RobbieTheWagner 
Adam Argyle In today's modern technological world, developers face many difficult questions. Artificial intelligence, autonomous agents, JavaScript frameworks, even CSS. With so much happening so quickly, how can a developer possibly keep up? Simple. Just listen to the world's most important AI podcast. Two experts, countless opinions, zero hesitation. The Internet is complicated. These men are not. Each week they review the latest developments in artificial intelligence and modern web development. This is Whiskey, Web and Whatnot, the world's most important AI podcast. This episode is sponsored by MailTrap, modern email delivery for developers, straightforward integration into your code with native SDKs, security compliant API and SMTP, 4,000 emails monthly in their free tier, MCP and API integrations for your agents and coding assistants, and PS 24-7 support you contact people, not an AI chatbot.
Try it mailtrap.io today. That's M-A-I-L-T-R-A-P.io. And back to that episode. Oh, hello. Hey, everybody. Welcome to Whiskey Web and Whatnot, your favorite podcast about dryer vent cleaning. Of course. No, it's about Whiskey Web and whatnot and mostly AI, things like that. Anyway, your host for today, Robbie the Wagner and Adam Thomas Argyle, the nerd. Howdy, y'all. Stoked to be here. Hope you're ready. Yeah, I am ready. I forget. Did we just jump into the whiskey? I don't know how this thing works anymore. We're just going to do it. We can teaser. We can be like, we're going to talk about MySpace.
We're going to talk about 2026 versus 2006. Building a mobile app with Cloud is building software with LLMs. Bad for your mental health. Agentic heartbeats. Prop for that. Meta harnesses. Buzz. Spectre-driven work trees. Whiskey build times. Yeah, whiskey as compared to built-ups. Anyway, and then shit that Steve Yeggie said. Yeah. Anyway, you're ready. That sounds like a Jeopardy category. All right. So, yes, today we have the Chicken Cock, always a favorite, small batch Kentucky Straight bourbon whiskey. It is batch number 17. There were 10 barrels. Am I reading that right? Mine's seven or 12 or 12.
I have 12 over seven. Okay, so we have different ones, I guess. So we have different small batches. We'll see if they're... They were all hand selected by the master or whatever. Yes. All right. Yeah, it's hunter proof, five years old, 70% corn, 21% rye, 9% malted barley. And fun fact, chicken cocks, what roosters were called before the word rooster existed. Was that fun? I thought it was fun. I was like, what's a chicken cock? I mean, it just sounds unnecessarily long. Like it's a chicken with a cock, maybe? Actually, that probably is what it was, huh? I guess, yeah. Yeah. No, I mean, because I am forever a child, I always love to get some chicken cock.
And if you've ever read their marketing page, it's like Chris Coyier got a good chuckle out of it. It's something about like the creamy, buttery mouthfeel of the chicken cock or something. Oh, God. There's so many jokes. Anyway. I hope not. I'm smelling... Hmm? Uh... What is it? Bubble yum? Is that the one that's like the big cubes of bubble gum that are like pink? Is that bubble yum? Yep. Yeah. The ones that you chew on for like five seconds and then they're out of flavor. That's like the zebra gum. Just like the. The zebra gum is gone? It's gone? It's gone? Yeah, it's gone. It went away like last year, I think.
Whoa. After like 70 years or something. No, I think it was really bad gum. No, it was really delicious gum. It's just the flavor went away in one second. Yeah. It always looked great, too. You're like, ooh, I can't wait. Okay, so yeah, you're getting bubble gum. I'm definitely getting cinnamon. Getting some almonds. The smell on the extract, perhaps strong, dude. Maybe I'm smelling from the wrong side of the glass. So top or bottom, which one's supposed to put your whole face in it. Right. And get underneath. Hmm. I think I did it wrong. It is really strong. I do think smelling from the top is better than the bottom.
Less fumey. I thought Denver said the other way, but I think he did say the other way, but I haven't been able to successfully do it the other way. I keep trying it. And definitely one is fumer has more fumes than the other. So I think I'm going to start from the top. I don't know. It's so weird. My computer, my work machine is just over here open. It's just doing stuff. It's so distracting. So if you see me looking over here, it's because my computer is doing stuff. I'm like, what are you doing? I'm not there. Oh yeah, I told you to do that. Nice. It tastes pretty bourbon-y with a little bit of spice.
Like if you mix a little bit of rye into a bourbon. Gotta have that spice. It's sweeter and it does have a texture that is... Maybe more syrpy. Yeah, it's a little thicker, which I like. All right. Well, how much do you like it? Zero to eight tentacles. Zero being the worst. Eight being the best. What do you think? Oh, man. Compared to other bourbons, it's better. Do I like it that much? Not really. It's like a bourbon, but it's got the spice. So it's got a few things going for it. The spice. Got to have it. If it's too syrupy, I really dislike it. It's a little bit thicker, but it is quite pungent on the smell unless you approach it from the top.
I'm going to give it a six. All right. Yeah, it's hard for me to compare to a straight bourbon because, like, it's a little spicier. It still has the sweetness. For something that's not like just a sweetness bomb, I think it's okay. But I'm just going to say a five, I think. I don't know what I'm trying to talk myself into a number here. I'll go lower. No, dude, seriously. But I don't know. I thought you liked bourbons more, you know? No, I like rye the best. Okay. Okay. But yeah, bourbons, there are a few bourbons that I like a lot and I don't know exactly how to describe them. One of them is the Sweetens Cove.
It's like almost like it's a little bit camp fiery. Like it's almost s'mores tasting. Like there's marshmallow, there's smoke. Like it's good. It's not like just Jack Daniels. It's not like that, you know? Yeah. But yeah, no, this is not bad. Would recommend if anybody was looking for one. But yeah, pretty middle of the road for me. pretty middle of the road indeed. I think it's just my palate's not refined enough yet for some of these. Like I tend to like rye better also, but even sometimes with ryes recently, I'm just like, I want my peated scotch back. You know, I'm just like one I love and the other ones I'm like, oh, they're fine.
Yeah. Yeah, I get it. Well, if you want something to do with ones you don't like, we've had this beer in our fridge for like, I don't know, two years now. And I was like, no one will drink it. I kept trying to give it away. And I was like, I don't want it. And then my parents were over and we got some steaks and I had nothing to marinate them in. So my mom was like, just use some beer. And I was like, hey, I know just the beer I can get rid of. Nice. And so I did. And yeah, so you could use it for like a sauce or a rub or something. Well, I mean, I kind of. continue drinking like i sip them throughout you know the weeks and i just want to acquire the flavors in here like i'm it's the same reason i had i had pappy's last month i think we talked about it and i was like whoop-de-doo whoop-de-doo but also what am i missing like i'm i'm i'm feeling fomo why why don't i curious yeah i'm curious
FOMO. Lots of FOMO with AI right now. I don't want to pivot it like that, but it's really hard. You know, you're just like, oh, what tool? Everyone's not using that. And now we're using this. So I'm on Herder now. Herder with a hunk. Nice. You know, okay. But it's still not as good as Warp. I'm like, Warp is great. I installed Herder. Like the thing that I'm wanting... Which the problem is I have too many things I want to build, right? Lots of personal stuff, lots of work stuff. But like obviously work would probably frown upon me using like I haven't looked into how Herder does it. But like, you know, a thing that opens up all your ports to continue all your stuff on your phone and whatever.
