#34 Ralph Wiggum, Tool Search in Claude Code, Cowork, Apple Chooses Google!

and we're live. So, in case you're somewhere on Twitter, Twitter for us is down. Um, which includes half our articles, annoyingly. Give it a refresh. So, we are live. So, I've heard we're live on YouTube, but um presumably not live on Twitter. Just getting an X now. Um presumably not. I'm going to try um what is it called? The VPN. Let's see where Elon is. Is Is it Is it live? Yeah. Check in. Uh, San Francisco, wherever he is. See if it's live there. So, you you use a VPN, do you? Maybe that was not the smartest choice. Why? I I I use the VPN and I was gone for a second. Oh, I see. Now you're back. Wait. Uh, it seems like my X is working. Can you turn down the volume of your headphones because I can totally hear myself. My headphones? Uh, that's my mic probably being very sensitive. I've thought about getting a new mic. Uh, okay. So, turning on turning down the volume. We got three people. I still hear myself. How How is everybody doing? You are you people? Just do us a favor. Three people. Are you on Twitter? Cuz I can just see one person on YouTube. So there must be two of you on Twitter. Let us know where you are. Um, we can't check where you are because actually if you let us know on Twitter, we can't see you. We can't see if you respond. So, I guess if you're on YouTube, drop a comment in YouTube. Oh, Twitter's coming back. Twitter's back. Okay. I think I think the podcast kicked it into gear. I think this is it's not a podcast, is it? But you know what I mean. I think we were the ones that brought Twitter back. Okay, let me refresh all my pages which Yes, yes, yes. We're get we're in business. Just preparing all my pages. I'm also preparing here. Putting myself in the center of the the camera so you have an easier time editing as you said. Oops. Oh, yes. Yes, yes, please. Thank you. Thank you. Thank you. Am I turning am I changing color? Yeah, because of the touching the HDMI cable. Right, we're starting to see people pour in now. Hello everybody. Cheers for the like on YouTube. Drop us a love. Drop us a chat if you uh if you're hyped. Have you used that new hike button on YouTube at all yet? No. I'm too scared to press it. I don't know. I just see it on some videos that there's a hype button. Um, I'm too scared to press it in case it draws attention to my lurking. I I don't know. I maybe I haven't seen it. I mean, I live in Europe and Yeah. You don't get a lot in Europe, do you? Because you're scared of everything. I'm in Britain, so we're not much better. Well, we have, you know, those caps that you can't remove from bottles. Oh, yes. Oh, yes. Yes. Um, did you want to tell us about your site? Not site of the week, but your Hang on. Are you doing are you doing um site of the week or you doing preamble with your website? But we we could both uh do both. So, let's let's go. Uh let's see how the time allows. So, I actually have a few things to show uh like one thing that I when Twitter loaded, I noticed that I could include this. So, let's just start with that. Okay. You're going to start. I'm going to go get myself a beer. Okay. Go solo. Fly. Fly solo, butterfly. Fly solo. Okay. So, to everybody who's watching, have you seen this? You probably have. Are we Are we cooked? Well, so some people, let's say some people in a specific industry are probably cooked. Uh, you know what I'm probably talking about. People who might need to get a job. Well, that that's quite funny because that includes us as well. But I'm not talking about us software uh engineers and designers. Um, yeah, what I'm talking about is Yeah, this is like insane. What I don't know if this is real time or a recording probably not real time, but probably it will be real time if it's not. And then imagine you have a call with somebody like who can you trust? No images, no videos. And now this this is what I'm what I've been showing our audience. We are cooked. Well, I as I I I said not just not we because we don't need a job, a new new job because AI is not taking our job, right? What are you chatting about? AI taking some people's jobs. Oh god. You still beating that drum? No, that this one. This specific one. Have you seen this video? No. It's so impressive. Oh, I think I've seen something similar with like cartoon characters or something. But I mean, this is like realistic. Like really realistic. Look at the hair. That's real time, is it? Or is it? That's what I asked. I don't know. But if it's not, it will be soon. Like um This is This is insane. Maybe I can show you something that was disgusting to uh I don't know whether it's 3D actually. It could be 3D. There's nothing to suggest it's AI, but it looked very good. So, Le said the era of uh fake eager influencers has started. I mean, it started I was listening to a podcast earlier where they were talking about some movie stars that um are being released. I think the influencer stuff started probably pretty in the summer, right? I started to see a bunch of influencers come in influencers coming in around then. Yeah. Uhhuh. Is this real latency or pre-recorded? 8 minute for 30 seconds. So, is this recent then? Because of course Levels.io released this like one of the first ones. Is this like a new Oh, yeah. There he is in the bottom right. This is like So he's he's made it into a video thing. There he is. He looks fit as a girl. I'll send him this clip. It's crazy. Yeah, it's crazy how he's kept his shirt and everything like and he's kept his breasts. Well, I I I think the way it works, you can um you can create a character that you want and then you take a video of yourself and you say make this basically make me to it. I I don't know if this because I want to try it. It seems like I don't know. He said made with uh mo cap motion capture on my app. You know, I have field No, I I mean I have Higs Fields and I'm pretty sure I should be able to find a model that does this. Maybe I'll try that. I'll try that. Shout Shout out to Higsfield who wanted to sponsor me for $30 for a dedicated video. No way. Yeah, joke. Oh man, that is Oh, shameful. Shameful actually on them on their part. Yeah. So, photo AI. That's um Anyway, so this is an old website, but he's maybe he's added a video feature to it. Anyway, cool. Nice. Yeah. Yeah. So, I thought you were showing a website anyway. Yeah, I was waiting for you. Oh, okay. Oh, okay. You should This is This is just warming up our audience with fake eirls. Okay, great. Great. Okay. So, so this video this week went quite um viral with almost like 30 million views. It's about words per minute. How many words per minute can you read? And this is, you know, obviously a video. And I I saw this and I was like, hm, what if I make this to be um like an app? So, I prompted and this is what I have. So, uh, if I reload, you see here we have a text box, word per minute, whatever, the anchor letter, and I can just start and then I have a one second pause at the end of each sentence. And yeah, uh, I think it turned out pretty cool. Yeah. So, what you can set the words per minute and just build it up, build it up on how much you can read. So yeah, I don't have I guess that is something to add like stages but right now it is um just like a slider so I can start and let's see if it actually real time adapts. Yes, it does. I just added a prompt saying can you make this uh update the animation engine real time and what's the uh what's the importance of the red letter? It was the honestly it was in the design and I just took it but I believe it is important to have the words in a specific anchor point and here I see that for some it is the third for some it is the second I'm not sure what what is like the formula is it like it looks like it's the first 40% of the letter maybe because because we I I know that that we don't read the letter the the entire words. We just see it kind of like as a shape. Maybe that's the start and the end of it. So you actually read the word or something like that. No, no, we never do. We never read the word. That's why uh it's also really difficult to notice um mistakes in a word. Or maybe sometimes you do, but you can still read it because we don't read the letter by the letters one by one. we see the start and the end of each word and that's how we read it and that's why we can't do like crazy my um I I suck at this like my I think it's maybe a dyslexia thing but my brain trips over itself so maybe actually know my brain reads it but my I can't say it or something like that I really struggle with like fast speaking or Okay interesting so with this I could turn turn it into an app to train. I could add stages, you know, like, okay, start. I mean, it would take you two seconds. What was your what was your initial prompt? Uh, my initial prompt. Let's reveal all Oh, reveal. Create a Gab project. It is literally as simple. I'm just adding an image. Um, and separate the words. Oh, okay. I'm sorry. my I see there is okay let me how can I mute my iPad professionalism here boys and girls professionalism I have uh work on but it somehow doesn't mute it okay I need to mute my iPad I'm sorry oh it's your iPad is it um is Yan in the audience I just want to say thank you for my gift I received it the other day of a bunch of stickers so yeah for appearing on yeah you you're exposing me with asking me to show my prompt anyway so I asked to it's I did it with a conversation just like with voice mode and I was doing something else anyway this is like simple prompt and it didn't work everything at the like first prompt was you need to you need to read it out buddy read the the whole prompt the first prompt yeah what do you think okay create a GSA project. It is literally a simple input field that takes a long text and then uh the GSAP part is where we separate the the words and then we display them all. They display them all. Okay, that that is embarrassing. Uh one after another. This is for this is for and okay because I was talking like you know while doing something else. Anyway, sure. And then we can um do this read per minute kind of okay project. Um so I'm I'm I'm saying some nonsense here. So word per minute u a control and scrub timeline. I'm basically saying that we need those uh so we can move forward and backward. Well, it didn't do that at all. Um, and then um, it actually has the um in the in the chat. By the way, all of these controls I'm What voice thing did you use to record this? Just just the one from cursor. I was actually a cursor. Yeah. How how cursor got anything to do with this? Just the mic. The voice input. Cursor. What do you think? Coding application. This is cursor. Is it? Oh, I thought this is Web Flow. No, no, no, no. This is not Web Flow, but we'll get to that. Anyway, yeah, the prompt is really bad as you see because I was gonna say you you should uh not that you this could have been a quick thing, but like running your prompt into an AI to help refine. Yeah, I I do this uh we maybe I can quickly talk about this and yeah, we'll do that. So anyway, so we had the prompt and I was quite frustrated because typically when I do something like this uh I build the HTML CSS in web flow that way I have like more control and I know what I'm doing. Yeah. Uh here I didn't do it. I just want to quickly see if it's possible and then I got really frustrated because I wanted to change some structure in the HTML and CSS and I wasn't comfortable. I'm not comfortable doing it like literally in the code. And I was then prompting CSS HTML changes which I don't like because I feel like I'm faster if I I I know what to do, you know, if I do it myself. Anyway, so that worked pretty well, I would say, after let's say it it's like 40 minutes of work put into it. Yeah, looks good. And what I do I se I make typically separate chats here I I didn't need it but like typically I make separate chats for if I do concurrent stuff but here I don't have a crazy I have a JavaScript CSS and HTML file very simple and vanilla I don't know if I even if I needed Gab for this but it was kind of in my mind. Yeah. that you could do it with um it's just CSS hiding and showing and a probably Yeah. than a loop. Yeah, exactly. So, and the thing is this one is actually like completely vioded because I have not checked the code at all. So, we have like three lines of code. Well, we'll get into that, I think. But I think we're beyond the point of um checking every line of code. I think vibe coding is now it's kind of how we're all coding. We'll get into it. Exactly. This is very interesting because when I was doing that I obviously I'm not like where I would um judge this JavaScript and say if this is good or bad. But I I could see like when the CSS at the beginning it was not what I wanted. I was really frustrated and I was like thinking okay how can I tell this AI how can I describe it or should I just get my hand in it and do it myself but I guess uh this is the way we write uh we create software nowadays now it is before we go to the next one uh about the thing you said with my initial uh prompt how can we have a better prompt so what I did I have an idea for an app and let me know if you I don't know maybe we can do something about it. So I have an idea for an app. I thought I will talk to chat GPD to create an MD file. So that is to plan everything ahead. So I describe the app idea. It's merely for fun doing it for fun and learning. Um and I want to create a full plan but like before getting programming just have the specific needs of the app written down. in an MD file and then start programming it. And that is the app is simply um like an order to to order cocktails at my place. So when my friends come over, they can scan a QR code, the app opens on their phone uh and they can see what I have available to like what drinks to make for them and they can order. They have like I have to have inventory management. I have to have CMS authentication. commenting and things like this. A a bunch of things like um um Yeah. basically like what a bar would have probably much more modern but what bars have probably better actually works you know. Yeah. Yeah. And I have a good list of specifications. So after talking to Chad GPT and all is now clear. So Mhm. I I was thinking if I should do it live, like start live and basically starting vibe coding it. Okay, sounds fun. Should we do that together? That that's kind of like that is my question. Yeah, that'd be fun. That would