¶ Intro and Claude Skills Review
Hello and welcome to another episode of App Stories. I'm John Voorhees and I have Federico Vatici with me. Hey, Federico. Hello, John. How are you? i'm doing really well today i'm doing really well i want to talk to you about our latest experiments because this is the time of year as as we've talked about many times before where we experiment a lot but before we get to that
do want to mention something about Claude's skills that I discovered over the weekend. You know, we talked about this on an earlier episode, and I've been a big fan of Claude's skills because it allows you to package together scripts and prompts. in kind of a nice, tidy package where you can just kind of use them and stack them together and things like that. And it turns out there's a limit of 20, which is a real big bummer because I was making...
cloud skills that were very small and self-contained. Because since you can stack them one on top of the other, I thought it made a lot more sense to have the flexibility of having smaller units of automation. But it turns out no matter what subscription you're on with Anthropic, you're limited to 20 right now. I think because it's a beta feature, but I hope they expand that soon. Yeah, 20 is not that many.
not when you consider we have hundreds of shortcuts i mean that's that's how kind of i look at it it's like this is like what if shortcuts but you could only do 20 of them yeah that's not especially when you start making smaller skills or skills that call other skills you know exactly when you start doing that sort of thing it's easy to run into that limit i think yep yeah hopefully that that'll be out of beta soon i hope
I don't know. I kind of feel like there's a learning curve to skills. I tried making... This is actually part of the experiments that I did myself. I tried making some of these skills that... talk to web services that I use and actually run code instead of using MCP, for example. But there are so many issues with how the...
how the code environment of Cloud works. And sometimes it doesn't connect to some services that use a specific version of SSL for authentication, for example. Other times I even ran into Cloud looking at the skill. and deciding rather than running the script that was ready for Claude in a folder to rewrite the script. And that is like non-deterministic automation 101, right? Yeah.
Yes, I sent to you, I think I sent you a conversation I had with Claude to that effect where it was like, I caught it. It was like, oh, let me write you a Python script. I'm like, don't write a Python script and use up all my tokens. You've already got a Python script. Yeah, it's not great that way. It also...
gets complicated when you have a lot of API keys and stuff you're trying to juggle with using skills. It's kind of when you're in that area, it gets maybe easier just to write a Python script that you're running from a Mac instead of trying to run it inside a skill.
Yeah, I kind of wish if Anthropic is serious about skills as opposed to MCP or maybe both will coexist. I don't know. But I wish that there was, first of all, like a keychain system where you can securely store things like API keys or stuff like that. in a global environment in Cloud that the code interpreter can securely access. But also there needs to be a better UI for putting together these skills. Like the fact that you cannot even replace a skill.
If you have a folder on your computer and if you have a zip file on your computer, you have to delete the skill. No, it will replace them now. Oh, it replaced it? I think that they updated that. I've seen little changes here and there, but if you have... a skill that's named the same or close to the same it'll ask you if you want to replace it interesting interesting so you can do that and they do now also a lot of times when you build one it'll
download a dot skill file which is really just a zip instead of because initially when i first was experimenting i was getting zip files back from claude now it gives you a dot skill now it gives you a dot skill which is their format which is really just which you can change to dot zip on mac os and it's going to open as a zip file in a folder exactly it's just a it's just a zip file by another name basically
¶ Open Source AI Model Experiments
But I think these are part of the experiments that we tend to start doing now before the holiday season. Some of them will carry over into our holiday breaks in addition to spending time with family or playing video games or doing all... the things that we do but we try to keep some of the tinkering on the side some of the experiments on the side and we and we carry those over into the new year when the new season begins in january so uh i figured you know this is a classic app stories episode
where we can just round robin some of the things that we've been playing around with, some of the experiments that we've done. And I wanted to start, I talked about this before in App Stories Plus. for subscribers. We are going to talk some more. in App Stories Plus today in the post show about automation and the changing role of shortcuts and classic automation in our workflows. I have spent more time with the web app that John doesn't really like.
called Type in Mind. Now, John, we know that John doesn't like web apps, first of all. Oh, come on. I've used a lot of web apps, and I'm not anti-web app, but I do find... Typey Minds UI to be a little cluttered and messy. There's a lot going on. There's a lot going on. You're going to some of those tabs. There's a lot of overlap when it comes to like plugins and agents and global settings and things like that.