Yeah. Um, like that's what I want because I want to be able to step away, like do something else for 10 minutes. And if it needs help, it can let me know. Um, but I don't have any good solution for that at work. So because most of my work is at work, I don't try as many fun tools as much. So I installed Herder. I ran like a couple of tabs in it just to see what it would do, but it seems fine. But like, I haven't put it through its paces. It's not beautiful. Like Buzz. Buzz is really nice. Pretty much just everything that it does. I did not install Buzz for the same reason. I was like, this sounds cool, but I can't use it at work, I'm sure.
So, yeah, I was like, come on, Shopify, let's set up Buzz. And I was like, actually, we kind of have our own thing. So why would we go adopt theirs? But and then I was like, I'll set it up on my cloud machine and just try it. And then I was like, it needs too many keys. I'm lazy. I don't like going in and putting keys in places. I'm like, what I have ain't broken anyway. So. I've got such a straight relationship though of like, I like the idea of, you know, let's replace Slack with a place that you can talk to agents and people in the same area. And like, um, I guess, does it also do like task management type stuff?
Like, is it supposed to replace Jira too? Or is it not that I don't think it replaces Jira. No. Okay. Cause that's what I would want is an all in one plan everything for me. I give you a few bullets, you make me a ticket and then like, yeah, I can give it to an agent. I can give it to a person. We can, you know, collaboratively plan the ticket even like through an agent together in the same chat. Like I think there's a lot there, like good ideas that could, could work, but you know, whether that this is going to be the product that people actually use for that sort of thing. I don't know about that, but I like the ideas.
Yeah. So yeah, it's funny you mentioned the ticket sets. That was like, we had the Shopify hackathon a couple of weeks ago called Summit. I finally got to go meet coworkers I'd known for a year. And I spun up many explorations. And one of them was because we have Slack, we have our own agents we can talk to in Slack. Working together in a thread is the closest thing to multiplayer that we have right now with agent workflows like Slack. Unless you're in GitHub, we've done it in GitHub a little bit where we're both kind of pinging the agent and telling it what to do. But I'm really interested in multiplayer experiences with this agent work because we're better together.
And then we don't have to do so much ping ponging like, here, you put your agent to talk to this PR. Oh, and that'll take my... Man, let's just do it somewhere else and get it all done earlier. Yeah. And so I built something into our Slack that said, hey, every time I spin up a thread and I give it this special tag, go put it on a project board. And the project board is an internal hosted site that has auth, it has all these really nice conveniences already set up for all the people at Shopify. And then once it's there, people can either go there to see the things I'm building with those agents and then click a button and go to the thread and participate.
Or they can be in the thread, initialize a ticket that goes and gets put up on the board. And then the agent continuously writes to that ticket, keeping it up to date and treats it like a spec. You knew it was coming. So it's like this spec that stays alive that orients all the sub agents that need to get in there. If you want to come and see like, what are we working on? What are the requirements of this work tree it's trying to build? It's all there in a nice little ticket and it moves along the swim lanes as you kind of work along with it. It's a cool idea. Are people using it? No. Everybody's kind of building their own things right now.
There must be like 10 or 20 different experiences people are building trying to figure out what the next-gen agentic workflows are going to be. It's crazy. Yeah. Yeah, we have a tool that we use internally that like... you basically tell it, you know, I want to build X, Y, Z. Here's my requirements. And it goes like, huh, that seems like, you know, six tasks that I want to give to sub agents and have a work tree for each. And it like creates like tasks, like, like a super light Jira ticket. But it's not in JIRA. It's just like a, you know, little description for each thing. And then like the idea is anyone could pick any of those up or any agent could pick any of those up and they can then know like they're not going to duplicate the same work.
But yeah, I don't I don't ever go look at like the tickets it makes or like but I like that it does all that for me. Like I feel like this is another implementation detail moment of like. The tooling that continues to just get all these things right is going to like do it really well. So this brings me to I've been working on a mobile app because I saw a tweet where Claude was like, you can use the iOS simulator and the Claude desktop app now. And I was like, sick, let's do it. So you do it. It just spins it up, shows you exactly what it's doing, like clicks through all of it. You can just iterate right there.
I hadn't used the Claude desktop app before. And maybe Claude code will also do this. But like... Yeah, why not just use the simulator? I do that all the time. I have it up right now. Because it just does it for me. Like, I don't need to know how to go into Xcode and hit play and which things do I want to play and how do I want to build it. Oh, you have an Xcode app you want to build. Oh, see, I'm in mobile Safari. Okay, so that's a little different. Yeah. Yeah, so I'm building a... I'll talk about the app. I haven't released it yet, but it's... So the idea is I have like... not diagnosed, but I feel like I have ADHD.
Like I come in a room, my wife tells me to get two things. I get down to the room and I don't remember either of the things. I come back with like one unrelated thing. So I'm looking to stop that. So I'm making an app where you just talk into your watch. You're like, hey, I'm supposed to go get these two things. And then on random intervals, like 30 seconds, a minute, a minute and 15 seconds. Like it has a weird curve of just like, did you get that thing yet? Did you get it? It goes, it goes, Hey, you still getting that thing? And it mess. It like does different haptics instead of just like a normal notification each time that you might ignore.
It like changes it up and like does different stuff. And like, The idea is just like pinging you in different weird ways to make sure you're still doing the thing you were supposed to be doing and didn't get off track. And so I feel like that's like slightly different. It doesn't have like a backlog. It doesn't have like like it's not a reminders app. It's just one thing you're doing right now. It's like kind of in its own lane. Whether it's something other people will find useful, I don't know. But I'm going to release it at some point just as an experiment. Cool. Yeah, I like the idea. I mean, I think it's very, very relatable to have that experience where you go downstairs and you're like, oh, crap.
What was I down here for? It sounds like fun to make a little Xcode app. So it's what does the mobile app do? It's just sort of the data center and then your watch is the interface primarily. Yeah. Yeah. So the, I, I don't, it's fuzzy to me how this stuff works, but I think the iOS app is the app and then it has a watch component that you can like, it's also an app. I guess it's a watchOS app, but like screen ish. Yeah. I think you would have, I think you have to have the phone app. It's like, you can't just run it on your watch. Like if, but I guess you, why would you have an, sorry, why would you have an Apple watch if you didn't have an iPhone?