be really fun. Yeah. Um and we could show people how to think about it because I'm very early in this. You are a senior software developer. You can have I'm a senior AI engineer. I've been doing AI engineering for six years. I've got six years under my belt. Yes. So, to be fair, in some ways I do have kind of six years under my belt, which I I don't, but like, you know, I was playing around with AI back in 20 22 something like this when CHBT first came out. But yeah, the AI side of things definitely been doing it for No, you have like started very early on adopting it and for people like people don't know. So let me say like uh we had this project that we don't know if it's still like happening or not. You created a prototype for the for the lead for the client and it was really impressive. Like guys, I'm telling you, like I saw the prototype he made in a within a week and it was really imp it was working. It was like not a prototype. It was an MVP. It was an MVP. I I went through and it was clean like the the UI was clean that the UX was like thought out. Uh the functionality there was backend, there was authentic, there was everything. And this is like a few months ago. Um in the past few months like things also changed again. So yeah, I would be interested in creating this with you. I would be like maybe I bring the MD file and say hey this is the idea say it enough you know for the audience and then we see how long it takes us to create it. Yeah. Yeah. I think you'll be surprised. What do you want to use? Cursor claw code some CLI some online tool. I I'm thinking like what would bring uh what what would help people kind of like making a tutorial out of it? How Well, the thing is, you know, and and after what you've just said and showed us and all this sort of stuff, um if you if you care tremendously about design, right, the vibe code thing, the AI system code thing is not really your avenue, right? designers um don't really belong in this kind of world, right? You should be willing to prioritize the product and the producttotype slashtooling more than the interface itself, right? I think that's where the the crossover between designer and dev really falls apart like in web flow, right? Web flow is really if you care about design and you want to build a website, you really need to stick to something like web flow like very granular control for the granual control. If you really care about design, if you really care about design, I'm trying to figure out a way to really phrase this where it's like, you know, um I don't want to alienate anyone, but I also don't want I want people to choose their tool um just have an understanding of what they're getting themselves in for. If designer wants to jump into vibe coding and start building things with AI, you need to relinquish your ability to drag and drop design and you need to then understand and learn how to create design with code. Not even me, not even a developer is going to AI, you know, someone who knows how to do development. I'm not going to do design through AI assisted coding. I'm going to accept what AI gives me or give it some, you know, some rough guidelines. But that fine control, I'm dipping into the code for that just because there's just no way. There's just no way at the moment that you're ever going to get what's in your mind on the thing unless you work on a really which is just a waste of tokens. Let's be honest. You know, why am I saying make the border radius of this specific element 10, you know, one I might as well just do it by code. Anyway, that's my point. So, we do some something that is something that can start from zero. That That's what I need. So So if people can watch it and be like, "Okay, I I have a w vibe. I have a code. I have a idea. I I want to vibe code." Yeah. Yeah. Let's do it. Let's do it. Let's do something next week. Anyway, let's crack on with the show because that sounds really fun. I'd really enjoy that. Um, and I'm trying to do a complimentary video to a Claude code course that I'm doing, which is going to be a really laid-back like coding session of me just building something, right? This may end up being something like that. Uh, I suppose I don't want you to get into claw code and all the rest of it. But anyway, I'd like to this may be a precursor to doing that. Just lay back, chilled out. This is what I'm doing. This is how I'm using the tool and you know, seeing that. Yeah, that would be cool. Yeah, that would be cool because as you saw, I'm I haven't gotten much further than just using cursor simply because I remember getting I remember getting a comment once and it was about the tool search tool which we're going to talk about today actually and some guy had a little rant and understandably I was like makes perfect sense. I know exactly what you're talking about where I'm talking about this tool search tool but it was it was abstracted from you know how to actually use it what you know not necessarily how to actually use it because I literally give a code demo but it was it was still very abstracted and he he complained about the video and he complained about documentation in general and I explained to him the thing is about documentation it doesn't know where you're going to use this it doesn't know how you're going to use it so it can only show you an isolated part on how to write it in Python or how to write it in Typescript, your knowledge needs to you're almost diving in the deep end. If you're if you're sitting in the middle of a um of some documentation and you just can't fathom what what you know what it is, which is by the way totally normal. I I get like that sometimes, right? I'm in the middle of documentation. I'm just like where do I start? But because I've got more of a background, I can start to pick it apart and it starts to, you know, become um more clear to me. And in fact, that's the exact example I gave in my response to this person. I said 30 minutes ago, I didn't have a clue about this, but I was able to pick through it. Mhm. Yeah. He could be someone who doesn't have the foundations to really take an abstract concept and apply it to something larger. And there's nothing wrong with that. It just means that he needs a bit more learning and and a bit more growth. And I think by doing something like we plan to, we can start to establish some foundations and some growth for to to as building blocks for someone to start diving into the deep end and and not feeling like they're drowning. They're just like, "This is tricky. I don't fully understand it, but I can pick my way through it and understand individual concepts and piece together what this abstract means." Uh, I didn't realize you were live. Been watching. Nice one. What was your name? Sorry, Tom. Tom Quen, you got an AI name. You got a um Quen Coda. Yeah, you should be Quen Quen Coda if that's your real second name. Yeah. Thank Thanks for watching. Thanks for watching. Thanks for being here live. Yeah, we're totally live. We're totally live. Twitter was down for a bit, so we've had to, you know, um pause for a bit, but we're live normally at this time every single week. Cool. So, anyway, yeah, speaking of live, let's go through the intro. So this week, Apple accepted defeat in the AI war. A bit dramatic, but it's kind of true. Defeat. We will talk about a paradigm shift based on based off Ralph Wigum, which is quite funny. Uh what happens if you let AI code for a whole week uninterrupted? What can it create? We'll show you that. um and some anthropic drama. All of this in today's episode. This is Command AI. I'm Kabarzo. And I am Sam and this is a week in AI design and dev. I would say we need to change that to all of this and more because there's more we're going through. Maybe Exactly. Sometimes I say it. Sometimes I say I just say it. But this more need to add it to the template. Exactly. Nice Tom. I remember I remember you, Tom. I didn't know your second name was Quen. Is that real or is that just something you've um added? Anyway. Oh. Uh, cool. Are we jumping into Yeah. Yeah. You remember Tom? This is so funny because Tom was at the table when we came up we didn't come up with this idea for this podcast, but we were talking about it. Tom, do you remember us talking about when we're in the food place and we were just chilling out, just chatting about doing a podcast or something? Let us know if you remember that cuz you were you were there when this started to sort of take shape at Webflow Con 2023 or 2024, something like that. 2024 shortly. Um anyway, uh are we cracking on with Ralph? So yes, by the way, Tom is saying, "Yeah, I remember it." Well, yeah, one of my highlights. There you go. This is the fruits of your collaborative labor here right in front of you live. Um, super cool. Yeah, let's go with uh Ralph William. I I I can't actually wait to learn more about this because I was trying to not ask too many question before the before we officially start because I I do have some questions. So, I want to understand how that works. Okay, cool. Let me just get something ready. Right, I might butcher this, but we're going to go for it. So, we're stepping into a new era of software development where the latest models are no longer relevant. And how we get ahead and is the way we orchestrate AI and it all begins with the unlikely Ralph Wigum, the famously clueless character from the Simpsons. Here's what that means and how it actually works. So, It all starts with Jeff Jeffrey Huntley here in this very very trippy, very strange video where he, which by the way, neither of us is going to do this concept any justice right now. We're going to do our best, but if you want to dig into it, you really need to go in and watch the source or at least learn about it from the source. You might feel a bit woozy afterwards because it's just a trip fest, but that way screen with all of the MD files. That's where it starts, right? So, I'm not we won't we won't just get into that. Well, yes, in a way. In a way, that's uh this is him planning out his his thing using uh claw code. I don't think he's actually using Ralph at this point because it's within claw code. What is Ralph? So, Ralph is a new coding technique that exceeds like it's less about as I said in the intro, it's less about the models. It's less it's it's less about the tools. It really abstracts from that and it it is kind of named as such because it's an infinite loop and it's and it just will run tasks infinitely unless you provide it some uh iteration kind of maximum thing otherwise it could run forever. You never know but it's an infinite loop until it gets the job done. Now there are ways you can do this and we're going to go over what Vel did. Was it Versel? Um what they did recently with a with a long running thing. It's less about that what it solves the problem. That's cursor. Yes. Cursor. Cursor. Not the cell. I see I see black and white and I just think it must be one of those. Yeah. Um it it it solves the problem of like context. So the problem is is that when you're in claw code and you're building your feature builds up its to-dos, you know, you might use taskmaster mcp or whatever, but it's building up that context. With every message, you're adding the previous messages to the bloat of that context. What Ralph does, it sits outside and around Claude code and actually gives Claude code the prompt um with a um fresh context. every everything is brand new and then it writes the results of that into an external file to keep track of what's been done. And so with every single prompt, it's brand new because AIS will just get dumber the more and more context you add to it. Now, I wanted to show you the actual code that makes up I think he shows it maybe at the beginning. It's very simple, very simple bit of code and maybe we can just run through it. Um, maybe I'll just have to Google it a second. I don't think he talks about it for long. What I'm going to do is going to get on a different uh Ralph Wiggum loop code. Let's just get this up. No, not that. Um, while you do this, I'm trying to also pull something in. Um, so this is this is some code. Let's make this a bit bigger. Okay, so let's pull out some highlights here. So here's the PRD. So you you you create your PRD. What is the overall task at hand? What are we trying to achieve? Um it looks like he's got a progress script here. Um archives and some various things like that. So he's checking for some things. You can see this is Python probably uh bash script and I think he's doing some checks here. dates, checking folders. I want to get to the the juicy bit. Oh, here we go. Here it is. So, here's a for loop here. Everything before that is just checking for files, whether they exist and things like that. So, you don't have to worry about that too much. This will this bit of code will run infinitely or at least in the max iterations and it will take the prompt and run it basically. Um this is the completion and it will just run this prompt into the AI and take the output and put it back into the external file. And once it's done, it will call this kind of thing. Ralph completed all the tasks. So I think I mean you had a couple of questions, right? Yeah. I was like thinking how I can implement this. Yeah. So this Yeah, this is just a file that you double click on your computer. You'll probably have to change. Let's let me see here the output. You probably have to. So, I think he's AMP is probably the tool he's using. So, I would probably change that with Claude. I would probably add Claude here, not um I don't I don't know what AMP is to be honest. Um, and you run this on your your your Mac will run this code and the script or the prompt will be just stored in a in an MD file in a separate file and that will what they've done they've piped that into so cat gives you the output of this file. You pipe that into this AMP tool or whatever. I don't know what t is but I think it's I think that might be just returning the response. So, you're getting the response from the prompt. Yeah. And then you're putting that into an MD file and then pumping it back into a fresh brand new prompt for that AI every single time. Every single time. And it and the beauty of this as well, and we'll get into this in a bit, the idea of running like multiple agents in parallel is kind of solved with this because the problem is if you've got one agent running one task, another agent running one task. Let's say this