Yeah, it's a little hard. It's one of those places, you go to that app and you think, oh man, where do I even start? Because it's a little bit, it's got an intimidation and learning curve factor to it, I think. Yeah, you should see. You should see some of the other things that some of these services that, because I have done a lot of research about this, like running your own AI, whether it's local or an open source model.
hosted somewhere else. I have done a lot of research and there's a really popular... I actually... I didn't even bring this up with you because I didn't want to waste your time and it was a short-lived experiment. The really popular... way to run these models is with this open source application that you need to host yourself or host somewhere else called Open Web UI. It's like, imagine typing mind, but like 10x more confusing. And I hosted this for like two days.
on a server from a company in Switzerland or Germany. I don't even remember. And it was so confusing. I was like, you know, I tried this for 48 hours and I'm done. And I went back to type in mind. Okay. type in mind right now, this is sort of like my test bed for open source models. We have seen a lot of progress from open source Chinese models. And I'm going to go through my list right now. I am experimenting with...
four, but really there are three in the sense that Kimi K2, this has been my go-to for the past few weeks. Kimi K2 now comes in two versions. There's a traditional Kimi K2 and a Kimi K2 thinking. So they introduced a reasoning. model. My favorite one continues to be Kimi K2 on Grok with a Q. It's really fast. Sometimes I can even hit like 400 tokens per second. It's incredibly fast. This is a non-thinking version of Kimi K2, which is not as great as, I got to say, the new Kimi K2 thinking.
I am testing via the official Moonshot API. It's not fast. it's a lot it's like 60 to 70 tokens per second which is still faster than claude i was gonna ask how that compares to like sonnet 4.5 yeah son it's like 40. 40 tokens per second and when you hit like the peak hours like i don't know about you but my clock gets real slow during the day Sometimes you see literally like one letter at a time showing up. Like that? That's like 10 tokens per second or something.
So Kimi K2 thinking is faster than Claude, not as fast as Grok. The Grok infrastructure with the, what are they called? The LPUs that they have, custom chips. for running AI really fast. But Kimi K2 thinking surprised me yesterday. I was doing some research on a topic and...
On its own, when it realized that one of the services that I wanted to use was not working, it automatically decided to reroute the research to another set of tools. That's the sort of agentic behavior that Kimi K2 Thinking has. which is really impressive. I need reasons through every step. It's really, it's like... Imagine an open source version of Claude that you can host yourself in theory. I really anticipated the moment where Kimi K2 thinking gets up and running on Grok.
That should be interesting to see. Is Croc the only place where the regular traditional Kimmy Cake 2 is available? You can't do Cerberus. It's available on other... It's not on Cerebrus, unfortunately. Cerebrus. It's not there, but I'm going to talk about that in a second. No, Grok is your best option for Kimi K2 right now. And hopefully they'll put up the thinking version as well. I have also been testing Minimax M2.
This is another Chinese model. Also a model quite, I wrote about this on my story, it's very similar to Claude, supports interleaved thinking steps. So basically when it's doing a task, when it's generating a response, it thinks. in between the responses. So it's got this intermixed thinking blocks. And I've also been using this via the official API. Not fast.
I would say I don't like it as much as I like Kimi K2. Okay. I have been... really impressed with the latest addition to the Cerebras infrastructure, and that would be GLM 4.6 by ZAI. This is another Chinese lab just called Z.AI. I think the full word is Zipu. I'm not sure the proper way to pronounce it. I'm sorry. GLM 4.6 is a really good coding model. It's not a reasoning model, but it's running on Cerebras at like 1,700 tokens per second.