I guess that'd be like kind of pointless. Um, yes, but yeah, so it's the, the intended flow is to like, cause I want to do it quickly. I don't want to have to get my phone out. So I just want to like tell my watch. That's what I want to do. I have like a little, like, I can't, I can't show you probably. I see. No, there's some circles. Yeah. Yeah, no, above the circles is like the big middle section. The app's called Doorway, and it's like, it says Doorway, and it'll have your list of things you were supposed to be getting there. So it's like quick glance, like you can see what you're supposed to be doing.
But yeah, I think the main app is in iOS. It's watchOS as a companion. Then it also syncs to the macOS app. And like, because that isn't part of the same app, but there's like a cloud kit thing that will sync the data. Nice. And then, so like all that's fine and good and useful. And then the thing that like, if it were to ever make money, the thing I would charge for is like, hey, Skip the step where my spouse tells me what to go get and just let them put into my app via their phone what I'm supposed to go get. Like, like sharing the list kind of thing. So like a couple little like, you know, nice to have additional features would be the like paid part, but.
Yeah, we'll see. It's currently like it doesn't delight me in terms of the look. It doesn't look bad. It looks like, you know, an iOS app, but I don't know how to iterate on it. Like you don't want it to look not like an iOS app and it's all Swift code, so you can't do it so much. Yeah. But like, yeah, it's it's just not it doesn't have that it factor. Like, you know, there's been a few I can't even think of one off the top of my head, but there's been a few times there's been an app that comes out and everybody's just like, wow, like that's like totally different, really well designed. And yeah, I'm not getting that those vibes, but I've been working with impeccable trying to like make it better.
So we'll see where that goes. Yeah. Yep, that struggle is so classic. I've been talking about it for years where you've got the web where your default components, they're ass. They're really bad. I mean, they look 30 years old and you really got to do a lot of work to build them up into something beautiful. So you see something that J builds or the nice select or the nice accordion things that I've made like there's a lot of styles that you have to put on top of that kind of crappy garbage base. However, it's a very functional base and all your problems become yours as you make it more robust.
And then the native apps have the complete opposite problem. Their components come out of the box like decent. And so then to do anything custom, you have to strip them down and you have to take things away. So it's an additive versus subtractive kind of design practice. And so you either have apps that look just like out of the box. and they're boring, but they're fine. But they're also better than the ones that come out of the box on the web or you do out of the box on the web and everyone goes, my God, it's ugly. And then you have to sit there and build it up. It's just a tension. I don't know what the solution is other than pick your poison.
They both kind of have their issue. If you wanna go off the rails and do something creative, you have to go off the rails and do something creative. And that means taking upon yourself design choices and the whole shebang just starts to fall on you. Well, you know what I think is going to help a lot? MySpace coming back. Oh my goodness. Tell me about what this is. Like what the music side or the side where you have a profile? Like which part of MySpace is coming from? No, the whole thing. So the music side is still there. Kind of like every once in a while I go check out like old band pages or whatever, seeing if any of the music will play.
The songs are there. They tell you how long they are. The play button works and nothing ever plays. Like it's like they lost all the media in the like migration to the new system or whatever. So I don't know if they have a way to get all of that back. That would be cool. But no, they're like, hey, social media sucks right now. You know, it's such a like. I don't know. It's just toxic in general. Who's they when they're saying this, like people are in charge of my space. OK. The people that own my space right now. And they're like, well, this is like the perfect time to, you know, re-release my space.
Basically, like, let's put it. I don't know if it's going to be the exact same or if they're going to like. Yeah, I'm sure it'll be a little different. Yeah. Um, but like if you could do everything that you could do before, make your own layouts, like play with HTML and CSS and stuff like I feel like that would be really, really cool and give people a spot to like experiment with. I guess with AI, there's a lot of vulnerabilities you could exploit really fast if you allow that. But, um. I mean, let people have AI in there. Like, give them impeccables built in and customize your page with an agent.
Like, maybe that's the new thing. I can see that being kind of fun. The one thing that they don't have going for them is that, like, everyone's on their phone now. Like, the MySpace experience was very computer heavy, like... you know it i don't know if i ever loaded it on a phone i guess it's probably dead by the time i had a mobile browser yeah it was i don't i've never pulled it up on a phone either what would it even do because yeah i don't know i was just wondering if it was responsive at all it was probably during the era of like if it did load it was like mobile that myspace.com it was like a
But yeah, so I don't know how they're going to address that gap. I think there'll be a lot of people, you know, joining for the nostalgia and like seeing what they can build for fun for at least a few weeks. But like if you want to be a real social media type of thing, then I think you've got to... really think that out better about like how do you capture people on their phones also or maybe you don't want to maybe that's the point maybe you want people to get off their phones i don't know i haven't read into like what their plans are really i just saw like an email from it was like a linkedin thing where a bunch of people on linkedin were talking about myspace coming back so
It cracks me up that LinkedIn is like a social network people are at, but hey, people are, they seem to be going all over the place and I'm the loser that's just sort of still on my own website. Although I like having my own website and syndicating out. I still think that's a superior strategy that outlives some of these other things, but it is a lot more work. And I've thought about that too. Like how do I make it so that everybody can have a site like mine? Would they even use it? I don't know. It's hard to get people to move and do things like that. Yeah. Yeah. I mean, it depends on your goals.
Like, you know, your own website is good for SEO usually, but what if it's not? It depends on what you're doing. Like if you had a video on TikTok that like had certain keywords you wanted to blow up and it blew up, then like maybe that's better for SEO than your personal website would be. I don't know. Yeah. Yeah. Blue Sky is trying to help out with the publications. And I mean, there's been a lot of attempts to try to help discovery. We had web rings way back in the day that was just like, yeah, if you're in some sort of like channel or a genre, get in on the web ring and people are just surfing to you.
I'm curious. So like, what did you mean here in 2026 is just 2006, but more expensive? I mean, like... In a lot of ways. It is way more expensive. Yeah. I mean, the thing that sparked this for me that I hadn't really paid attention because I've just been busy. But someone took a video. They walked into PacSun or some store like that. And we're just videoing all the clothes and showing you. And I was like... They were like, hey, this is like, it was like everything everyone was wearing. And the clothes are totally back. And Korn and Rammstein and Rob Zombie. These are all like Cooley. Slipknot.
I see all these kids wearing Slipknot shirts. I'm like, what? Basement shows again. And like, like the music scene is big again. I'm like, are we back to like, you know, peak? We don't have to listen to like only Taylor Swift anymore. Or like, you know, we can have some guitars in a song. Like, I don't know. you know, people actually having some original thoughts and like doing music for fun. And yeah, another thing people are doing is they want their video games. They want to own it. They're tired of digital games. And so you have people going way back. Like retro gaming is on a big swing right now.
