#56 Fable 5 is BACK and How We Like to Run Our Agents Remotely

Not not that. So tell tell the viewers what you just told me. No, no, no. Like the the Yeah, actually I want to talk about it on the broader sense. Like would you do you still care about Webflow at all or Figma or Framer? And I'm not just talking about like why why of course like we are okay. So I had this ready you it's not in our docs but this is like this plugs right into this question today I've been on the side uh searching things about uh vit about nexjs about um headless CMS um like open source projects what's out there like I'm kind of like I I want to try it. I want to see like what can I make with them because I see that we have this potential to capture kind of like a new market like building websites for let's say smaller clients. Should we do it like in Webflow and do the whole Figma thing and you know the the whole branding thing then Figma then Web Flow then CMS like do we need all of that or can we just you know do something quick and something that can be done really fast and good enough? I'm not talking about I'm not talking about like enterprise level of like building websites that are systems that are like you know for made for campaigns and can marketing teams access like if you want to create I'm giving you a scenario there is a there there is like a small uh small company like a local business doing whatever like they have a service it's a service-based company. They their website is from 1999. If you know the guy, it's the website. You don't know it. It's fine. The website is really old. Uh it's on WordPress. It looks awful. They don't they will not pay 10K for that website. What do you have a Do you have an offer for them for 2K? Can you do something proper for 2K? No. No. Why not? Proper like as proper as like you'll get something that's AI generated. Yeah, that that's proper for them. Like look at what they have. They have something that is like from 2002. It's preWordPress even. So to to to bring it back to the Did you mention re there was a lot there. Yeah. Yeah. Yeah. No. No. I didn't mention thing. No. Oh, no. I didn't like So, the this ties into my answer ties into it a little bit. Yeah. Go on. Tell tell the tell the gorgeous viewers. Um, so so I said like Room has the the new MCP and is this relevant? Yeah, that's kind of the question, right? It it has like you have access to tried and trueue components. Yeah. Does this matter anymore? Yeah. If you have So, obviously, obviously, let's go [ __ ] hard right now. Let's cause some controversy. Right. I actually don't know if I can cause controversy. I I'll just cause controversy for No, no, no. You can cause controversy. It has been tried. Then so with regards to reusable components, right, this idea of like shortcutting into success that's completely been democratized when it comes with with regards to AI because all of that information is already in, you know, in AI, right? all that base level knowledge that a that average nor of averages like nine times out of 10 we're not designing websites that reinvent the wheel. We are leveraging most probably something that's been done before. Not by not intentions. They're just there are just a lot of things that we we know that work. So why not just re reuse things that work? Okay. So reloom are dead because like you that if if if we're talking about this one like um aspect of the reusable components because people can just like use AI to do that. The reason why I asked you to bring that back in when you asked about like is this proper, it's like you either go the AI route or you hire someone to like do it properly. Like you really you try and figure out you might they might inevitably end up doing something that's been tried and tested before. Maybe they'll use AI to you know to figure that out or come to that conclusion. that let's say for example the output was in a Figma file or something like that where they've they've crafted that and they've got the evidence to back it up and it's not just they one day they had no designs and then next day they had designs right um so like so it comes back to like that are gone I don't think there's any need for but do you think I I'm Not sure if I can agree with that because I think it comes back to uh people who know how to use who who know like what's the underlying I think methodology the fundamental fundamentals I'm I'm talking about I'm not talking about what is HTML what is JavaScript type of fundamental I'm talking about like like a website um analyzing like a web page page and understanding what each part should do, especially like plugging it into the business and strategy. And most people are not that. And most people who have websites that are five, 10 years old, they don't know that. And they you there is always this thing that you can ask AI, but you understand basically on your level, right? your understanding falls back to the level that you are currently are like you can literally ask AI to explain quantum mechanics to you but you will understand as much as you are ready to understand right so how does that tie to these tools I think there there are tools that are essentially harnesses around these problems for example the problem is creating a proper landing page and you could see Reloom as a harness around that as a tool made kind of like in a linear way to get you where you want to be. I fully agree with the idea that AI you can just ask it to do something and it does it but it always goes back to if you are not a pro how do you know that that result is good enough? But if you use let's say a harness or like a tool that you are you can trust maybe I'm not saying like just trust it um blindly but if there is the trust it's easier to make sure that the thing is um doing it doing its job correctly. That being said I would love to do a test. I just I I think it would be fun to see um maybe maybe I find a like an ugly website and then ask maybe Fable to redesign it. Um ask Claude Code yeah Fable to redesign it. Maybe Sonnet and even Opus to design it separately. And then one of them to one of them maybe opus I give it uh the relum mcp and ask it to do it with the rum mcp and do it with the ru um basis in mind and just to see if that creates a better result that would be interesting to me the annoying thing about these just as a side note the annoying thing about these tests because I've been doing some tests today using a new local uh open source model right coding It's just so undeterministic. So like you need to do realistically, we don't have time for this, but realistically you need to do like 10 tests. Yes. And then even then it's like a vibe. But anyway, that's a side subject. Back to the point. I I didn't mention anything about people, right? No, you're absolutely right. I'm not I'm not saying that this company should go off and just try to do it themselves with AI. I'm saying they should absolutely hire someone who knows how to [ __ ] take take [ __ ] down boy take AI down boy control and command AI. Yeah, this you still need someone to know what they're doing. I mean for sure if the CEO wants to give it a go themselves and absolutely but they they're going to find that they um they're probably going to run into those limits sooner. I think that's uh we don't that's we don't need to say anything more about that. But if they were to hire someone, let's say someone with design, like