Yeah, that's wild. That has become my go-to Vibe coding model for drafts actions. In that, yeah, it does not one-shot a thing. Like Claude can, for example. Drafts automations are based on JavaScript, right? Yes, sir. Pure JavaScript with some drafts-specific libraries and modules. Right. Claude tends to one-shot an action or a script, you know, most of the time. And maybe GLM 4.6 doesn't. But because it's running so fast.
yeah you're just iterating over the iteration that it allows you to do it's like demo like when i see claude doing like line by line in a script Cerebrus just spits out like a script in a second. Don't even mention the fact that if Claude, if you're on like a pro account, which is what I'm doing still. Not only are you iterating slowly between requests because of the speed of the tokens coming in, but once you cap out on your tokens...
You're going to take a break for four hours before you can get to the next stage of your project. Yep. Here you can just vibe code away. And yeah, so really my main two open source models, right? now are Kimi K2, especially the non-thinking one because it's so fast and still very good at calling external tools. I built myself a collection of plugins.
Give me an example or two of some of the plugins you're building for yourself in drafts. Like what kinds of things are you building? I tend to build plugins for a couple of reasons. The first one is that TypeInMind does not support... external connectors yet. They're working on it. They posted a teaser and you will be able to connect to Notion, Gmail, all these connectors, just like you can in Claude.
So using drafts as an intermediary to move the data back and forth? No, no, no. So drafts I just use as my note-taking app. So those are just... My drafts actions are things like for Markdown, for Notion, for exporting documents. Those are like separate. The plugins that I built in Type in Mind, I built them because Type in Mind didn't have...
easy to use connectors. That will change soon. But also, I've mostly been building plugins for services that don't have MCP at all. Like Readwise Reader, for example, doesn't have an MCP server. Right. So I built a plugin that uses the API. And in type in mind, I can just, now I can save URLs. I can search my Readwise reader library. I can export, you know, highlights from there.
I just have a way to communicate with natural language with my Readwise Reader. That's nice. That's something they should really build into Readwise Reader. I'm surprised. All they have right now is the ability to... use ai to do things like summarize and one article one article at a time which i don't really like i mean i'd like to query my entire database of things that i've saved that's what i can do now i was just testing it today like how many articles do i have
saved from Mac stories and it runs through and it's, you know, it's very fast. But yeah, right now, these are my two non-CLOD models, GLM 4.6 and Kimi K2. All right. Very nice. Very nice. Well...
¶ John's App Discoveries
So I'm going to share a couple of different experiments I've been doing. Some of it is AI based, but I've been working. I'm going to start with apps because I've been using an app called Bloom, which I did recently mention. Bloom is a finder replacement. You're a brave man. It's super actively developed. And I've used forklift in the past, which I like.
But there were a couple of friction points. I had some issues with it properly allowing me to manage Google Drive files, for instance. At one point, this may have cleared up, but at one point I was having trouble. deleting the file that someone else had shared into a folder that was uh
you know, a joint folder. And Bloom doesn't have an issue with that at all. The main thing I've been using it for is, well, there's a couple of things. One is that it has very easy ways to set up multi-pane views everywhere from just a side by side to a four by four to a, you know.
three three pane setup and that can come in really handy when i'm managing a lot of files moving them around for podcast production the other thing that's pretty nice about it is you can set up those views with particular folders and define them as workspaces so you could have a workspace that is your writing workspace and maybe it has you know where you keep your screenshots and where you have your your
markdown files, or you can do what I do, which is with some of the podcast stuff. It'll be like, oh, here's the inbox where everybody's sending me their files. Here's where I'm going to put it. to process it and here's kind of like the third folder with the final final work product that's going to get uploaded to youtube or libsyn or wherever so It's a really nice app. I've been really happy with it so far. I haven't gone as far as to try to, I think I can essentially force.
the mac to open bloom instead of finder when it does you know when it automatically opens something from safari say or some other app haven't done that yet still kind of feeling my way around it but so far it's been very solid and i've i've really enjoyed using it so i'm going to keep keep doing that and exploring some of the
The other features it has, I believe that you can also run some terminal commands and things like that directly from within Bloom as well. That's the kind of thing that I haven't really messed around with too much yet because I haven't really had a need for it. But yeah, there's a lot you can do there.