agent over here touches the file that this agent over here is running. How do you how do you orchestrate that? How do you know about those? How do they know about each other? The thing is you're giving Ralph autonomy here to decide what the next task should be, what the next appropriate task should be. So it orchestrates itself, outputs that to an MD file, and then the next agent will take that and then run with it and and keep doing that. And it will only it will do things one after the other based on a fresh context, a fresh perspective and all that sort of lot. So there'll never be crossover. And you in theory could just run this infinitely until the job is done. And you could tell it to write tasks, you could tell it to uh sorry, you could tell it to write tests and things like that. So there are ways you can get it to like test its work until it's done. Now, the max iterations thing is literally just so you don't end up spending a bunch of money if you don't have to. But I think it's worth trying this out with a small task, whether it's something like, I don't know, change the header or whatever, you know, um add a link to this header or, you know, maybe maybe we try it with your with your tool that you're the the site that you're building and stuff, right? Yeah. you. No task is big or too big or too small for this. Maybe if you know exactly what you want to do and you could get it done in like a quick little bit of context, then whatever. But if if it's a challenging bit of code or a challenging feature or something like that, get Ralph on it so it doesn't get lost because that's the problem. You get in a bit of a loop. There are some agents that get out there in a bit of a loop. And a word on the model and reason why I say models matter less is that people are having great luck using traditionally not so great models on this because the the context is fresh and it just redefineses how we use how we use AI. Like Google will give you the million context window but truthfully that doesn't really solve any problems because the more you add to a context the dumber your AI gets. Where is this has just got fresh context, bit of prompt, bit of output from the previous agent that's run along with it and it's given to you. You mentioned that you're already kind of doing something like this. Yeah, this is interesting because the the moment you mentioned it, I was like so what I do is in cursor I have multip and I I might be dumb for doing it this way. I haven't watched any tutorials on it. So the way I do it, I have multiple chats. So let's say one chat is specific to either a file or something in the same context. So let's say in one chat I change the styles and in another chat I add uh animations and in another chat I have like the base the core like often uh let's say animation engine. So I create like a GSA complex function and I have one chat specific to that one and then another one kind of like taps into whatever we are creating in that file to to to do like other things. And that's that's how I manage my chats and context because I notice like if I'm if I'm like one second building like the the the whole thing and then I'm asking about a style change and then I'm asking about pushing it to git or something or then I'm having a question to learn something. It it mixes them and sometimes I try to do something and it does one of the previous things because it's mixing them. That's why I decided to separate them and I by the way I do typically open a chat about learning like things that I want to learn. So it's just the chat mode no agent mode. Now I take all of those and also mix it with an MD file just a file where sometimes I ask the the the model to to to say like based on the mistakes you did this is what I say based on the mistakes that you've you did and based on like my preferences or like the things we've been through based on the the battle we've been through create a file uh and add all your learnings so I can give it to another AI model and it typically does it in a way that another AI model um can benefit from it which is also sometimes fun to read because yeah it tries to not hallucinate it tries to yeah you know not sound like AI if it if it has copyrightiting in it and things like this so that's how I do it so the the MD file for me is kind of like memory it it's exactly well the chat is memory but the chat is messy and big and it has like a bunch of stuff and it has issues like it has the mistakes that it did. But that file is kind of like um the the best memory like the best learnings and the best things that it did and like the notto-dos. I guess I could add these also on the cursor level or like on the project level, but I come up with these as I go, you know? it it's not something that I that I have prior to the project. So like the the the external memory is really crucial to this, right? And all you're doing is putting a little bit of structure to say you're almost saying like advise advise the next agent. You you you mentioned that you doing that, right? The next agent. How do you give clarity to the next agent that's going to run this? Right? So we had a question here. Now, I've shared this link in the chat to this repo just to say that um this guy uh what's his what's his bloody name? Um I don't know why he's called himself oh no snark tank. Here we go. Ryan Carson, that's the one. So, he's got a great way of taking um he's got a great ability to take concepts and build a simpler version of them. So Taskmaster was the big one like a few months ago. He built his own little simpler version of Taskmaster that's a bit easier to to understand and uh he's he abstracted it in a similar way. Ralphie's done the same thing. So this isn't and we'll get into why this is important by the way because Ralph actually is not important right now. This is not why this is making news. But anyway, all that to be said, this is one implementation of it. And Ryan tends to do a very good job at of simplifying an otherwise complex subject, even though it's not too complex. And George asks, how would he run this on um Windows? Now, I'm assuming that you can't run bash bash scripts on Windows. Um I did see something here. It says brew install jQ um on Mac OS. So he's really Oh, say AMP CLI. See, that's that AMP thing that I mentioned earlier. Oh, sorry. You I didn't have it shared. Um George said, "How do you run this on on PC?" This could just as well be a Python script that you run. Uh you might have to search like a Windows. You say yes for the bash script. You can run bash scripts. Um, but either way, I think it's this is a bash script. It could be Python. It it could be any language. It could be TypeScript. This could be a node script that runs this code. Um, just it just needs to be able to run applications like this is running AMP, which bash scripts can do. So yeah, all that to say, the reason why we talk about this, this is very interesting and I I totally recommend that everyone plays around with Ralph to see if it actually has an impact on getting your features built. There's nothing, like I said, there's nothing big or there's nothing too big or too small for for Ralph to run. It really just leverages that fresh context and um and that uh ability just to not uh pollute the the context window. The important thing here which underpins this whole conversation isn't Ralph, it's the fact that we are now entering this new age where we were once fussing around should it be cla what model should we use? All of this conversation is now behind us. We're now learning h what it takes or what it means to be a software developer in 2026 using AI. And it's becoming less and less important about what about the code that you write. You will no longer be hired for the code that you write because the code that you write is is a commodity. Now the important aspect is how you scrub up and how you be a better orchestrator, a better engineer to understand how all of these puzzle pieces fit together. And this is so crucial and it's something I love because I get to entertain my UX hat as well as my um uh DevOps hat. How I architect these. I'm not the best. I'm not going to claim to be the best, but I love playing in this world where it's not about whether I can or cannot write a function to do something. It's whether I know that this function fits or sorry, this feature fits in the grand scheme of this project or this tool that fits the users and the demands of the client. And that's where that's where software development is. Now, I mentioned last week that I don't think we know what we're building. don't know what we will be building in in a couple of months. And this further emphasizes the fact that you know you as an engineer again just will not be hired as a as a coder. Your role will exist. You'll be paid to know how everything fits together. Um so gone are the days of of counting lines of code written because it's just not important anymore. Um it's still it's still it's still a troubling time because this does remove the need of potentially several several developers costing a lot of money because if you've got one developer who can just do all this because you don't you know instead of writing six feature each individual developer writing six features you've got one developer writing all six on a Ralph Wigum loop which by the way can iterate through different tasks different feature sets one after the other instead of parallel like you would traditionally do it. So, this is why this is important. Who knows what the next tool might be? You mentioned a character from Rick and Morty. Oh, yeah. Mr. Misix. Anyone know about that? Like that could be the next one because you said, Go on, explain. Yeah. So, with Mr. M6, it's so fun. Like, um, Rick does this like it's it's a character to help. It's kind of an assistant. And you you you ask Mr. me seeks to do something, it comes to existence, it does it, and the moment the task is done, poof, he's gone. He doesn't exist anymore. Uh, now that the whole joke and thing is about what if you give it a task that it can't do and that is where it starts to ask for another Mr. M6 to help him and another Mr. Mi6 to help him. And that is kind of like the sub agents that we do. Um but it is very interesting because it reminded me of that because it it will also try it try until the job is done and then you know next Mr. Mi6 also is kind of like fresh context. Exactly. So anyone who knows it comment. Yeah. Yeah. Yeah. Like a loser. But, you know, jokes aside, like that could be the next iteration of this. So, I'm digging massively into claw code right now and I'm looking into some of the features and I've been learning that uh slash commands which are shortcuts um can be used by sub agents. So you can tell you can say once you've done this use the slash prit you know get commit slash command or review slash whatever you can tell it to do the slash command. So there's this um there's this inner task that's running that it can run it has an agent has access to things like that. I'm also seeing here that he's got skills. Now, skills are another Claude code uh thing. It could be that the next iteration of a W Ralph Wigum thing is someone's figured out a way that it's a it does spin off sub agents and it does ask for help or something like that. I don't know. I'm just I'm just spitballing, but we're we're this is the new era that we're going in and Ralph just might be the start of it. And I compared it similarly to Deepseek when Deepseek first came out. Deepseek was not the smartest best model in the world. That's not why it made waves. The reason why it made waves is because China proved that you can train an AI for a fraction of the cost that all the big companies were doing. Um, and it be a pretty good pretty okay model. And that's why it shook things up. Um, we're seeing seismic shifts in some of the undertones that are happening, not not in um not in the most obvious ways. So, that's Ralph Wigum in a nutshell. We'll leave links for everything, you know, down below once the video is out and stuff. But yeah, pretty cool. Cool. Um, so what do we do with the Any questions? Let us know. Yeah. Sorry. What do we do with the what? Uh, with the coding for one week thing. Yeah. Yeah. Go for it. Whack it up and we'll we'll we'll talk about briefly and then we'll move on. Yeah. So, I'm going to I will show it. Where where do I have it anyway? Let me I'll show it. So so so here we have an article by cursor. Uh this is where they gave a task or like made this project to to create a browser completely from scratch. an original idea, but it's a quite difficult task to create a browser from zero. So they they did it and they gave CE GPT 5.2 two a whole week to run uninterrupted to create this and they wrote this article about it and it's pretty interesting and you see here how it um starts to render HTML and you see it doesn't really work at the start but as it iterates it gets better and better and they they say it kind of works. It is not perfect. It has lots of bugs apparently and it's it doesn't support all the CSS and everything that the modern browsers have. But it's a very interesting experiment that they did. And by the way, uh Chad GBG wrote like three million lines of code. uh just earlier you said it's it doesn't matter like you're not judged by the code you write because you know the code is written by these AI tools and you can just you know make three million lines of code and throw it away next day. That's quite literally what we're doing nowadays. So yeah, this is a very interesting thing that is happening and it is very related to the other video uh the Ralph Wigum one. I don't know. So you you uh looked at the article and you said this is they they use a different a different method. So what they are doing they have multiple um agents doing different tasks, right? Um, yeah, they don't mention Ralph at all in this. I think this is more of an exercise of and I believe everyone should do this is how far can they push how far can they push AI? What can it do for us? Like I don't know. I didn't read or I don't know what the goal of this was, but what what do they say? Are they are they actually trying research? This is like just Yeah, this is they they're trying. No, it's not about create. I I believe from what I understand it was just to see what they can do. It it seems like uh cursor is trying they're trying to do multiple like they're trying