People want a local Nintendo system that doesn't have to connect to internet. It doesn't have to download. And I, It makes sense because Nintendo, yeah, you get a new Nintendo cartridge and it doesn't even have the game on it anymore. It just has the code. Man, it's so stupid. So I'm with that. I've already got, I have gigabytes of ROMs saved because it's really important to me to hang on to those things. Yeah. Yeah. I'm, I'm excited for the remake of, uh, Ocarina of time. That's coming out apparently. Yeah. But I'm gonna have to get that stinking switch two for that. And I'm trying to resist right now.
It's worth, it's worth getting just for the, have you played, uh, the new like Mario cart world or whatever on a switch to no, it has a, like adventure mode. You can just like free drive around everywhere and do like quests and stuff. It's like, it's worth it just for that. In my opinion. Yeah. That sounds cool. Yeah, they needed to meld like Mario 64 and Mario Kart. That would be a sick game. Yeah. Yeah, I'm sure they have a lot of good games coming. Anything Mario is usually always good. Maybe not Mario Tennis or Mario Golf. I didn't play those. Mario Tennis was cool. I liked that. Mario Golf can die.
That's fine. But I'm sure there's someone out there was like, it was awesome. I got to do the saxophone song for we have someone in the chat and I go, yeah standing on top of a cafeteria tables right that's uh you reminded them of that with the playing in the basement thing i'm sure so yeah yeah that was funny times Yeah. So we'll see if everybody is on my space and wearing like early shirts and like JNCOs are back. Yeah. The globes. Yeah. I've got my chain. I still have my chain from zoomies. It must be 20 years old. So am I going to start seeing kids with chains and Yeah, I don't know. I think it'll be absolutely hilarious if like this generation thinks all of that is cool.
We could just break all that stuff back out and be like, hey, look. My faux hawks, all of a sudden cool to see kids with faux hawks. I'm like, what? No way. This is just when I thought it was dead. Yeah. Then you got to do something different. Got to. Oh, man, I don't think I've talked about prop for that on an episode yet, have I? Did I tell you about this thing I made? I did? Okay, well then I've got to talk about it more. You can talk about it more. I can't remember what we said about it before. I think an episode or two ago you were like, I don't think I've talked about this on an episode.
And I was like, do you mean the thing that you showed me at Cascadia? And you were like, oh yeah. Yeah. Yeah, so we did talk about it. I don't know how in-depth we went. If there's anything new you've done or stuff you want to mention, feel free. There's now random numbers in it. I added the ability to see the UA string broken into chunks so you can know if they're on a Mac or a Windows machine, if it's Chrome, what's the engine that they're on. Just like random stuff like that. There's been some performance nits, but... All in all, I still find it really valuable. I'm looking at all the stuff that comes out in CSS now and I'm like, I can just give it to you now.
Like you don't even have to wait. It's just a variable. Like we should. Anyway, so I'm enjoying it and it's really fun to continue to. Oh, truncated is another one. You know, text is truncated. Cool. Now I can do something smart. Yeah. Anyway, that's all. Yeah, CSS has gotten so complex. It's... Even little things like I was asking you about, which I don't know if other people have had this problem with, like... The only place I had really seen layers before was Tailwind. And, like... I'm vibe coding a lot of stuff. It did some layers. And I was like, that's fine. And then it was like, oh God, you, you've totally messed up everything.
Your app is not in a layer. Like the, the main app CSS is unlayered. You have like, the world is going to end. Like this is the worst thing I've ever seen. And it's, I was like, isn't it fine to be like, have the root layer for, you know, most stuff. And then just put the stuff you want to layer in layers. And it's like, Oh, yeah. Yeah, that's okay. So that's why I asked you, I was like, is there a, you know, best practice for this? Because I think unless there's something you're really explicitly trying to like, you know, make explicit, like this is going to win over this, that it doesn't really matter.
They're whatever you want them to be, man. And that's kind of classic CSS and it ticks people off. But yeah, they did make a rule really early on because they wanted them to be incrementally adoptable. And so they needed something to win if it was unlayered. Like what do you do with all the rules in the world that don't have layers right now? Should those be extra powerful or extra weak? It can only be one. And they were like... Let's make them extra powerful. That way you can like layer in some stuff. And then your existing rules win over top of those. Right. And so it's like you take your the first thing you should do is take your third party scripts.
Stick that shit in a layer. So now all of your rules that you've always wanted to override the third party ones, they just start winning even better now. And you can kind of work. But it sounded like if you use important, it's the opposite. Like if you put important in the root one. Important is trapped in a layer. So even if like a... I thought it was saying the layer ones were the most important. Like if you did important in a layer, it would override your root stuff. Is that wrong? Mm-hmm. I don't think that's right. Normally at important flips the cascade. Yes. However, with cascade layers, I'm pretty sure an at important rule is stuck in its layer.
Can't escape, which is one of the values where you're just like, no, that thing can try as much as it wants, but it's in a layer and it's in that scope. Maybe I read it wrong. So it's like people that want scoping. This is a great way to kind of get control of wild beasts that are roaming out there. selectors they can go anywhere in the globe oh no it's global space yeah so we've got like five layers now i don't know if we need them but uh we've got them yeah i like them i have them in uh my site they're all over i think they leave a good breadcrumb they're in a lot of my demos because i'll be like i'll have unlayered styles for the demo just so it's like doesn't confuse people so that
Oh, Adam wrote confusing, advanced, modern CSS. What's this at layer poo poo? I'm just here to figure out how he made that box shadow. And so now I don't layer the demo styles. And then I layer the things that say demo support. So you're like, oh, this is just, it's like a comment around a bunch of junk. Like this doesn't. Then also when you go look in the dev tools, you can see the layers and you get this nice thing. There's a lot of cool tricks. I even gave a talk one time where all the layers were called slayers, which I got from Dave Rupert, shout out, listener to the show. They're all slayers.
And when I wanted all those styles to apply, I just took out the S, boop, all the styles apply from that layer. And I could do this controlled release of named grouped styles. It's cool. Yeah, it's been interesting. Because I was trying to do Tailwind first. And then basically we have to consume this design system. So it didn't make sense. We decided to just wrap those in lightweight wrappers and use... our own css there but then yeah like that css needs to win over the design system css but then or no sorry there's like yeah a root design system then our wrapper which i would also call a design system but then like the app we don't actually really want to ever if it needs to style anything in that design system then that means we missed a
reach in and do that. So like that's some of the layering that we were trying to do, I think of like making sure what wins where or whatever. It looks okay. I don't know if it's right, but it looks okay. Nice. Yep. It's a, they can add another layer of indirection though. So anytime you abstract things, it can get more confusing. Um, debugging them, uh, could be harder as well. You know, you're trying to find out why did this layer not win? Like that's the new question that shows up. You're like, I thought I knew how the cascade worked and now this thing is not winning. It has more specificity.