The other thing, this is a surprise for you. I've been messing around with superhuman again, Federico. Oh. now he's seen the light yeah well we'll see we'll see jury's still out jury's still out i um you know shortwave we were both on it for a while and i do like shortwave i think the thing about shortwave that i like the most that i've missed
the most going to superhuman is the ability there's two things i guess first one is they have this this concept called bundles in shortwave where you can get you can bundle together sets of emails from like the same company or the same domain so that say you go into your your unimportant email folder or whatever it's called your other folder and you have a whole bunch of things from i don't know some some
company so you know sending you you know like uh one x player which is the handheld manufacturer who sends out like three emails a day yeah and i finally i'm unsubscribed because they were driving me insane with their uh with their one next flag, but...
Yeah. I mean, it would bundle those together. And then you could like, if you're like, I just don't want to deal with one X player today, you can just hit delete and there'd be like five emails gone all at once. They don't, there's not really that same sort of thing in superhuman. It's more of a label based approach. which is a very Gmail-centric kind of thing, which is fine. And what I've enjoyed, I think, the most about Superhuman is the keyboard aspect of it, how quickly I can, on a Mac...
go through and archive or delete stuff. I don't like that deleting an email, this is very specific, but I don't like that email is deleted with shift three, you know, the pound symbol, because that's a two key. combination and i'd like to be make it easier for me to delete my email and there's so what i did is i made it a hot key on my logitech creative console so why don't you just archive everything because
search becomes a problem. I know they have good search on superhuman, but you know what? I feel like I need to at least make a rough cut and make sure that the stuff that is archived is at least stuff that's worthy of saving even a little bit. I save a lot. I archive a lot, but I get a lot of junk too. I don't want to save all these, you know, discount codes for Black Friday. I don't want that stuff popping up. Okay.
Like if I have 10 emails with discount codes and offers from JSaw, and then they email me about a sponsorship, right? I don't want to have all of those. those promotional emails get mixed in with the emails from actual people at the company. That's one of the things. That's probably something that you've not run across very often, but it happens, you know?
But yeah, I've been enjoying it. I mean, it's good. I still feel like I've never yet met an email app that I really am 100% happy with. And I think that that's probably going to be true of Superhuman 2. It does have some good tools for moving quickly through email. Yeah.
Yeah. Okay. Well, I'm pleased to hear this. Finally, maybe if you stick with it, let me know so I can share superhuman links with you. Oh, yeah. Well, for the time being, you can. At least, you know, I'll let you know if I drop off.
¶ Federico's AI-Powered Services
Okay. All right. I want to mention some services that I've been testing lately. Most of them are integrated with different AIs and they're just a plug and play in the sense that I can use them with cloud. I can use them in type in mind. And all of these were new to me. So this is part of a bigger topic where like using these AI tools, I'm actually learning a lot of things, which is interesting.
It's a fun thing to be learning about a lot of different startups, a lot of different tools. I'm just going to mention this. Firecrawl, this is a much better provider for Cloud. If you need to... fetch the contents of a web page. If you need to scrape a web page, Firecrawl is much better than Claude's built-in web fetch tool. Oh, okay. I actually asked Claude, I told him, take this Mac Stories article and fetch it via your own tool.
and fire crawl and then be honest and let me know which one is the better output. And Claude said, I actually prefer fire crawl because everything is better organized in Markdown. I think WebFetch is a Python thing. That's what Claude's using. And I actually got Claude to output. the results of both. And the web fetch was like a mess. It was a jumbled mess of like, I think they're using like some really basic, beautiful soup for in Python.