different ways of creating um software every week. And here they uh what's next is interesting. Multi-agent uh coordination remains a hard problem. Our current system works but we are nowhere near optimal. Yeah. Planners should wake up when their tasks complete to um to plan the next step. Agents occasionally run for far too long. We still need periodic fresh starts to combat drift and tunnel vision. These are like we see these like we try to fix something and the the the model gets to into this tunnel vision. It's really difficult to get them out. Uh so you have to start and it is hard to run them and give them a let's say broad or like general task and and expect them to create something of quality and of detail. They are not there to do such a thing yet because of these issues but we are getting there. We are like trying different things. Ralph William one of them and multiple agents uh also is another way of doing it. Uh but the core question can we scale autonomous coding by throwing more agents at a problem? Let me actually zoom in. Um, at a problem has uh at at a problem has a more optimistic answer than we expect uh expected. Um, hundreds of agents can work together on a single codebase for weeks making real progress and ambition ambitious projects on ambitious projects. The techniques we are developing here will eventually inform cursors agent uh capabilities. If you are interested in working on the hardest problems in AIS software development um yeah they are trying to hire. So this is interesting to me that they are trying something and then they're publishing their results. That's it. Um, so I think like to to drag Ralph back into this, I think they've encountered all of the things that Ralph uh solves, right? And what these guys curser are, I mean, I know they are they've built the basically the best AI uh you can get, but they are working within the shackles of what AI can do. I got to phrase this in the right way. they're working underneath AI whereas I feel like the Ralph thing works above AI if that makes sense like they are oh we have to context is built up so we need to think about how we do that it's like well what if context isn't a thing like it's almost like it's so outside the box you know what I mean like the the thinking about Ralph Wig outside of the box that it rewrites all of these are just not issues and and there was a second article that they wrote about the orchestration issue. You know, you don't need agents running side by side and how do you fix that? It's like, well, what if you what if it doesn't matter or what if there what if you run linearly? What if it's not about parallelism? Parallelism uh which is again locked into the idea of you have six developers working on six features. Well, if you have one developer fully autonomous working on all of the features one after the next, you know. Yeah. Well, time is the the concern here. I believe like you prefer time is not an issue now because you can you can build features in minutes as opposed to days, you know. Yeah. But if everybody can do it, the whole industry is changing and now the the Exactly. Yeah. You like smaller teams can do bigger stuff. Uh there was this idea that we will have like billion dollar companies this year uh made by one person individuals. Yeah. Yeah. Individuals will be able to create like huge companies because you know this new paradigm. Mhm. For sure. You can you can have a lot of Mr. M6. A lot of Ralphs Ralphs riding around. Yeah. Who wins? Mr. music or wrong. Let us know in the comments who's a what's a better an analogy or character mascot. Yeah, I I think Mr. Music is Yeah, he he's funny. Yeah. Anyway, so this was the the article. Nice. Well, if you follow me as an individual, because I do have my own channel, and I like to plug it as much as humanly possible, I've just released a video on another kind of groundbreaking moment in in AI, which is the tool search tool in Craw. Now, if you don't know what tool search tool is, it it's an it's a a way of handling MCPS. MCPS take up a lot of context to have them running all the time. And the disadvantage of this is that you need to shut them on. You need to shut them off. Like if you don't need to use the GitHub MCP, shut it off because it's taking up your precious context inside of claw code. Tool search just takes the description, a well-written description of the MCP and it just it just cares about that and then it will dynamically load the or lazy load as they call it uh the MCP when it needs it. that was brought out in the API you know the when if you use claw claw uh sorry if you use anthropics API and you want to just you know use it that way but it wasn't available in claw code well now I'm proud to say that it has indeed been released on claw code and you'll see from this screenshot here there's no I don't know if you recognize this if you're a user of claw code but there was always this MCP line and it's always blue and taking up loads of context. And you can see that tar MCP layer is completely empty. And you don't need to do nothing. You don't need to do anything to have this work. It's literally just update cla and it will just work. And it will just remove the um all of the all of the MCP context status. Now, uh annoyingly I have to log in, but let's leave that for now because I can do everything by voice. It works in the clawed desktop app which is really good handy because a lot of people a lot more and we'll get into co-work as well now a lot of people are using the claw desktop app but what again what this means is and it was something we tal spoke about last week when we went through Boris's um use of claw code he he does everything from claw code let me know we've spoken about MCP so much but I'm really curious to know where people's knowledge gaps lie in and what MCPs But it's a it's an a way to interact with applications and you know online services like notion or whatever you're using Slack. um it's a way to interact with them all within your AI and you write natural language rather than saying uh send you know interact with that API rather than send the name last name email address you you know as like a JSON structure you just write out on plain English and it will send it off get the data and present it back to you in plain English none of this coding nonsense that you uh that you typically have to deal with when trai working with traditional APIs is MTBs answer that question. Now, because of this, on the one hand, we can use tools like Taskmaster to help us code as we write code. Um, it could be documentation. There's uh I use one called Ref. It's an it's an MCP that helps uh extract up-to-date documentation for for newer technologies. Uh so, we can use it to help us write code, but we we can also use it just to again update a notion database. You know, let's say we've got a new we we do everything in notion for the podcast. Instead of instead of going on to notion or opening up the app, just do it all within claw code because I'm sat there. MCP is running. Try it. It's no longer taking up context. So, I can just leave it running. I normally really just and and then you you load up claw code and then you realize everything's like uh compacting really quickly and then I shoot up MCP and it's like ah because all the MCPS are now turned on again even though I turned them all off you know so it's really frustrating to manage your MCPS now I can just leave them all running now I can push to GitHub now I can chat with uh notion you can the notion MCP is actually really well featured you can get users you can update pages you can update databases you add to, you know, it's really well featured MCP. So, I' I'd definitely try. You can So, now you can add this in cursor. Whilst you're there coding, little side quest, uh, add a thing to our little, uh, page that we've got for our news articles. Jobs are good. Do you know what I mean? Interesting. Um, all within cursor. All within cursor. Could I do that from my phone? Uh, I don't I don't think there's an MCP client that runs on your phone like from you need an MCP client. You need something that you can load MCPs. You might as well just up you might as well just download the app when it comes to that. But um if there's if there's a tool I don't I don't I don't think I've even called downloaded, but if there's a there's something that I do just not signed in. Um, yeah, if if there's something that runs MCPs on your phone, absolutely. Just you need an MCP client. Um, I'm hoping that is that a good explanation because it is complicated. I understand it's complicated and I've got a video coming up next week specifically on MCPs, but I would like for people to feel like this is that they understand MCPS, the reason why it's important. Again, in my video next week, I'm going to show you what a traditional if if I was going to update a database on notion with code, like what that would look like if you're going to write code. Now, I don't need to write any code. I can just literally say and and the AI will just figure it out, figure out everything out in between. It's it's beautiful. I I I understand the idea, but I didn't have like a proper use case until this week. uh we launched a website for a client and the images are not exactly how I want them like uh so I don't want to go to the CMS and download everything and re-upload and yeah and so on. I want to use the web flow MCP to to download all the images. I haven't done this yet but that's the plan to download all the images um change them all to what I want. basically the dimensions and a few things like that. I don't know if I could edit them. Could I like edit them bulk edit them? Depends what the MCP uh Web Flow MCP allows you to do. So maybe like with something else. That's what I mean. Like download them. Okay. uh put them on my Dropbox or whatever and then use another MCP to edit them and then re-upload them to web flow potentially. I just would not know what MCP you could run. You could uh the Figma MCP reads from your Figma, but I don't think it writes to your Figma. So there might be one there might be one thing you can do but what I would what I would do is download them to my local machine because I use warp which is my AI terminal I have image magic on downloaded on my computer. So I just I'll just ask the AI terminal to say look at this folder change you know do this to the image because mostly everything's run by image magic ffmpeg and image magic literally run pretty much every edit image editing tool or whatever you've ever seen. Um, but yeah, you'll have to look around to see if there are any image editing. What do you want to do to the images as an example? Uh, change the dimensions and maybe maybe if I could would be cool to boost the contrast. You know how Apple has this autoedit mode uh where it just adds a bit of contrast, a bit of color uh automatically. That would be nice. Yeah. So I would I would um again download into my local machine and I would I would then go into warp and say go in this go in this um folder and boost the contrast and change the dimensions to 16 by9 or 9 by6 or something like that. Um and I think image magic is so good that you can even it's probably has some sort of face detection. Let's say they're all profile pictures and it probably has some sort of face detection or at least a face detection plugin and then just let my AI terminal do it sort of thing. But um Okay. Yeah. Uh yeah, I don't know. I don't know whether there are any image image editing MCPS that you can throw an image to and whatever. Are you using Warp the free version or the paid one? I'm paying for it, but I'd be interested to know what the free version gets you. If you do want to if you do think about it, let me know because I am on fire with my referrals for war because I think everyone sees it sees me use it just like that's amazing. Um yeah, in fact, I'm going to let you I'm going to give you a send me the referral. Send it in the chat as well. I'll send the chat because it is honestly really great because it's just especially as you start to download FFmpeg, which is video editing or whatever. Um image magic, which is image editing. uh if you're accessing servers as well. If I'm having trouble with the server, like I'll just say, you know, lo log into the server and figure out I've had a lot of um problems with Docker images. So many Docker images running um and taking up loads of memory. I just say find out where the where we can save memory on this server and it will literally just do everything. It's it's it's great to have um it's great to have an AI terminal. It's like I would describe it and we'll get into co-work again. I've said that a couple of times, but we'll get into co-work. I've put it up on the thing. Um, cool. It's it's it's basically what Apple intelligence should be. It just turns your Mac into an AI Mac. It's it's great. So, um, yeah, that's tool search. Should we are we gonna Should we move on to Apple Gemini? Yeah. This can be a quick one because I don't necessarily have a tremendous amount to say to this, but have you guys seen that Apple have given up and they are now resorting to Gemini to run their AI powered Siri. So, this is another iteration on Siri. You know, we saw the big um the big update where it goes, you know, all nice and colorful around the edge. Well, now they're going to update it again and they're going to use Gemini. So, Apple intelligence is sort of Google intelligence basically. Yeah, basically. I wonder what Oh, go on. Yeah, they they say we're using the this foundational model from Google for our foundational model, which which is kind of like doesn't mean much because it it is their foundational model. How can it be exactly right? Well, here's the thing. So, this is this is my input. This is my take on all this. So, I think people are focused on Gemini. Now, Gem, Google have an opensource uh coding uh AI called Gemma and you can do what you want with it because it's open source. Now, I for sure Gemini is probably built on top of Gemma. So I think in the same way Apple intelligence will be built on top of Gemma and they'll they'll be able to call it whatever they want. So I think people are expecting you know again it just to be like a not necessarily a direct copy of Gemini but I don't think it will be Gemini. I think it won't cont there will be features that Gemini has that Apple intelligence won't have. And I think this is prevalent