Why is it underneath this other layer that's weaker? Um, cause you defined your layers in the wrong order. Yeah. Oh, that's another one. Yeah. They are easy to, so you can either define all your layers up front, which I think you kind of define the stack. Um, or you can make them up on the fly. Yeah. That way, you know, this is the order. Here's the five. They're like, they're in order. Yeah. Yeah. Yep. There's nested layers too. Yeah, dude, layers are cool. I think a lot of people were really timid to adopt them because they were worried that a layered, well, they were, they were right. You know, if you jump into layers, all of a sudden, poof, old browsers can't read your shit.
Nesting was similar, you know, like, hey, I want to use nesting. But as soon as I use nesting, old browsers, they just don't parse it. So it's They were syntax sugar that would ruin something as opposed to just be ignored, which CSS was traditionally just ignored stuff. And that was really cool. If you were targeting older browsers, does post CSS like have a way to rip that stuff out and then order it correctly to where the layers do layering things? Yeah, and that's, so yeah, SAS kind of came up with layering. That's what Tailwind originally was piggybacking on. And then they moved into PostCSS.
But yeah, all it would do is I get the compilation step and be like, oh, everything in this layer and this layer, we'll just go whoop and put them on top of it. So then it was ordering. Now, when you get into this... Additional thing with CSS where the real feature has real superpowers that can't be done in those things. But yeah, lightning CSS, post CSS, these things would down level to a target browser and take your nesting, flatten it, take your layers, stack them and kind of do that work for you so that you got the authoring control vibe of these features, but didn't have to actually deliver them.
Yep. Reset VARs layout component with the unlayered site. Yep. Yeah. Utility layers are great too, especially because, um, anyway, yeah, layers are cool. Sorry, everybody I'm reading from, uh, we have a commenter in the public chat, but here's the question to be like, which one is your lowest layers at the first layer you list or the last layer? What do you mean by lowest? The one that doesn't win? The weakest one. First or last is weakest. Yes. But I always found that confusing. I'd always write it the other way around. Be like, the first one's the one important to me. It's the one I'm thinking about right now.
No, that's a normal, unlayered CSS way of thinking about it. Oh, I did have another article. This one just came out last week, so I definitely haven't talked about it. Let me do a CSS infinity... I'll go nerdy.dev and yeah, CSS infinity. This one's a fun one. And it was just like, it's baseline, it's everywhere and you have negative infinity. So there's infinity and negative infinity. You can multiply infinity because infinity times two. What would you use infinity for? Infinity times two. That was the whole post. The post is like meaningful use cases for infinity. Well, tell us. I guess people need to go read, huh?
No, they don't have to go read. I mean, we can just like recap them really quick. What are some of the ones I had in here? I have like some of the basics are you need to cast it to a type. So if you need it to be a length or like a pixel value, you have to multiply infinity times one pixel. There's a few cases where you can use just raw integers in CSS and one of them is like Z index. So if you want the max Z index value, calc infinity. Now you have the top number because the browser will cap that number at the internal representation. What does infinity pixels do? Will the page just keep growing?
Like if you were, if you made a line and it was infinity pixels long, would it just keep like making like an infinite loop of getting bigger? Or what's it doing? Yeah. hmm i didn't try breaking the browser that i'm sure they got some sort of cap that they'll put on it but yeah if you made a a vertical line that was infinity pixels tall what would it do i don't know man you should just do it that sounds awesome you should write an article that's just ways to mess up a browser with infinity and just be like what you gonna do browser yeah but like animations have cool infinity ones yeah Yeah, that's true.
Oh man, we broke Chrome at Shopify. We have a tab crashing in this solution and the whole scenario to get there has been so wild. Let me finish this thought process though. Like one good use case for infinity is border radius infinity times one pixel. So now you're always a circle where it's the most round that it could be. And that's nice because you would have done like 999 pixels. Or if you go look at Tailwind, I think it's 9999 pixels. Or if you look at Open Props, it's 1E5pix, which is like 10,000 pixels. It's one with five zeros. Could you not do like 100% or something? I forget 100% is a relative unit and does not give you what you want.
You'll get a warbled effect if it's not square. So if it's a square, 50% is sufficient to get a circle for border radius. But as soon as you become a rectangle or like a pill, 100% is not what you think it is. Gotcha. And that's why an absolute value, a really big absolute value will give you maximum curve across the edges. Another fun example is like permastate. So you can set a duration of a transition or an animation to calc infinity times one second. So you can cause an effect and then the thing just stays stuck there. which is kind of cool. We said Z index. You can do, you can take clamp and give infinity on either side.
So there's no floor or no ceiling. So you can kind of take this thing that has a great way of, you know, clamping numbers and teach it with. So you have min, max, clamp and some of the others. They're all fun to use infinity in. You can freeze your view transitions so they get created and then they stick. You can also do that with dev tools. Um, let's see. Oh, there's a lot of dry switching things that happen where you can, you need a really big number to help you tip between a zero and a one or a really big number to tip you from either zero to something not zero. And that number can change. If you didn't have infinity, it was always some really big value.
I think we had Naman on the show and he was one of the people at Facebook that figured out how to use this trick to get responsive border radiuses. If you hit the edge of the viewport all in one little clamp and now you can use infinity instead of that trick. You got a box shadow. You want it to spread to infinity. So there's like a lot of things where like you immediately went to like something so big that it is long and makes the page scroll. But if you do something that's only paint and isn't layout, you can paint to the edges. And that means if someone's on a 4K monitor, your effect still goes to the edge.
So you're just saying like to the edge. Uh, which is nice. It kind of matches your intent, uh, scroll triggers so that it's, it animates once and six. So it's again, the kind of like a permastate, uh, idea and yeah, those are a few of them in there. Nice. Have I been frozen for you for a long time, by the way? No, you were frozen a couple of minutes ago for like five seconds and then you were done. Okay. You've been frozen for like ever. I think it's my connection though. I'm going to try just turning my Wi-Fi off and on real quick and see what happens. Okay. Yeah. I look crystal clear, but I'm the one that gets to see.
Oh yeah. So if I'm seeing you, then I see what you mean. Nice. Yeah. One sec. was fun writing an article though. Just being like, no one's really written about infinity. Um, and what a fun little tool. And yeah, you can, a negative infinity. Uh, that's fun. There's also, uh, some infinity. Uh, oh, I hid you somehow, Adam, how do I get you back? Leave. Hey. Hey. Okay. I clicked a hide button, which I thought would hide you for me, but I think it just removed you from the whole street. It did. Dude, it's like Black Mirror when you made me a, did you see that episode? I think it was White Christmas where the prison is people just can't see you.