to parse web pages. Whereas Firecrawl, it gives you like a really nicely formatted markdown version of a page. You can use this with AI. You can use this with shortcuts. They have an API. And if you need a really good parser, for web pages to give you a markdown. Really well done. They have a credit system. Have you tried it with any JavaScript-heavy sites? Sites that aren't really articles. Maybe like the Apple.
homepage or something like that? Not yet, but I can try. Yeah, I'd be just surprised. I'd just be interested to know what it gets you. I mean, a lot of times... The Apple homepage? The Apple homepage. Yeah, which is mostly images and a lot of animations and that kind of stuff. I can ask right now, fetch this with fire crawl and... give me the markdown markdown
So Firecrawl is very specifically a Markdown output type of tool. Well, they extract plain text and they tend to favor Markdown. I also want to mention Parallel. The parallel is this new... This is one I've tried, yeah. This is a new startup. And there's a whole bunch of these startups on their eyes, like alternative search engines.
Built for LLMs. There's EXA, which we mentioned before. Turns out there's a whole bunch of companies like this. John, I got to tell you, the Apple homepage is a pretty nice markdown document. that i'm gonna send to you it's uh you can see uh i think it did a pretty good job oh yeah getting the markdown of apple.com yeah It did, it did. It does a nice job of, yeah, it's got all the headers and it's got the images. Yeah, this is good. Parallel is another search engine built for LLMs.
And it actually does a bit more. They support three modes. So they have search, extract, which is similar to fire crawl, but I still prefer fire crawl, and deep research. And I got to tell you, I tested Parallel. First of all, I signed up and they gave me $80 for free. Same. In credits. Nice. Federico, they gave me $78.66 of credit. I don't... I have no idea why. No idea why. And the search is pretty good. I will say that the search is better than EXA, better than perplexity search.
I tested it with Claude. It got some really good timely results for recent news. A really good search engine for LLMs. The deep research is fascinating. You can kick off a deep research task remotely. It starts running in the parallel cloud. They give you a URL and an ID for the deep research task. You can go there, you can check. the progress. It tends to, depending on the complexity of the query, it may run for up to 45 minutes. I know, I saw that.
It scans thousands of web pages and it gives you a report at the end. I found that these reports to be much better sourced than Claude or Gemini. or ChatGPT. It's really, really good. And I configured the parallel web app so that when it's done, it sends me a notification via pushover. using zapier so we mentioned pushover before and so this is something that i set up a few days ago and i did some research i have i have this this like test evaluations that i do all the time like um
do research and give me a list of modern markdown editors for iOS that supports user-created plugins. There are really three of them, but I'd like to see how many AIs fail at the job. For example, Claude... told me about editorial as if it was like a new thing turns out turns out you
Markdown text editors with plugins, you really have three or four options. You have Obsidian, Drafts, Noteplan, and there's another one that I don't remember the name. All the other ones are discontinued. So it's really nice to see the different AIs fail.
That's a good question. I really like that one. I personally know this information, right? Because I really follow this space. And so it's a good test to see. I treat them like interns, basically. Like, do this research and come back to me. I want to see what you gather. And Arlal did an excellent job, I got to say. That's cool. I like it. I was reading through their blog over the weekend. And their whole thing is they say that their methodology of doing the search.
uses fewer ai tokens and it's faster as a result now that the the deep research does take a while because it is doing a lot of hitting a lot of web pages before it gets back to you but this is the kind of thing where I think a lot about... like why why is it that google doesn't have these tools i mean it's because they're an ad-based business and this is searching the internet without ads uh i suppose but but it is pretty if you if you do heavy duty internet research
There's a lot of potential for these kind of tools. Yeah, yeah. So yeah, those are the other. And did I have anything else? Firecrawl, Parallel. No, yeah, these are the two that I've tested. And also Firecrawl. When you sign up, you get 500 credits for free. And so all of these things, you can just sign up with your Google account, get some free credits, and take them for a spin, and you don't have to pay.