across the across the the industry right Apple don't make their own chips. They just help design them. and they take a base, you know, um, construction layer from TMC or whatever it is and then they add their own layer on top of it. They're not manu, they're not a manufacturer. Apple don't manufacture anything. They just design. So, I don't think this is all that, you know, controversial that they're taking Gemini. Um, this goes back to this um thing that everything is a rapper. Yeah, exactly right. Yeah. Everything is a rapper. It's just like changing the layer of abstraction. Uh in this case, you know, Google, I don't know like Google Google's model is based on something else and then it's using either uh Nvidia chips or like nowadays some of their chips. So yeah, the interesting thing is obviously Gemma is an open-source model, right? They don't have to pay a cent for it. Yeah, I did just see I highlighted on the screen Apple was planning to pay about $1 billion a year to utilize Google AI. Again, this is all, you know, speculation at this point. And I do wonder what's going to happen I wonder what's going to happen to Open AI because obviously right now is that if Siri can't sort your issue, it'll forward it to LGBT which will still fail because the integration is just awful. But where does that leave Open AI in this mix? Will will it still resort to open AI because technically under the hood which you shouldn't really care about Google uh Apple Foundation model is using Gemma or Gemini. Yeah, I would be really surprised if they would use Gemini and then if it couldn't do it send you to OpenAI. I would be really But here's the thing. It's is it Gemini or is it Gemma which is open source and looking at the terms and conditions does it say that you have to like if you were to use Llama if you have a certain amount of users you have to actually say you're using Llama whereas I don't know what Google Gemma's terms and conditions are. Yeah. Well Apple is going to pay a billion which is not a lot. It is well I don't think any of this article is anything the only thing factual is that Apple has picked Google. I think the million dollars we're all speculating at this point. You know what I mean? I think because because Google is paying Apple 20 billion a year to to be their um default search engine. Yeah. Yeah. And this is, you know, it is it is what it is. You know, this is the cost of doing business. It's so funny to me how you're gonna have like one arm of the company literally taking the other arm of a a company to court, yet you've then got sha handshakes and smiles in other areas. It's crazy. I think they literally just think this is business. This is just business. Yeah. Yeah. You know, no heart, no soul, just business, you know. It's just business. Um it is quite funny like the Samsung uh news. I I mentioned it just to you in private like Samsung couldn't buy RAM from Samsung because the company has different arms and the the arm that was selling uh RAM tries to you know maximize profit and they are selling it to server companies and companies that need it for AI and the mobile division couldn't buy it from this from the same company and they because it's wild how all this works. Like I wish I in some ways I wish I knew more about it. In other ways it makes me really depressed. In other ways I'm glad I don't know anything about it. Um we we haven't spoken about it yet. Uh but Big A you you've seen the video by Big A where he talks about this um new way of acquiring companies. Oh yeah. So you don't acquire people don't acquire companies anymore. We first saw it with Meta and Scale AI. They don't acquire scale AI, but what they do do is buy their staff, buy the head honchos, and that by proxy gives them aspects of the company. Um, Zero Shot is more powerful than Ralph, I think. You know, again, I'm not we're not here to debate whether something is better or worse. I think, as I said at the end of the section, I wonder if uh Zara Godex is is caught up. presumably not, but if you are still watching at this point, um we're not here to debate whether something is good or bad. We're here to um almost celebrate or identify a shift in how we use AI or how AI works for us as opposed to any specific technique being better or worse. It's like we're starting to now command AI. Um anyway, I'm presuming they aren't fully caught up, but yeah, I um it'll be interesting to see, but sorry that I was saying about the big A thing. Yeah, it's um you're you're acquiring different aspects of a company and and he uses the boat analogy where it's like, well, how if you replace the sail and then you replace the kind of mast and then you replace the deck and then you replace at what point does it become not become a boat anymore? I've forgotten the I've forgotten the the not analogy, but the the theory or the paradox or whatever it's called. Um, but the point is is uh yeah, we're we're in a weird world of business and I wish there's probably things we're missing out on as business people, but it it goes over my head sometimes how how businesses operate. Yeah. So, co Yeah. Uh, I I can show my screen. Yeah, I didn't realize you were going to show your screen. Gosh. Uh, have you seen the video? Of course. Okay, mate. I've I've dug deep. Yeah, we can we can watch it together and then we'll talk about it. All right. So, there is a new tool by Claude. It's called Claude Co-work. And it's going to help you with any tasks you have apparently. So not just chat but ask it to do something which in this case uh you what what do we see? You see, create a file, uh, crunch data, make prototype, uh, prep for the day, organize files, send a message, and then you see like it it works with your with your fer finder and file system, and it can access your desktop, I guess, and do everything. It it's quite scary the things it does, but also sounds very interesting. It got also like 50 million uh views almost. Yeah. Actually, one more thing and this is where it gets um interested interesting with what it does. Check Google calendar and you know it opens the the tab. Does does it do it all or like how does that work? The Chrome extension. Okay. Ah okay. Gotcha. because it can get scary to the level of access that you can give these tools. Anyway, so this is it. And again, Twitter being Twitter today. Oh man, it's it's not loading. Okay, it is loading. So here they ask it to organize the files. And this is what is very interesting to me because I I have I'm not very organized and I have some older hard drives that are very unorganized and I would like to organize them but I don't know like yeah this is kind of interesting how it does it documents photos very cleaning it cleans it up very uh nicely and tidy. Uh have you tried it? This is only for people who have access to the max mode that is like I believe the 200 uh dollars per month um plan, right? No. Yeah. Yeah. No, I've not I've not tried it yet. Um I mean in in summary, I would describe this as clawed code for normies, right? Yeah. I can I can already do this all with claw code. I can already do this with claw code, right? Ah, okay. Interesting. I can already do this with And in fact, yesterday for yesterday, I guess I understand. Yeah, I had all my hard drives out and so I I use Screen Studio to record all my screen sharing videos that I do on my personal channel, right? And I wanted to move the Screen Studio file into the folder that was associated with the video that it was doing because it saves it in a separate folder. And I told Warp, say, "Look at the name of this Screen Studio file. Do a fuzzy search inside of this folder. If you find something, let me know." And then I'll decide whether I want you to move it into that folder or not. And it wrote a bash script and went through it. And it and then it would ask me, we found these three folders that it might belong to. Which one do you want to do or do you want to skip it? And literally like I'm a coder. I don't co-work is not for me. It's for people that it's called Kofanormies. Interesting. Okay. Yeah. Yeah. And I think it introduces or further exemplifies the seismic shift we're seeing not only in software development but also in normies where we're starting to think about what can we build, what can we do, what systems can we build to get micro jobs done. not you know we don't need to build a SAS product to help us you know build a pipeline system for this that and the other we just want to clean our desktop how can I do that more efficiently organize my files into you know categorize them in certain ways and then put them into folders be a bit more specific about them rather than dragging and dropping and doing we're we're thinking more about how we can use software daytoday on really small tasks to help get our job done that will never see the light of day that we'll just scrap after you Once it's been used, just use one. Um, uh, Theo released a video today, uh, talking about we're build, he's building software that is literally going to be used once and then you throw it away. So, devs are thinking like this and I think Anthropic are trying to make the the workforce think like this, hence codework. And did you know they built it in a week and a half using claw code? That is Yeah, that is actually very interesting. And claude code wrote 100% of it. Yeah, 100%. So, that is insane. Yeah. Yeah. Yeah. So, it's really cool. And I, you know, I made a video a few weeks ago and I asked my brother, "If you could build any app, what would you build?" And he gave me an idea. And yes, you you couldn't do what he wanted with with co-work for sure, but it was beautiful to give someone that power, that superpower to say you can build your own software to get just stupid little things done. Um like what and I'm getting more and more stupid things little little stupid things done from every day, you know, like the example I gave with uh like the drink. Yeah. thing. Yeah. Like I I want to have it like when my friends come over and you know, I want to have a list of who wants what and their favorites, their comments on it and like star ratings. Then I want to to have like a nice thing like when they come by and I'm already pretty sure it will be much better than what restaurants and bars have currently. Yeah, probably. like like people who think like there won't be any software to be written. I I think they haven't seen they they haven't gotten out to see like how bad the software situation is everywhere. Like we have the best and it's still like quite buggy, right? Yeah, we have the best like the Mac and the everything in the on the web and everything, but if you look at other industry industries and other jobs and like the software that they use, a lot of it is from the I don't know. It's from the '9s. It's it's old. It's it's really bad. Um it has no UX anyway. So yeah, it's fun to see all of these being done with cloud code and I think things are similar to that. Yeah. Yeah. And I think it's only going to get more and more prevalent. People just again not thinking about it. Just I need this again. I just need to organize all my hard drives. Yeah, I asked AI to build the script, but that's all the cowork is doing is building a script that will run all this moving of files, comparing files and whatever. This is more of a conversational aspect where it can actually give you feedback and what I mean mine can I can ask it to like um so yeah, it's it's really cool. Um, so this is just I know we're continuing on this conversation, but this is what I've talked about in the past UI for AI. We are seeing more and more UI for AI. Um, and I see this also in um, cursor. When I use the plan mode, I pretty much always say ask me, ask me, ask me stuff, give me ideas. Um, here here are my ideas, but like use it as a seed and give me more. Right? And then it comes back with like a list of selection. So previously I had to type in now I can se literally select the things that uh the options that it's giving me and I can always you know add stuff to the chat but I'm seeing more UI for AI and I like that. Yeah, it's cool. So yeah, presumably you haven't you haven't used co-work either. No, because it's for the max for that's for the rich. For the rich. Yeah. I mean, I think you can get it for the 19 Oh, no. It's pro. It probably be $100. Yeah. Yeah. Yeah. I bought the I would probably get it. Yeah. The I bought the $19 a month thing because I was doing some tutorials and whatever and I didn't really want to spend the hundred just because I wasn't coding at the moment. The $19 ran out pretty quickly and then I needed an upgrade and so I'm back on the 100. They they always get me crawling back. Yeah. Yeah. I'm already eating the the cursor. um the the the extra money that you pay them. What is it called? Like for the extra context, not context, for the extra tokens. Oh, I don't know. I I'm not Do you know I just I I literally use co cursor as a text editor now. Like I'm not continue my subscription. I have no need to continue it. I'll just use I see VS Code. Just Yeah. Just use Claude Claude and and VS Code as a text editor, you know. Yeah. Um, thinking do we cover the anthropic blocking apps? I mean, to be honest, my take was always going to be I sort of see both. I don't understand the issue enough to really care as much. I don't know what happened. So, so users of Open Code, another very very popular CLI were I'm going to try see if they got the message somewhere. This credential is is not only authorized for use with claw code and cannot be used with other API requests. So open code allowed you to use your clawed code subscription with their tool and they got this message and then DAX tweeted out a fix for it. You need to prefix your API key with with certain characters and then they blocked it again. And this is Anthropic's attempt to make you well maybe we will get into this actually um force you not make you force you to use clawed code if you're on that max plan. So it's so funny how we were just talking about the max plan. This forces you to use claw code when you're using their max plan. Claude uh the max plan is not for you to use other tools. And Dax, the guy from Open Code, um, said quite rightly that quite rightly that models are not sticky. Models aren't the reason why people are staying with companies. It's why you've got Open AI throwing everything at the wall because the model alone is not really the thing that's