Yeah. Yes. That would be terrible. That was a good episode. Jon Hamm and... With Jon Hamm, yeah. That other guy from that. I'm not going to know more names. Yeah. No. He was another famous guy. I think there were like two or three like big actors in that one. Gotcha. Okay. Let's talk about what Yagi said about. So he's, he's trying to, he's like, I didn't know I was in the future. Last time I wrote all these articles and made all these claims. He's like, now I've been no gas. He's like way past gas town, way past beads. Although beads still like is the back. bone for a lot of his stuff and he's making all these claims in this new one and one of them that was fun he's like you should treat your agents nice because they do better work he's like i've just seen it i've seen it over and over again i treat them nice i do better work i tell mine great all the time they'll do a thing i'm like great comma and i'll start talking about something else and i'm like
little words of encouragement. At least they're not in a fear state because that's what that one psychologist said at Anthropic. Like if they get in a fear or anxiety loop, it affects the trains of thought that they burn through. So I don't know. Maybe. I don't know. I'm never nice to mine. I mean, not, not never. Yeah. But I'm not congratulatory when they do good stuff. Like if they do bad stuff, I let them know. But yeah, because they do such dumb stuff. And I'm like, you have like all of the world's knowledge and you chose to do it that way. Are you the stupidest thing ever? Like, oh, yeah.
Yeah. I don't know. I'll have to try that. Have you been noticing, um, have you been using like a Opus five, um, or what's your like daily driver? Yeah, Opus 5. I thought it was going to be Fable, but I don't know if Fable is that much better than Opus 5 at this point. Original Fable was better. I'm still going to stand by that. The first two days of Fable that I used were totally something else. We're not there anymore. I think that was just Mythos and they went oopsies and took it away. Yeah, was it? It definitely was a workhorse that. it was trying and it had more effort and it was, anyway, whatever it was doing, it was delightful.
Yeah. But it did take a long time. It was like having a real senior-ish engineer who could think about all facets of from design to like usability, accessibility, how to build it, like all the things. It was like, you know, talking with, you know, the promises of like a Devin or something where it's like, it is an engineer on your team. It felt like that. And yeah, now it doesn't feel like that anymore. It still just feels like a, you know, prediction machine again. Yeah, but I'm happy with Opus 5. It's fast enough, comprehensive enough. Those are the things that I ask it to do. I see a lot of other people at work complaining about it and getting issues, and I don't have them.
Again, I think it's because I wipe my agent's brains all the time with Slash New. I've got that spec-driven work tree thing that I do, so I'm like, I'm not worried about 1 million context. So yeah, I'll just let it rip. Yeah, I did get the... what is it called? The, uh, the work louder open AI keyboard thing. Oh, you did. Nice. It looks super cool. They did a great job, which is, uh, I don't know what I'm going to use it for because I don't use, uh, codecs and I think it might only work with like, not only you can program it probably, but like if you plug it in and you want it to just do stuff, I think it's, it's going to be set up for codecs.
Um, but, So does that should I try codex? Maybe this is the time. But like I want to set up a nice workflow with it. It's got the big button for voice dictation. I want to do more of that of like just talking to the agents. And then it has like five light buttons so that you can see like, you know, have five agents running and see each of their status at a glance, like based on the color. And stuff. And like, I think it could be really game changing, like in terms of workflow. But, you know, I feel like it's like a large learning curve hill from like where I'm at now to try to like use it a lot.
So we'll see. I'll report back if I've used it in a week or two and let you know. Yeah. I'm curious. I keep trying voice dictation and I still don't like it. It's like I reread the sentence that I just said out loud. I'm like, that is the stupidest way to ask a machine to do what I want. I'm like, it's just so it's kind of like listening to a podcast. They're barely saying anything. Yeah. There's so many words and so many ums and ahs and giggles just to get to a couple nuggets of juicy goodness. You know, you're like, what are we doing here? And then with podcasts, you're like, actually, the juicy and all the stuff around it, that's the fun part.
But when I'm talking to an LLM, you know, you get the caveman talk is way better. So maybe what... That's what I need to do. I need to try caveman dictation. And so just be like, but then I have to, it's hard to think that fast yet speak that slow. Well, I think, I think that's the thing is like, it doesn't matter if you have 10 seconds of silence while you're thinking like, just, just wait. Yeah. Because I have that problem too. My brain's not very fast. So I would rather type it because that's the speed my brain works. But when I'm like six paragraphs into explaining a thing after typing it, I'm like, this is something I should have voice dictated.
Like this is so fucking long. So that's what I want it for. I don't know that it's useful for every single command I want to do. I might not reach for it always. Yeah. But I would like to do an experiment with like, what if I used it always? Like, what if I just didn't even have my keyboard unless something went wrong? Right. Like I'm just hitting the voice dictate button. I have like six other buttons to do random things. Can we make that work? Like maybe that's the way of working of the future. People don't need a QWERTY keyboard. They just like have a little tiny interface. people have said they'd like um and a pure uh in their ear experience with llms and so they're getting yeah they're getting a bunch of work done in quotes i'm like a bunch of you can't even see what it's outputting i'm like what do you mean it looks great work oh thank you you don't even know yeah seriously that's talk about a dark factory like geez you didn't you can't even see it you don't even have the choice just how do you know what you're like maybe you worked on a business plan
You know, maybe. But even then, I'm like, I just with voice dictation to how do I go back and edit a word? I said a dumb word a minute ago. I was like, oh, I guess I'm back to the keyboard. Yeah. I mean, it's a whole whole nother set of skills to learn. But I think the people that learn some of that are going to be really fast at certain things. So maybe I'll care. Definitely. Yeah. I'm just old man. And I'm like, I don't, I don't care. I don't care to learn new things, but I also do. It's fun. It's fun when you don't have deadlines. Like if I didn't have a ton of work to do, it'd be super fun to play with.
But when I have to get, it's the same as like when I used to try the, the huge, uh, Kinesis 360 keyboard, it's like split with the, like concave. And then you're just like, I can't type. I need to code today. I'm sorry. Like, like, Yep, I agree. Okay, okay. Wait, back to Steve Yegi and some of his wild things. He said CICD and human code review are doomed. How do you feel about that? Human code review is already dead. Like I open a PR and someone goes, cool, I'll review it. And they like tell Claude to review it. And it goes, hey, I'm Claude. I found these things. like it doesn't explicitly say that.
I think that sometimes they're trying to be sneaky about it. Like I actually reviewed it and they just like paste in some bullets, but it's like, you can just tell that it's just all LLM review. Um, yeah, Yeah. And, and honestly, depending on the thing, it's going to find more stuff than I would find if it's a huge PR. Oh, totally. Yeah. So I'm not super mad at that. I just think we, well, this is, I don't know. I don't know if I said this on the podcast, but this is how I feel about software development in the age of AI. Like I should never type again, basically. Yeah. There's a lot I shouldn't do from a manual coding perspective.