So this reminds me all very much of the early web when no matter what you signed up for, you got incredibly huge. credits for everything like, you know, I don't know, groceries or dog food or whatever you were buying on these, on these services. And, and that all went away when, when we had the, the web crash and that, I think that that's all going to go too with the.
with AI if the bubble pops. But for right now, you can get a fair amount of service for almost nothing. And if you're not an enterprise, if you're just individually using these services... It takes a long time to get through some of these credits. Yeah, just use them for free. Yeah, because it's billing you in cents.
single digit dollars at the most in most cases. So yeah. Yeah. Yeah. Very good. Very good. I, I mean, I'll close out by mentioning that for me, I've been continuing primarily working in Claude with.
¶ Custom Automation Workflows
cloud code in the terminal, building Python scripts and connecting them to Notion. Because I've really found that one of the nicest things...
One of the things that has always bothered me about spreadsheets is they're all so designed for numbers. And we don't deal with numbers a lot. We do deal with numbers, but we deal with text more than we deal with numbers. And what I like about Notions databases is they're really... flexible in terms of the kind of data that you put in each field you know urls and images and files and text and numbers all kinds of stuff and
With a good Python script, I've been able to do all kinds of things. I've already mentioned before, and especially in the pro show, I mentioned some of the stuff I've been doing with Amazon's. They call it, I think they call it the purchase API. It's their API that shows you what all the deals happen to be at any one time at Amazon. And I started off with an analysis of Black Friday deals because Black Friday.
the spreadsheet that amazon offers to people in their affiliate program is like 450 megabytes it's literally a million rows of data that you could never analyze on your own but using python and coming up with a a multi-factor filter like i filtered out used items i filtered out uh all kinds of
you know duplicates things that were like just differences by color and then i did a ranking system based on whether it was something we covered on mac stories before you know what the amount of the disc the magnitude of the discount was all kinds of different scoring systems that i came up with and run through in a python script and then it just populates straight into into notion and i took that that ranking
algorithm that I had developed myself. And I built it then into a daily Python script that runs at 4 a.m. every day on my Mac studio. And when I come down to my office in the morning, once I get up, there it is. And it's replaced all of the expired deals, indicated which ones are new, and then ranked them all. And so I can look at that database.
And I know exactly what I've got there and what's good for putting on Mac Stories deals. So this is like, that was all an experiment using an API that can return tons of data just to get my feet wet and really understand. the kinds of things that I can do combining Python and Notion. And I expect to do...
a lot more of that in the future with other APIs. It was just that this was a good place to sort of experiment with it and actually have a little benefit for the site as well along the way. Yeah. That's actually stuff that makes us money. So that is a good thing, right? It's a good thing. The final thing that I will say, I know a couple. First of all, I built myself a little drafts action that basically acts as a quick way to go from something that I'm reading in Readwise Reader to a linked.
post on Mac Stories. It finds all the articles that I have in my Readwise Reader queue that are tagged with the tag linked. And from there, it fetches all the highlights. from the article some things that I highlighted in the article and when I click export it prepares in drafts a markdown draft with the title the name of the author, a link to the source, and all of the highlights as markdown block quotes.
so that I can basically get a head start on linking to somebody else's article very quickly, going from something that I'm reading and highlighting in Readwise Reader into a draft in drafts. All of that was done via JavaScript and Claude. No, actually Cerebras and GLM 4.6. Because I absolutely have no idea how to deal with these things with JavaScript. But now I, you know...
Instead of cobbling it together with Stack Overflow, now I get to do these things like this. Yeah, same with me with the Python. It's the same situation. And lastly, I spent about a week...
¶ The State of Web Browsers
testing all browsers again. It all started because a couple of web apps that I wanted to use didn't work in Safari. And so I thought, oh, maybe I should go down the Chromium browser rabbit hole again. And so I tested. Firefox, which is not really Chromium. Firefox, Chrome, Proplexity Comet, Edge, Atlas again, Vivaldi.