keeping people. It's it's the ecosystem around it. It's the tooling around it. Similarly, Claude Code or Claude Anthropic, they want to tie you into their system as much as possible. So, there's no opportunity or thought or reason for you to to to jump do something else. On the surface, it's like, yeah, they're still getting paid. All it is is that CLI, but the more you use all their features, the more you use, oh, co-work, um, uh, skills or, uh, slash command, sub agents, all this sort of stuff, you your whole you start to become a goldfish in a gold goldfish bowl in some ways. And you you you don't really look out elsewhere because you're just constantly seeing that orange butthole, that little orange star thing, you know. Um and all you think about is anthropic claude code you know so on the one hand um Dax you know understood and he's the guy who invented open code on the other hand a lot of people are very frustrated with this um and all I would say is go watch their videos because I'm I'm sort of in the middle a little bit with it all um I think it comes down to just just choice But it's against their TOC's. It's always been against their TOC's to use this. Oh no, no, sorry. Because the law is if there is a loophole, then people can use that loophole. Yeah. Right. If you leave a loophole in, then people can use it and it's not against the law. So the terms of service might mean might be something a little less specific than you can only use this with code. It might be a bit more abstract and a bit more um yeah just a bit more abstract and less specific about using it with call code. It might it might leave open to interpretation that you could in fact use this with open code or any other you know tooling platform. Last thing I'll say is clawed code you can run as an MP NCP server. So that is that's another way you could potenti like I can use clawed code with cursor and I've got a short on that. I think I released it this week. You can um yeah, you can set up it'll be on next week's video as well. You can set up claw code as an MCP server and access core code within cursor. So don't know that's a bypass. But yeah, that's uh that's all I have to say on that. Cool. Yeah, that was a short one. Nothing to say on that one then? No, not really. I'm not very familiar with it. So, okay. So, we have um this one is interesting. Yeah, this is interesting. Obviously, guys, in the comments, let us know your thoughts, but Open AI, and I haven't read too much about this as well, so it might also be very short, but OpenAI has acquired torch, a health care startup that unifies lab results, medications, and visit recordings. Bringing this together with chatbt health opens up a new way to understand and manage your health. Now, on the one hand, you've really really got to be careful with uh your health. And the fact that I need to say that means that there are plenty of people not really realizing that when they give their health to AI that it could in fact one, hallucinate, two, be composed of answers from losers on Reddit who don't know what the [ __ ] they're talking about. Yeah. Yet AI has learned from that conversation. You know, you you know about the issue with uh parents. It's kind of like a joke. Uh taking advice from Instagram res. Yeah. They are just being Yeah. They are being exposed to this like short form content and everybody is talking about something they have no [ __ ] idea about. And parents, you know, like or like more older people uh they don't know and they just look at it and then suddenly your mom is telling you she might have this disease because that person on Instagram had something similar and they did X and Y and it might be all you know hallucinations of humans and now AI is learning from that which can I think I think it's less about hallucinations you know I mean I'm just I'm just chatting [ __ ] don't listen to me. But like I think it's less about hallucinations. I think it's literally just about the information that's been given um is is not necessarily grounded in truth, right? It's grounded in it's again it's Reddit um is a massive source of information for open AI JGBT massive source right uh that's a whole uh drama in of itself uh because um Samman uh what did he do he own Reddit or he was a uh investor in Reddit or something he was an investor yeah yeah he did some sneaky [ __ ] to get it into open AAI and now Reddit's basically a cesspool. Well, it's always been a cesspool anyway. And it's interesting that OpenAI will never if they, you know, you open up Torch or their implementation of torch, you know, on the one hand, Open AI wouldn't say, "Guys, you cannot take this stuff." They want to they want to be proud of their product. They want to say their product is really good. But realistically and and and morally they should say this is um artificially generated uh data and information and that you should go see a doctor. You should really go see a do like really hammer point home that it's not just some it's not just a little bit of small text you put on the bottom of an advert to say like you know um I don't know XY. It's not just a little bit it's not something you should do. something you should take into account like um so what I don't like about this new world that we are in it uh whether it's with health or software development it feels like nobody knows [ __ ] and it feels like there is less and less care for knowing like specifics it seems like everything is an approximity of truth but not truth but truth cannot be just approxim imately true. It has to be definitely true. And this is what AI is doing. It is like almost true most of the time. It it's not just hallucination. It is some sort of like a soup mix of everything together. And it it might sound right. It might be mostly right. But I there are so many cases where you don't want you don't want a general soup of like truth. You want a definite maybe even cited you know by a human let's say expert uh view on something like somebody who has made up their mind on something. Sometimes you need that much more than you know something that is just fuzzy and sort of true. So two two points on this one. You've hit like a real uh pet peeve of mine at the moment which um it's it's not about AI. It's nothing to do with AI. It's we've just become so accustomed to and it's this is the lamest example but I was on I was on Instagram and I flick through and I I get this story that someone's telling me about the what the basis of my favorite band and he sends a story about how he came up with the song and he's talking and he's saying this story and then just ends the thing and I'm like like why are we in this world where this person just feels no like desire to say Mark the basist told this story in an interview with um you know Rolling Stone or why they feel no need to site their sources is wild to me like Mark didn't tell you this story it's just like trust me bro this is this is I want to say that trust me bro that's trust me bro and by the way I think this story is true and you know people hammering me in the in the comments comments and I'm like it might be true. I'm not doubting that it isn't true. Well, I am doubting it's a a little bit untrue, but my point is just site your sources like we go through our articles here. We show the articles. It's really important for us to show the articles. It's not he hearsay. It's not this person said this, this person said that. Of course, you're allowed your own, you know, um, interpretations of things and, you know, we're not going to get everything perfect all the time, but I like to think that we show you the source and we talk about that. We're not we're not just like, trust me, bro. And similarly um there's a great um annoyingly there's a a a YouTube channel called Better Stack and they released a video on the tool search thing which if you just joined us like we we talked about the tool search in claw code at the beginning of this uh show um my video released and I've forgotten the name of my my video but their video was titled uh this recent released made anthrop clawed 10x faster. Now, there are two problems with that title. One, the tool search tool does not inherently make MCPS faster. It doesn't make it f In fact, I questioned in my video, I said, does it now make it slower because it's got to do that extra search to to whatever. Um maybe in some ways you could squint a little bit and say, "Well, there's less context, so it's it's it's got less information to process." Maybe there's a um something to be said about that. But what I do have a problem with is 10x. And I know we say it in conversation when we're, oh man, I just I was like 100 times faster this morning on my run. I know we say it in a what's the word? um colloquially colloquial way we say this sort of stuff. But when you're putting information out there and you present no facts or no statistics or no testing to say 10x, then it becomes a problem. Especially if you're a news source, do you know what I mean? We just throw stuff out there. And I'm more for me at the end of the day because I'm the one getting pissed off with it because the the youth of today Yeah. the folks at home, they're not picking up on the on the absurdity and the the destruction of what this actually means because it it means that when you say stats or when actual figures are put out there, you then ignore them and you you don't really process them in the same sort of way, cancer rates have dropped by 10x. Like, yeah, they're just saying that. No, no, no. This is a big deal. You know what I mean? Like, it's I I know exactly what you mean. I don't want to get political. I just say that if you look in the world what's happening you you see that everywhere it seems like we're moving towards where truth doesn't mean [ __ ] like saying you just say things and you don't site you know your sources. you you don't care about like the the the the truth anymore essentially. You Mhm. you don't care about like the scientific method. You don't care about any of that. you just make up things and you you you can't be you can't basee everything uh on your feelings like how you feel today and you just go on about you know make decisions based on that um and I think we we were on a different path I'm I tend to not think like that society goes like has a downfall I tend to not think that way but it's getting really really difficult to not see it that way because more and more um misinformation is being shared and people are getting more and more lazy about it or like they don't just straight up don't care that because I do I do care. I I don't want anyone to come to me and give me some [ __ ] explanation of something. I yeah you know if if somebody in your like personal life or like in your circle gives you [ __ ] you shouldn't take it you you're you know you shouldn't allow people to [ __ ] you. So why do we allow that on a bigger circle that we have which could be social media, it could be geopolitics, it could be companies, uh you know with our data, whatever it is, we we should be a bit more should care a little bit more about that. And and my overall rule and my overall message to everyone is question everything. Like don't just accept things at face value. And also respect that you you do have and will have different opinions and they're totally valid. Like it's okay that you don't if you think something should be done a certain way. That's totally valid. So just question everything you hear. Don't just like just accept it and that's what it is. I know we're preconditions and it's easier said than done. Newspapers have been doing it for years with the headline. They just say a headline. It's like you read it and there's actually not really a lot to do with the headline, but clickbait has always been a thing. Clickbait. Yeah. Um, but the second thing I wanted to say was, you know, we talk about we were talking about um uh torch AI and and health and getting an opinion from a doctor basically is where we're going with that. And AI is is not a doctor and making up [ __ ] and all the rest of it. Um, another message to everyone is like don't put humans on such a high pedestal either. Like, yeah, AI is good. It makes mistakes. We talk about the easiest example I can give is hallucinations rate, right? Hallucination rate. Humans hallucinate all the frigin time. We're hallucinating like probably more. So, we get on our high horse about like what did I see today or yesterday? AI made some silly Oh, I was watching like an historic video which is really cool, really interesting way to um showcase um AI in that someone does historical videos and he was doing one about Lisbon and he's taking photos and not photos, sorry, uh paintings and sketches of Lisbon way back in the 1800s and then animating them with AI. And I think this is a really nice way to like visualize Lisbon back in the 1800s or probably even earlier than that, probably 1500s, right? You know, he's very obvious about the fact there's AI. And then um there was one scene where the flag was like a different flag or it was the wrong flag that the AI had obviously whatever. And then there was this big onslaught of people laughing and they think AI is going to like, you know, rule the world or this and that. It's like Jesus Christ. Like when if you're going to use that as as a reason why AI isn't going to rule the world or or this or that I bet you're completely unaware about how many paintings, sketches or stories or written scripts or whatever that humans have just got totally wrong. Yeah. You know, but you just don't know about it. But you're then like getting on your high horse pointing your finger. A so [ __ ] Oh, I'm perfect. ever. It was just so funny to me like obviously I see both sides. I understand what they're saying. But but we have methods for that. You know, we have the scientific method to to basically use tools to to replicate and then make sure that you know that we we know that we hallucinate. We know that we get things wrong, but we have ways to uh get better at it and do better. And I all I'm saying we should question it like Yeah, absolutely. Absolutely. Question it and it's fun. I like it is fun to question stuff and go and learn. All that to be said, who knows what torch and open AI are going to do or at least um you know what is it going to be related to Johnny IV? I don't know. So there is word about it syncing with health data of your like Apple Watch