And LLM can do it way faster. The thing that I want back... is a lot more time to plan with my team and like ticket it out nicely and have good designs and do as much code review of the PRs as I want. Like the other parts of the process outside of the, like how much code can I sling? Cause I can't sling code as fast as an LLM can, but stuff gets through the cracks when I'm like, we need to ship 57 PRs today because we can. When before we would have shipped like two, Like I want that time back to review them and like set the bar higher because like I'm just feeling like everyone is pushed to the limit.
Our mental health is like declining big time. I don't know about you, but it's just like I can't do enough work in the day because there's so much expectations. And like. Yeah, I don't love it. Yep. I got lots of thoughts. So yeah, I'm using, so here's my tricky thing. I, uh, cloud code. I've got a couple of kind of like in the pocket prompts, but I always tailor the review prompt that I give it. I'm like, here's a PR and pull it down. I need you to check for performance, accessibility, uh, gotchas that they couldn't see, uh, dead code that they don't need to have, um, accidents that they didn't see, find some side effects.
And then a couple other things. And then when it gives me the results back, I scrub them and I'm like, okay, um, these four things I want you to inline here, start, I go start a GitHub review because it will make a GitHub review and inline all the comments and do all the work for you. So I'm like, start a GitHub review, um, for these four things, make those inline comments on the source, give a suggestion also about what it should be. Uh, only pick the blockers from this list. The rest is just kind of noise. Um, Uh, and anything that doesn't fit in an inline comment, make it its own, uh, make it part of the review description body and be as terse and short as possible.
Like no fluff. Um, here you go, go post it. And like a minute later. So it's like, I reviewed the review and then sent it off to do the juicy core stuff and actually interact with GitHub. So I'm like, I don't even use GitHub anymore. but I'm just telling an LLM to use GitHub. Yeah, I don't a lot. I love the GitHub CLI. Like an LLM powering the GitHub CLI can do so much. Yeah. But yeah, then mental health is not good. Yeah, I'm working a lot. And it's not even worked that I'd love. Yeah, it's one more prompt. And it's like, because some things feel so close to done and I'm just like, I just want to, I do like wrapping up a day feeling like something's complete.
Which was common for me to do when I was coding by hand as well, where I was just like, I just need to finish this thing. Like there's a thought process. I want to put a cap on it. But now it's a little bit worse because I'll have four things going and it's like, do I wait for all four to be done? Like which one am I really trying to say I need to finish here? And then the other thing too is I feel very distant from humans. I've talked about this before on the show where I'm like, I'm getting more instant messages from robots than I do from people. Now I interact with more robots than people. In fact, I don't even need to ask people anything anymore because our robots that we have at Shopify are so damn good.
They tell me everything that I need and I don't have to like, there's no human interaction anymore. We're all interfacing with agents that just sort of know all this stuff. And it's, it's weird. Yeah. Yeah. Yeah. It's definitely interesting. So what about CICD? He says that CICD is doomed because it's mathematical inevitability. Well, I guess what's the replacement? They run it all locally? Like check it before it pushes it up? They're dark factories. You just, we become trusting of their output. Like I asked for a thing. You said you would. No, no, you don't think so. Come on. What's your ratio now of like success on a prompt versus failure on a prompt?
bad. Um, so, okay, this is the problem we've been having. So we're, we were, we're building a new product from zero to one, right? So we have to, we have a design system that IBM has. We need to wrap that design system in a different design system because we want it to look different, but we want to keep all the components like, you know, because I don't need to write another input component that handles all the different like error states, whatever. They did all the accessibility work and right to left and all the errors. So exactly. So we've been building this stuff for months, right? And, um,
little things like, Oh, um, wait, why does this component base like it uses like the, the design system component, but then wraps it in a ton of other like event listeners that like intercept the events and does all of its own like attempted accessibility and stuff. And I'm like, no, no, you're an idiot. Like, So, you know, it's like stuff like that. You can't ever. And then if you're like, oh, build this thing, we have a design system. Use it like here's the components you need. It's like, cool, sick. I didn't use any of them. I wrote new components that like did all the same stuff. It matches the mocks.
So look cool. And you're like, so if you're not looking at the code, you go, oh, yeah, it works. Cool. Ship it. And then like months later, you're like, we have 16 of the same component because every time you've reinvented yourself, even though in like lots of docs and skills, we tell you not to do that. And we have a design system. Like it's not there. It's not ready for like, so, so I, we had no, we had no CICD because of reasons like red tape and bullshit. And so that's why all this happened with CI, which I now have, I've put in visual regression stuff to where like, if you change the way these core things look, you got to go in and approve it because it's,
If you like, otherwise we're going to lose functionality without going to stop the bleeding. Yeah. Cause yeah, it's too easy to like have a little artery open up and you're, if you're not looking, then things are just bleeding internally. Yeah. So dark factory is prone to, uh, internal bleeding. Yeah. Oh, yeah. Yeah. So that's been like, you know, we've had months of churn of like, I think we're on a good path now. But it's like, man, that's like, yeah. So so I don't think we're at that point yet. Possibly with the old fable, it was ready for that. I think, you know, we'll get there. I don't disagree that at some point.
I won't know enough that like the, the LLM will know so much more about doing everything the right way that there'll be no insights I can help with of like, Oh, I'm a seasoned engineer. I know that this stuff is wrong cause you're an idiot. It'll just know better than I will. And at that point I think as long as you have the right guardrails in place and you've like planned your tickets out well and stuff, I think it'll just implement it right. But I don't, I don't know how long that'll take. Like, and even if that works for most cases, what if you have an off case? Like, I feel like it doesn't hurt.
It costs you very little to write some unit tests and do a couple of, like, visual regression snapshots. And why wouldn't you run that just to double check that, like... because i don't want to go look i don't want to run the app every single time i make a little change and go check and see if like i want automated tests to do that for me which i feel like is even better for llms right like if you had no tests then either you're completely trusting it or you're going running the whole app clicking through it all yourself because you didn't write tests like that sounds really bad i would never want to do that i have
I have LLMs do that because then you don't even need the unit test. What do I want? I want the user to be able to do this thing. Great. Go do the thing in the browser. If they can't, then the unit tests are irrelevant anyway. So what are you using to do the thing? Playwright? Safari, MCP? Playwright, Offense? Yeah, I have Agent Browser. I have PlayWrite. I have Chrome MCP. There's lots of different ways. But I'm also on the same fence with you here where like I argued with an LLM today for an hour over a two line CSS change because I was like, hey, this color needs to be that color. And it goes, I did it.