And then I realized, I went back to Chrome again, and then I realized, oh, this is making me miserable. And I just went back to Safari and decided that those two web apps that I need to use on my Mac, I'll just save them as PWAs with Google. Chrome in my dock. And so those web apps are in my dock using the Chromium engine, but Safari is my browser of choice. I just, the fact that Safari on iOS and on the iPad has a proper interface.
It supports extensions. Like there's so much going for Safari on iOS right now. Even beyond the fact that in the United States, not like in Europe, things are any different because of the DMA. But like... Even with the fact that all browsers on iPhone and iPad need to use the WebKit engine, fine. But the other browsers are just plain bad. They're like bad browsers. Chrome on the iPad is a bad web browser. Doesn't let you pin tabs. Doesn't let you show the bookmarks toolbar.
Like, there's so much wrong with these baby browsers for iPhone and iPad. Apple is the only company taking the iPhone and iPad seriously when it comes to web browsers. Maybe Vivaldi. He's actually doing a decent job on the iPhone. I would say that they are. Everybody else is bad. Yeah, they just have a long way to catch up, I think. They're still doing, like, fundamental.
features but but yeah they're pretty good i i think especially on mobile you're right i mean on the desktop you know i think sure yeah edge is okay on the desktop i'm not a fan of chrome myself but i do use it from time to time And also like in my test, Safari was faster than Chrome on Mac OS. I did a bunch of benchmarks, like speedometer, like those things. I don't know how they're called. Safari was faster.
And I have a whole bunch of extensions installed even. And I don't know. But like on the iPhone and iPad specifically, Safari is the only... Apple is the only company that is taking those devices and those users seriously. Instead of giving you like a...
companion browser or a lightweight browser, which is something that in 2025, I honestly cannot stand anymore. The treat in this device is like, oh, you want to use an iPhone to browse the web? That's cute. No, I just want to have a browser. I just want to have a serious browser.
my iPad Pro that I paid 2000 euros for. I want to have a proper browser with windows, with tabs, with toolbars. Like imagine this idea, a toolbar. These companies are so frustrating to follow. And so yes, Safari is the only good browser. That's my take. That's the hill I'm willing to die on today. So far, it's the only good browser. Any app that says, even hints that it's a companion app in the title, I don't go anywhere near it. Go in the punishment.
Box in the corner. Go away. Nope, not for me. In 2025, if you're still emailing me about your companion app on the iPhone, don't.
¶ Companion Apps Rant and Conclusion
Honestly, don't. And that includes Apple. When they were making those companion apps, remember when they were making Logic Remote for the iPad? They still have that. That still exists. Come on. That still exists. Come on. Come on. I know. I hear you. I hear you. Companion apps. That's a good way. That's a good way to end the show, Federico. I can think of one companion app that is a companion app because it has to be a companion app.
which is Raycast. Yes, I was going to say that, but they're making the best out of the situation. They're making the best of a bad situation. Yes. They are. And I think it's smart because they'll be ready if it ever opens up to do more. That is a good call. And actually, Raycast on iOS and iPadOS is like a completely different app. They've been very smart because they're syncing your data.
the things that make sense, like your snippets, your quick links, but they're doing things completely, they're taking advantage of the iOS platform with like the keyboard that they have now for dictation. That's a way to make a companion app, not to say, oh. We cannot make a toolbar on the iPad. That is like false. It's a false statement. I'm sorry. This is my rant for today. All right. All right. Well, I'm glad we finished with a rant, Federico.
It's always fun, always fun. Well, hey, folks, you can visit us on MacStories.net anytime or Club Mac Stories. which is club.maxstories.net. We've got all kinds of stuff going over there as well. And people are in the club right now. nominating the MacStory Selects Reader's Choice Award. And then the Club Plus and Club Premier members are going to actually do the voting on those nominees. So we'll be announcing those things in December.
I want to thank everybody for joining us. And of course, you can find me and Federico over at Mac Stories. And Federico is on social media where he 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-H-E-S. Talk to you next week, Federico. Ciao, John.