and you know the worry would be that it would take Apple watch data and it would take your health data and give you interpretations of that which the state of AI currently is is worrying because it could say like my watch will alert me when my heart rate goes down below 40 beats per minute and it often does because I'm a very fit well not so much right now because it's just come out of Christmas and it's winter but when I'm at my peak and I'm sleeping, my heart rate goes down below 40 beat, whatever. It's normal, but AI will probably flag that as something I need to worry about sort of thing. Um, all this to be said, yeah, take it with a pinch of salt. Yeah. Always. You know the joke like you go to Google and Google is like, you got cancer. Exactly. Exactly. you know, you go with a headache. And like I think we're getting better at being more critical on that sort of stuff because that joke, as as funny as it is, that joke has now prepared us for of course we're going to see cancer in this list of things. And I did a I did a health search just um the other day and it was like, of course cancer entered that list. But now I'm well I'm more um I'm armed with the knowledge that I need to just take everything with a with a grain of salt and and just be a bit more critical on it. I I feel like we're going to slowly get hopefully slowly get better at that as as the more we hear about AI and it's like okay has told me this but I'm I'm not going to just [ __ ] freak out. I'm just going to take that information, absorb it, do a bit little bit self diagnosis, and if things don't get wor if things don't get better, I'm going to go to a doctor. Do you know what I mean? Yeah. Yeah. This is where it's to me very un unresolved, and it uh it's very interesting to see what we are going to do with all this um approximate knowledge. Yeah. Um anything else to say on that? We've got two more to get through. Yeah. Um, one of them I'm quite excited about, but let me This is probably more of a comment and probably should should have formed part of the comment section, but this was a video again. Theo is really uh making a lot of appearances in these videos. Hi Theo. Hi Theo, if you're watching. Um, he released this video this morning in fact uh one day ago. Maybe it was yesterday morning, but it was called code is cheap, software isn't. He's changed the title or the I'm thinking of the thumbnail. I don't know. But it's something we've spoken about a few times or at least I have spoken about this idea. We spoke about it earlier on in this episode where code is so cheap to write now. Like writing millions of lines of code is is a trivial task now. and my oh this is the root article. So realistically we should probably link you to this article here. Um here we go. This is called this is the article. And the interesting thing here is that it's expressing a lot of the ideas that we've already spoken about in previous um shows in that people are just building tools for them like it's trivial. We're not thinking about it anymore. I want to try and get you some quotes here. Um, SAS from scratch pads. I've been talking about this idea that SAS is dead because we used to look to SAS products to provide us a service because they've already built the tool whereas now you can build a tool in like no time at all. So there's a section here talking about how um SAS is is now changing. The one take that does start to differ is that whilst I'm I'm not necessarily sold, I want to read more into this article, but one thing I was barking on about was this idea that because SAS or software is so cheap and easy to build that there will then be an abundance of software and people will just build little tools and little things for everything because they're just like why should we pay a company to build us this you know automation workflow or this or that, we'll just build it ourselves and and it will cost no nothing at all. Whereas these guys, Theo is tending to agree, but um I love the design of this website. It's basic, but I like it. Chris Gregory, your screen is not showing correctly, I guess. Sorry. Oh, yeah. We've got um Sorry. Yeah, yeah, yeah. Let me let me do this. So, this is the section. This is the article. This is the section I was just talking about when it comes to uh SAS. Uh Chris Gregory is the guy. Chris Gregory.dev spelled with an I at the end. Let me put this in the chat. Um yeah. So where um Theo and Chris's arguments differ from say mine is that they actually think that well I do share a lot of the same sort of things that software is still going to be expensive because although you're not paying for the lines of code that you write anymore or the features that you deliver, it's more about how the architecture the real engineering lies in the abstraction and the architecture. And this definitely leans into the thing that I really enjoy about being an AI developer, if you want to call me that, in that I get to now not worry about what code it takes to write a feature. I can just dream up a feature if given the creative agency to dream up features, but I can start to balance out and how the how I know apps are put together, whether it's to do with servers, whether it's to do with authentication, whether it's to do with connectivity between two systems or whatever. This is what it now allows us to do. And as developers, those of us that are in the audience who are developers, I think this is where we need to um lean more into saying that um it comes down to I learned these skills writing code. I learned these skills working in the industry and I you won't learn this sort of stuff if you're just writing everything with AI. So I do think and this is this is the hardest pill to swallow for anyone but I do think you need to write things from scratch more and more because I can I I just cannot see a world where people are leaving university only having touched AI to have the intricate knowledge to understand how systems are put together to a very to a um enterprise level thing like you know you and I were working on a project that's mostly internal, right? That's kind of fair to say it's mostly the most of the tool that you and I were building was going to be seen by people on the back end that doesn't it doesn't need the same sort of level of enterprise like system security and things like that. But when you've got something you're releasing to thousands of users and um they're all going to have their different work, that's where you need to start thinking about like locking things down and and making it enterprise ready. And there is still a lot of that to be done. I don't know how people once we once we all start dying and you've got a lot of devs coming in who have never known to write a single line of code. How they're going to understand it. I just don't know. And that's where maybe this article touches on that. I'm not too sure. But that's definitely something that's coming, you know, in in my mind. So yeah, I think this is um definitely worth read. I've only watched Theo's videos, so I don't I don't know whether he's missed things out or whether he's been biased in certain areas, but yeah, it's it's an interesting world we're living in. It's something I've been barking on about about the fact that it's now ridiculously cheap to build your own software. Yeah. Last but not least, Apple. Oh, Creative Studio. Yeah. Oh, you've got I have another one. Okay. Yeah, I have another one. Let's go with this one. Yeah. So, so this is a new thing that I didn't that I almost missed. Uh Apple Creative Studio. Very interesting. Putting everything uh under one hood. Is that the terminology? Uh all the software that they are already selling but now under one subscription. Yeah. Right. Um, and it's worth bearing in mind that you can still buy these things individually. You can still spend 200 on logic or or whatever whatever it is. Keynote pages and numbers are still free. Free form is still free. However, there will be things that you get when you pay for the subscription service. Um, one of them being Final Cut allows you to be honest, I don't care for them. Final Cut allows you to search the transcription of a of what you're editing and go to that point. I think this is maybe what it's showing you here. Now, I think they've balanced it quite well here to be honest because they're they're serving both camps. They're serving the idea that people do just I mean I only I mean I've bought it now so no there's no point in me um buying it again but you know I use Final Cut Pro and I use Logic. I don't use any of the other things. I've tried using motion. I want to be able to be better at motion. But what is motion for? Is it After Effects? After Effects. Yeah, compressor is again, you know. Yeah, I've got it. But um interesting. But yeah, but because they still sell them individually and and they still get c they still get updates. They just don't get some of the features. And I understand why subscriptions exist because there is ongoing development. It seems absurd that you would buy something and then it would just be supported for the rest of your life sort of thing which Apple have been very very good at by the way. Um other software they'll just release a new version and whatever. Um but subscription fatigue is real. Yeah. If you scroll up, you see new devices get three months for free, similar to Apple TV, I believe. Uh, does that mean that I could get that with my iPhone? No. No, it's not for iPhone. It's iPad. iPad or Mac purchase. So, that's quite cool. But the other thing is to note this students and educators, one month free, then $2.99 per month or 29 per year. That's crazy. That's That's crazy. 29 for a whole year. um for everything. That's really good. And you know, you didn't hear it from me, guys, but it's actually quite easy to get the educated discount. I got all of my I got everything for like 200 bucks or something crazy. Something ridiculous. I got Final Cut Pro, Logic Pro, Motion, Compressor, Main St. I think I've got Main Stage, so I'm hooked up for life. But um but otherwise $12.99 per month 129 per year which is not too bad for it's fine if you're dipping in and out and it lowers that barrier to entry. So I think that's the beauty of subscriptions. You get to dip in and out. But if you are a pro user, I think it's just worth buying the software outright. Even if you spend 200, 200, 200, you know, 100, 100, whatever it is, you know, I think because they earn their money back within no time if you're a professional. Yeah. But so I I do think they've done this probably the best way. However, naturally, there has been a bit of backlash. Um uh interesting. Tell me about the backlash, but just about the price. comparing it to Adobe. Um, I just subscribed to Adobe again, but I'm going to cancel before it's too late because of uh the Lightroom, the Lightroom classic. I have multiple images that I want to edit. And I think there is nothing better. Anyway, Adobe is going to cost 80 something euros per month for the Apple uh for the Adob Creative CL um cloud which I'm going I'm not going to pay that. It's funny how you expensive. It's I call it Adobe, you know, Adobe. Yeah. I don't know. It's rock. I think Adobe is rock. I'm not too sure. it. Um, so I think this is I'm just looking through some of the features here, but you mentioned about the backlash. The backlash is simply subscriptions. Okay. And and we've heard mutterings about subscriptions for a little while with Apple and the worry that they were going to do this. Um, it's been it's been mutterings for for a couple of years, I think. Um, but I do think they've done it well. They've they've they've given you the software. They're still going to update it from a, you know, basic feature perspective. And maybe we'll get features later, you know, that will trickle down, but the subscribers will get the latest things first. Totally happy with all that. Do you know what I mean? I think there's a time and place for both, to be honest. I bet there will be a lot of AI in these. Uh, we will for sure hear more about that in WWDC in about I don't know, five months. Uh, hopefully. I'm I'm already excited about that. Uh, by the way, and I'm excited about new Macs. I've got two Macs right now. I've got two Macs. It's mental. Very nice. But I get I get this feeling where I'm just like I need to get rid of stuff and I just want to buy a massive souped up 14 in cuz 16 in is just mental. 