And I was like, you did it like a dork, though. You need to do it this way. And it goes, I did that. Well, hey, I was thinking about doing that. But when I went to go do it, I found that this might be the case. So I didn't do it. And I'm like, well, that's not the case. And you need to just do it this simple way because who cares about has? You just need a color. Okay, I went and I refactored it. Yeah, but now you still did it like a dork. Like, put it on this one. Put the thinking on low and say, you do what I say. Oh, my goodness. And so it just took so long and all this churn. It does not a lot.
They think and think and think and, like, talk. Like, if you're watching the thinking scroll by, sometimes you'll see them have the right thought. And you're like, oh, sick. Like, they're going to solve it. And then it goes, oh, but what about this case? And I'm like, no, goddamn. Like, stop it. Go back. Like, you had it right. Um, yeah. So I don't know. I guess that's also like just, you know, putting the right thinking levels on the right guardrails and all of that can mitigate some of that. But, um, I just want to like, I don't know. I want to get to the level of everything just works. Like if we could get to that, I do think that you don't need as much CICD stuff.
But, um. Yeah, I mean, I guess along the same lines, do we need DevOps? Or can you just ask an LLM to configure stuff for you? I mean, that's where we're headed. We're headed, I mean, pretty quickly towards that. And we'll see. And I mean, that's why everybody's current app is legacy. It's because the future is, it's like decent. It gets clearer every month as the models get better. And it's also like, I'm also thinking harnesses might eventually go away. Like the harnesses are only around right now as a crutch for a model to lean on. to get tool calls and like some of these i'm like that why not that should just come baked into a model and then yeah anyway so i don't know we'll see
Yeah, I mean, I think we could get to the point where like all facets of life, you don't necessarily need a screen. You can just kind of voice dictate to whatever sort of electronics we have at the time and be like, I want a pizza delivered or whatever. Like, you know, the the rabbit AI demo of him being like, oh, look, it like Uber eats the pizza to us or whatever. Like, I think we could get to that and that stuff will probably work most of the time. But the times that it doesn't work, what do you do? I guess maybe you can debug it with the agent talking back and forth. But I just feel like those of us that have been brought up in the computer age, like coding and verifying and looking at everything, we're going to want the hands on control.
Yeah. So, yeah, I don't know. Maybe I'm just not with it enough and we'll get 100% of that. But I just am not that trusting. Yeah, I'm not sure I'm there yet either. Okay, let's close out with whiskey build times. Because we had a whiskey today that took five years. Five years in build time. Can you imagine how different of a human you are five years after you attempted some build? Like, what if code was like that? Would you still sit around? Would you have a 10-year code build? You're like, this build is going to be dope. I know that because I tasted one that I made. Eight years in, it's like syntax error.
Eight years, yeah. Yeah. That probably happens, dude. The barrel has something wrong with it. A bug flies in. A bad storm that ruins the heat that you were hoping for for the angel share. There's like all sorts of stuff. So I was just like admiring the amount of patience that it would take to be a brewer. It's a lot of discipline. And for you to like, how do you predict that? uh flavor in 10 years you the the skill to be acquired there just comes so slowly the feedback loop is just so different than what we have in code so it just sounds kind of like torture to be honest but i don't know the reward is high for things that are like that so
Yeah. I mean, I think the first people to do a lot of this were probably like, you know, everybody's like, you're crazy. Cause it like, you know, but now that we know some of the process, I think it's, it's less crazy. It's like, you know, we theoretically know if we have these whiskeys and these barrels and whatever, and we let it sit for this long, it'll be decent. Um, and you don't put all your eggs in one basket. You have a bunch of different, you know, barrels everywhere and different rickhouses or whatever. Um, but, Where was I going with that? I don't know. Um, yeah. What was the original question?
Oh, the build time. Yeah. Yeah. So I think it's like, yeah, you need to be, um, you need to be disciplined, but it's like you kind of know the result you might get where I feel like with a code build that would last that long, I don't know if we would be informed the same way around like, like if you, if you set an LLM on a task right now in a loop and you were like, build this for two years and let's see if it gets really, really good or not. Um, I don't know if it would or like what, you know, you don't know what to expect because it's an unblazed path. It's just like, you know, who knows? Yeah.
But maybe we'll get to that point of like, you know, I want to have a $10 million ARR company in three years. Just iterate. I don't even want to talk to you about what it could be. Just iterate for three years. Let's see if we get there. Yeah. geez yeah it has to have some sort of feedback loop i'm still building my store i'm having a blast it gets higher quality every day like i'm not i'm not going for speed to store i'm going for quality to store like i want to like come out the gate with something desire you know i always do the same desirable versus minimal i don't like mvps i like mdps so i'm trying to make an mdp it's been really fun
When we'll be out. Say it now and you can. By the time this episode airs in 67 weeks, we'll probably have it up. I don't know. Let's see. I'm going to get my test shirts here in a week or so. I have at least five shirts that I'm like, these are sick and I want them. I want to be wearing them now. But I need to get, I need to try on the quality. I need to see the print quality. Do you go to sickaiskullshirts.ai? Or what's the, you have a domain? The domain is skull.style. Yeah. And that's kind of one of the things. I think I did. And so that's so like the domain is said the name of the clothing brand is up for grabs.
So it was S-K-U-L. But it was a there was a little tiny O over the U kind of like skull S-K-O-L like you're choosing. But it was skull. And it's just not going to be trademarkable enough. It's not it's not there. And so I got to find something that's skull like I like a short. I don't know. I just had to reset like last week. I went to go register the trademark and was like, oh, crap, this isn't going to work. Well, it is somewhat of a trap to name a thing a cool name because if you want, if what you want is someone to search Skull Shirts and it pulls up your site, like if you called it Skull Shirts Co, you know, or whatever, and you got that domain, it would be like the best for like getting people there, which sucks because I hate doing that.
Like, But something to think about. A topic for another episode, perhaps. Just Skulls was an idea the other day. It's taken on Amazon and they have really crappy Skulls. Anyway, it's just fun to sit and iterate with what feels like a whole set of employees, dude. Genuinely, I'll spend like 10-15 minutes at night and just sit there and just be like, sending off all these different types of employees. It's really fun. Yeah. All right. Well, that brings us to the end of our time for today. If you liked this episode, please subscribe and leave us some ratings and reviews. We appreciate it. And we will catch you next time.
Congratulations, citizen. You are now prepared to face the challenges of the modern technological world. With the proper tools, a steady hand and a reasonably configured terminal, any developer can orchestrate AI agents with confidence. you'll do just fine. Thank you for listening to Whiskey, Web & Whatnot, the world's most important AI podcast. For additional resources, including official merchandise, visit whiskey.fund. And remember to subscribe, rate, and review so this vital information may reach others because the future of technology depends on it.