would take all these deals throughout all the ones that didn't meet the criteria. And then I looked at it, and I was like, oh, there are tens of thousands of no-name iPhone cases in here. And... quite strangely, a whole bunch of pet products because I guess there's just a lot of tech-adjacent pet products that I had no...

I had no ability to really judge one from the other as to whether it was good or not. So I filtered out pet products in the iPhone case. It was this back and forth of, all right. I want to, skinning down this list is something that I can actually review and from an editorial perspective, pick the best deals.

But I also need to filter out a bunch of noise. I have to deal with duplicates. And so as I iterated... time after time after time that's what i did i would add another feature to it whether it's a filter or a sorting thing and at the end i got to a thing i got to a point where i was like okay this is a good list but Companies like Anchor have so many products that it was overrepresented. And so I created a distribution rule that basically went across brands and categories.

and spread them out a little bit. So it's not good enough that it just has a good score, but there has to be a spread of products. So it was an interesting list and not just a bunch of anchor battery deals. So yeah, that's how I did that.

Yeah, I guess the other thing for me will be that I've built a lot of just a lot of JavaScript, a lot of JavaScript scripts, mostly for actions in drafts. Yeah. on ios because you can build custom actions that run javascript and you know drafts has its own library and so something that i've done i've created a project in cloud with the drafts documentation

as attached PDFs. And also in the project instructions, there are the URLs that Claude can visit on the Drafts Scripting Library website if he wants to check again. to make sure that it's not hallucinating a module, for example. And draft sections I've just built with chat in the chatbot UI. And going in there, I'm like, hey, can you make me an action to append?

a line of text to my daily note in Notion. And I built these tools two months ago and I use them every day. Like, I just, you know, they're perfect for that. Used to be that I would spend, I don't know, a week. Copying and pasting bits of code from Stack Overflow and Reddit. I know because I used to do that 15 years ago with Pythonista and editorial. And now I've done it with drafts using LLM.

And for type in mind, I've done the same. So in type in mind, you can create plugins. Those plugins are based on the OpenAI spec. So it's a spec document plus some attached JavaScript that the LLM can run locally. And I've also created a project in Cloud and I use the chatbot UI for those. And some of the more advanced ones, I've actually set up a GitHub repo and I've iterated on them using Cloud code. But yeah, I've built a lot of them.

use them every day. I don't think about them anymore. They, you know, I've worked through a lot of edge cases and bugs and now they're pretty good. Like, you know, I got this, this essentially this. custom plugins for Todoist and Notion and Apple Music and Spotify and all these services that offer an API and I use them and... really a lot of JavaScript. Like these LLMs are really good at JavaScript and really good at Python because they are written in Python.

They're actually like most of the people who work at OpenAI Anthropic, those are Python programmers. Yeah, because they're all academics, which is Python is a very academic scripting language. Yeah, it's interesting. I've found that I really enjoy.

building things for notion with cloud code because the notion api is very detailed you can do a lot with it but it's also kind of complicated like very complicated right like if i had to sit down and learn notion api it would just take me forever but i can have an idea and say look i want to create a database that has all these properties and i want to run this on a schedule using like lingen x on my on my mac which is what how i'm

like the daily deals for Mac Stories deals. But I also, one of the things that I'm exploring that I think will work out really well is that Blue Sky has a very good API.

mastodon's not not as good but it's still it's it can get the job done for what i need i want to set up a system for collating all of the replies that we get across across all of the mac stories properties and synthesize that into some sort of database and report system in notions that like you and i instead of like switching through six or seven accounts to see what's going on can just kind of either just take a look at a quick summary that's been generated by quad

Or go in and look at individual comments or questions we got and maybe highlight ones that are like somebody's reporting that we have a bug or a typo or something like that. Pull that stuff out of the noise. so that maybe once or twice a day you and I can look at it and respond to those things and those people. Interesting.

And I think that's totally doable. It's just the thing that took the longest is just grabbing all of the API keys for all those accounts and everything. I've got it all ready to go. I've got to just find the time to take it over the finish line.

That sounds like the holiday project for you, right? Yeah, it's definitely one of those, yeah, for sure. Well, so yeah, I think... Before we move on to the post-show, I think looking ahead at 2026, I mean, this is obviously only going to continue. I don't think AI is going away anytime soon. I think next year...

we will see a lot of this friction reduced even more. I think, yeah, we're going to have our usual episode, you know, looking ahead at trends that we think are going to be important in 2026. But I think... We are going to see this rise of alternative app stores, apps that let you build apps. especially those that have a social component. So the idea of like building something and then sharing them with your friends and then your friends remixing what you've built. I think...

I think you're right that Apple revised their guidelines because they also know this is coming. And so keep an eye on these tools like BitRig, Wabi coming on iOS, Google, whatever they're going to do with Gemini. Google AI Studio. I think that's going to be something that we'll talk about again in the first half of 2026. And I'm curious to see if Apple will have something. There's an opportunity for Swift Playgrounds to become an AI playground for this new generation of...

vibe coders. I don't think they're going to call it vibe coding. And as we talked about, there is a spectrum to vibe coding. But I think this idea of typing something and it makes you... It makes you a piece of software that is personalized to you. I think there's something to that idea. And we'll have to talk about it again. Yeah, no, I think Swift Playgrounds is the perfect place for that because while Apple has promised smarter Siri...

I don't want to have to tell Siri to do everything every time and have it created from scratch. I mean, you could bake App Intents and string those things together right into Swift Playgrounds, and perhaps that would be the path to creating. little automations that way that are separate and apart from shortcuts itself. And ultimately, the reason we're doing this is because it's allowing us to kind of multiply.

what we're able to get done. You know, it takes, it takes something that maybe it would have taken you all day to figure out your plugin for drafts or, or typing mind or whatever.

and let you do it in an hour or two and same goes for me and that's that's been great because these are tools like i would not have created a lot of this stuff that i'm integrating with notion but for the fact that i can use claude and it still takes time a significant amount of time sometimes to build these things with all the iteration at the same time though you don't have to learn

Python from the ground up or the Notion API from the ground up. It's pretty nice that way. Yeah. All right, everybody. Thanks for joining us for another episode of App Stories. We'll be back in another week. In the meantime, you can find us over at maxstories.net. And we're both on social media. Federico is at Faticci. That's V-I-T-I-C-C-I. And I'm at John Voorhees. J-O-H-N-V-O-O-R. Talk to you next week, Federico. Ciao, John.