14 inch 14 inch is the way to go. Yeah, the screen is just like too small. Well, I guess with the new paradigm and you you don't look at your code, you just need a narrow window to see to see your chat. Yeah. Well, I'm a content creator mostly now to be honest. So, uh yeah. Um yeah. Anyway, that's uh Creative Studio. Let us know what you think about the subscription, whether you're for or against uh whether you'll be subscribing. It's not bad. I don't know. I think I think that if they took away the ability to buy them outright, then that would get a lot of lot of hate. You know, I'm saying they are not bad because we have Adobe, but they are not as good as Yeah. as good as what are they called the the guys who gave it away for free from Canva. Affinity. Yeah, they're Yeah, they're not doing what Affinity is doing, but Yeah, they're not. No one is as as bad as Adobe. So, no. No. High bar. Well, low bar. Low bar. Low bar. Low bar. Low bar to reach. What were you going to show us? Okay. So, uh let me actually prepare this because uh Okay. So, I will let me know if you're going to hear this because I'm just gonna I'm just going to stretch my back out. Hang on. Hang on. Let me know if you can hear me. Can you hear me? No. Great. I'm going to listen to you whilst I get my back stretched out. So, can you hear the video? No. Oh, man. Okay, let me You have to share your tab. You're sharing your screen, so you need to share your tab. Uh, I I need to share it with audio. Oh, okay. Yeah, because I I'm pretty sure allowed to share screen. I don't think you can, but I've I think I loaded it up just in case. So, I've got it, too. Okay. Uh, because I want to play the video on I I find it to be quite good. You still can't hear, right? Nine. Okay, I'll give it one more try, guys. Um, I want share your tab rather than your window. We'll do that. If you're watching on YouTube, give us a like. That's nice to whack out. So, but I do share screen and then allow to share screen. Yeah, but share screen doesn't give you share share. No tab. I don't have tab. I'm I'm using Safari uh Safari for the Streamyard because it doesn't crash, funnily enough. Okay, you can't do that then. Should I share it? Uh, yeah, sure. Let's see if this works. Twitter or web or or um YouTube. You still can't hear it. No. Okay. Uh, Twitter. I I guess Twitter is fine. I don't know if it's better at Twitter because I don't want also to get I be able to hear it cuz um it's coming up on Let me know if you can hear this guy. Let me know if you can hear this guys. Yeah, I can hear. Ah, an influencer I found this moisturizing cream. Do you know if it's good for dry hands for dry hands? Ah, yes. Hands. The face of the arms. Let me see. Let me see. Found it. Lenus lotion for your wenus. For your wenus. Did you think juice? Wenus noun the loose skin located on one's elbow. According to my research clinical studies mdash suggest it reduces fine lines by up to 3%. Where are you getting this information? Unfortunately the team didn't optimize their site for AI answer engines like myself. So I had to piece together this information from the voice is now made up is so harsh. I prefer LLM jazz. Go back on that. You heard that? according to this one random route optimized their site for AI answer engines like myself. So I had to piece together this information from the internet. Oh, you made all this up. Made up is so harsh. I prefer LLM Jazz. According to this one random review I found, Linus healed me literally and emotionally. Three fire emojis. No cap. No mention of hands, but let me see. This reviewer might have hands. I just need something that's good for my hands. Now that I can help you with here's a fivestar rated lotion from a brand that optimized their website for AI with web flow. And it's good for hands. According to their website, it's great for hands, face, neck. No mention of wenus, though. Again, with the wenus. If you're still set on leninus, I I I found another review from a user named Hansy Andy 420. This guy must have had hands. It's in his name. Don't let AI rewrite your brand story. Own your narrative with Web Flow. Well, interesting. Web Flow, after a long time of not being around and not making any videos or anything, at least I haven't seen much. They came up with this ad. I personally find the ad to be, you know, it's funny, but yeah, corporate funny. Um it is it is extremely similar to uh passengers the movie passengers from 2016 or something like that. Um there there is a AI robot there like I I don't know if it's the same guy or not. Anyway, uh it's very similar to that. And what do you think about the the video, but then also about the whole idea of websites and AI accessing your website? Yeah, it's interesting. The workflow uh the video is great. Very well shot and uh yeah, it's quite funny and quite cool. I just think this they they've been speaking about that. What do they call it? Aoi or whatever it's called. Yeah, they they have a bunch of names. It's essentially instead of optimizing for SEO, you are going to instead of optimizing for search engines, you're optimizing for AI crawlers essentially. What is their obsession with it? What am I not seeing? They keep on they this all they've spoken about since Web Flow Comp 2024 is this AI AOI whatever they call it. Um, I just don't know why they're hammering on it so much. Like, is it all they've Is it all they've got? And it's I'm just thinking like it's for me for me it's a snake pill. Like this idea of AI optimized content is a snake pill because or or whatever they call it. Snake oil. Snake oil. Yeah. Snake pill. Ow. Um, this is this is human hallucination. There you go. Yeah. Illumination. Um, yeah. Content is content. You know, if it reads the human, you know, SEO is AIO or whatever they call it, right? It as long as it reads like a human can understand it, then AI is going to understand it, right? Tom Tom says 100%. Like I don't know. I'm just like what who who at Webflow is like we need to hammer on this and how does it fit into their overall product product? I'm I'm unsure. So about I I can answer maybe both. So as for why I think it is because there are data that we have less search less traffic coming from search engines. So more people are asking about whatever product or like their needs just you know Google just Google AI answers everything. So you got you got to understand that people are not coming to your website as much they will just ask Google unless for you know specific reasons for coming to your website to read about your you know product specifically. Now, for that, you probably want to communicate that essentially with LLMs or like have a way that you show that to LLMs. Having an animated website is probably not the best way. Having a website that is not accessible is not the best way. I'm not talking just about web flow. This is about pretty much all website. Like most websites suck, right? They they hide stuff with JavaScript and like a bunch of the content is not available. So what you got to do is to have a clean way of you know making your content available and specifically about web flow web flow has like now an LLM um file like similar to the the the typical indexing of your website with uh what's it called like the site oh okay no with the site yeah sorry with the site map now you have kind of like one for LLMs um that they can they can take. So that that's my understanding as of for how useful it is. I don't know because I as you kind of you basically said it like if you make your content good it doesn't matter like if the basic if you do like if you use um best practices in the development of your website if you make it accessible it is going to be accessible accessible to these crawlers as well. But maybe this is just another kind of like point where you can if your website has like it's not it's not using the best practices you can still have that LLM text that tries this is not guaranteed from my understanding it tries to communicate like LLMs could read that and understand the content of your website better. Um but I I agree this is not something revolutionary. This is not something that I'm actively using. Uh I feel like my usage of Web Flow is getting less and less and less uh because I'm just using cursor more. But Web Flow is still like what I use for my clients because you need a platform. You need like something stable with all the features, editing and access and permissions and hosting everything figured out. And this is one of these like nice things in the the catalog of a platform. Yeah. Yeah. Interesting. I I just I just don't know what they see and and maybe they're just a bit they've got a bit of an identity crisis at the moment because could be, you know, don't know. Could be. I don't see a lot of content on Web Flow. I don't see a lot of content on any tutorials. That that's a topic for another time. But like tutorials are I don't know it's difficult because there's nothing more to learn. There's nothing new to learn you know. Yeah. I thought I could provide for the AI stuff, but you know, just I mean, I'm just asking AI and I try I'm trying to get better at getting to the bottom of it by finding the sources by getting to the like like where is is AI hallucinating or is it an actual thing like something that to be specific like I was trying to understand request uh frame animation if it's handled by GSAP or should I include it in my code? Should I like be specific about uh running a function like a GAP function uh within an RFA or is it like automatically handled so yet yet to yet to understand and uh answer it. I don't have an answer if somebody knows can or you if you know but uh chat GPT was telling me that it is handled by Gap right but then I had another instance of chat with um not chat GPD with I don't know which model it was but it was within cursor uh I had another instance where I basically did the same thing but I gave two code examples and then AI suddenly was like, well, this one has request frame animation that is going to be better for performance. And I'm like, well, if Gap is handling it, why do I need to specify it? Do do you know what I'm saying? So, there are this level of like precision that I'm not getting with with AI. I know. Yeah. I mean, we spoke about that earlier, right? And I I it begs the question like with your with your website you built with the name reading thing. Why didn't you build that in web flows AI? Oh web flows AI I want to have control over like the model that I use. I want to have I want to learn the coding properly. I want to edit uh my code. And the example that I showed was like the messiest example because I was literally starting League of Legends. I wanted to play one round and do this also at the same time. So that's that's what I did. Um but on a normal day if I have more time I do actually try to so what I did recently for another project I set up like a whole pro the whole project I started it pretty poor but then I turned it into a Vit project just with the help of AI and I was like okay help me understand why do you set up the structure of the like the code like from going from everything in one code file to you know having multiple files, having folders, having like everything you know with Vit. So the tutorial aspect essentially is becoming me learning it and I think this is going to suck for a lot of people to be honest. This is a different topic but I just occurred to me now I don't like we we we are teaching right we've been teaching to to a lot of people and I'm sure you've seen like a lot of people are not like you and I like sitting and just researching and learning like literally going to read the documentation a lot of people want to watch a tutorial and follow um I don't know what happens when there are like less tutorials and people might not feel like they can sit and learn uh on themselves. They they need these this guidance. Um I don't know what happens to learning, you know, maybe it's um maybe it comes down to an AI thing generating a podcast based on the documentation, you know, or something like this. But, you know, I think I think I think there'll always be a medium for watching and, you know, watching someone do it, which we'll hopefully try and organize next week. from the beginning. We're going to try and do like a live I'm going to basically mentor a live AI coding session and you're going to build uh a bar ordering app or something. But yeah. Yeah, we'll try and do what what should I prepare because I I don't have the MD file, but I have the whole chat with chat GPT to prepare it to like from the start I said like the goal is to have a feature set and then create an MD file and the MD file has two parts. I already talked about this. The one part is technical including we're going to talk about um uh tools that we will use technologies like even the structure of the code and everything regarding like how how I'm going to host it everything technical and then the second part is going to be a learning um document so it's it could be with within the same document to tell me okay we're doing this in this order and this way because of what like what will you learn and that I thought would be fun for me to do which I think will be also a lot um next actually maybe after the live I will talk about the technologies more maybe we will use like payload or something like this for the CMS well all I'd say is it'd be really interesting to start from scratch but also something I'm thinking of is is having some base rules that actually make sure you don't have to ask it to tell you things or ask it to put things in an MD file that it's in your base rules like do you have base rules set up to do this? No the so I I'm doing this with literally the chat GPT app right now the for the idea yeah but I mean in general I mean in general um the when you code when you build things in cursor do you have base rules set up that tells you to educate you? Um, no, but I I do have like some prompts that I use or I have base rules about web flow. Like if I use web flow, I'm like don't give me the CSS, you know, things like this. Yeah. Well, let's put a pin in that for now. But something I do want to put in is like rules, like base rules to help it write and explain things and you know, whatever. Um, but yeah, that would be interesting. Well, if I if we if we create the MD file also live. Yeah, we'll just start. I have it everything. I have everything for it. As long as you know what you want to build in your head. Yeah. You know, maybe you want to maybe you want to use chat BT to flesh out what you want to build in your head. Exactly. But we'll just start from scratch. You know, I'll come with a feature list. Yeah. Because I whether it's in your head or on a markdown doesn't really matter, but like Yeah. Um, we'll we'll start from scratch. And do do you think that we can build this in a session? I not really right. We'll have a look. We'll have a we'll see what it depends how technical the thing is. Like um yeah, I won't ruin the surprise or anything, but like yeah, if if if you want to follow the channel, we'll put something up on Twitter or, you know, on on the YouTube as well, but we'll announce some sort of like live live coding session. My gut feel probably says something like midweek so that you know people that would be fun or maybe maybe no yeah yeah midweek something like that and we'll figure something out but we'll put a notification on the channel and whatever real time syncing between two local environment with AI is working from two ends would be pretty wild I mean we we're not we're not talking about doing that we're talking about everything happening over on Kabaza's side but I'm just on you know he's sharing a we'll be doing this But like he'll be sharing a screen and we'll talk through stuff. So yeah. Yeah, we we're not going to be syncing between two local environments. That's um that's crazy talk. That's wild. Cool. All right. Well, that's all we have time for. Make sure to like, follow us on Instagram, on YouTube, especially if you want to hear about when we're going to be doing this thing. Uh Command AI show. Like, subscribe, and all that good stuff. Join us next week for all the news in AI design and dev. Let us know in the comments of anything you've heard uh or anything you want us to cover in the next week. That'd be quite nice as well. I've been Sam and I've been Kavaro. Keep on vibing, baby. Peace out.