I think I have UX sensibilities. Like UX is a UX is a bit of a dying breed. I was talking to my friend Susie about it last night. Like UX is a bit of a dying breed and that like most things have been done tried and tested. We're going for more optimization. There's nothing novel. There's barely anything that's novel anymore. So UX is really something that's been explored a million times before. It's just about how long it takes to find the the evidence and the the the reason why you go down a certain route there. Just leaves devs and designers who can consume that role. So you know um we'll get into it because this goes right into the open claw uh app and things like that, right? But yeah, I absolutely think that the um uh the the company should hire someone. It's just about how much time they invest into actually crafting something like go into town on their exact needs. It'll be like now we'll give you a quick interview and we'll we'll get some people who know how to build websites and we'll we'll get something and you'll smile. So, so I so uh the reason I mentioned headless CMS and other tools right now was that um I I see like the potential to do sorry I see the the potential to do to do tons of like projects that previously would cost maybe 10k to do uh because you would have to do design and development. Sure. Sure. that now day you could get away with the AI result. And now people in our industry like if you're watching and you're a designer and you you see another uh web website made with AI and you would throw up because you see that left side border uh two pixel added with the um curvatures. You see that you're like no but that's kind of like not the point. If you still compare that AI looking slop to what the abomination of like 10 year 15y old websites that these businesses have the AI slop is obviously better and with just like a tiny bit of like I would say web design sense or skill you can actually get away from the slob and not do what open claud did with their design. I'm not like saying like and you take it further. Yeah. Yeah. I'm just saying like if you if you genuinely just take what Reloom gives you in the design, I don't know if you've used their design thing. I I played with it. It's better than 95% of these like old business websites. Um so all I'm saying there is this opportunity. You could say we still want to ch charge 10k and and do more stuff which I'm trying to do but also there is an opportunity to instead of charging 10k you would charge maybe even 2k and potentially like literally spend one day even less on the project like just prompted away you could do it better but I'm saying like it's good if it's good enough for the business like who cares like if it's good enough you give them a website. Well, so I I I was on a call with Julian from Ship Studio um last week actually and we sort of briefly spoke about this and comes down to a few things, right? First of all, it comes down to how much and this has always been the case. This is no different. How much does the company or how much does the business actually value creating a website, right? if they really don't care and they they they literally do not see any need to spend let's let's go crazy let's go $500 right we've all we've all I think we've all built a website for 500 bucks right but like um they value making a website that little then it falls on to you whether you can bring yourself to build a website personally I could let's say $500 personally I couldn't right but they will probably find someone out there that will put the same amount of effort and energy into someone building a website who's, you know, who would do it for 10,000. Didn't mean to say the results are going to be good, but they're going to put a lot of energy and effort into it. Um, all that to be said, it's just it starts with how much do they value the website and then then the onus is on you whether you believe you can give them a good result for however much money they feel like they want to spend. if they know anyone who's buying a website like just just know that the more money you put in the more you're going to get if if the person you're speaking to has the capability to yes go as far as you push them you can just keep going there's so much more you can add this bottomless right features that that's a really good point and this is why I was looking for alternative CMSs like not use web flow I was thinking like these people who don't really care that much. They just want something good enough, proper. What if I can bring them in at 2K with an automated kind of like half automated way of like making a a website and then uh and then maybe upsell them later on if they want components and uh like they they want multiple languages, they want this and that then I then then I can be like let's bring what we have to a proper platform like web flow where you have like team members to log. Theoretically you could build all of that but you don't want really to get into this necessarily for like a small project but yeah that's the idea to bring people in and if they want something bigger later you can you can upsell it to them. But they're not they're only going to upsell if what you produce for them is is of good quality, right? Just because it's 2K, you know, doesn't mean they're gonna you're lower like you need to be able to produce the goods for that amount for them to want. Like I I've toyed with the idea I've toyed with the idea of oi o I tied toyed with the oier of charging nothing for an MVP like showing them what is what we're capable of before then saying okay cool now but like this is the idea so what if I like I'm I'm literally doing this with like two three businesses right now I'm taking their really not beautiful website and just re asking Opus just asking Opus opening a folder in my computer telling Opus re like read this whole website research it use workflows ultra think remake the homepage properly don't make mistakes it works but remake remake remake the homepage only and push it to my versal all in one go. It does it all in one go in 5 10 minutes whatever. If you use workflows it takes a few minutes more but then you have something that is in every case legitimately like better than like what they have. Yeah. But you got to the I don't know whether you've spoken to these companies but you got to know that that that they value the website. The reason why it might be [ __ ] is because they literally just don't care. It's like, "Thanks for doing that." I mean, I don't know what you've done, but thanks. Do you know what I mean? They could turn I I get it. I I fully get it. And that's the idea. I'm like trying it. I'm seeing I'm sensing an opportunity and I'm trying it with a few companies to see one of them, they cared. The other one, the the form was on WordPress. It was broken. Uh, it was quicker for me to to literally recreate the entire website with Opus and it was so much better than actually going to WordPress and figuring out how I can fix the form. I did it but I did it with help of that was with Claude help but it was faster to recreate the whole thing and I think this is a point that we shouldn't just like miss we shouldn't like ignore I think there is an opp I feel like there is an opportunity there well I mean there's an opportunity in so many different areas just because it's been everything's been flipped on its head the way we so we we can't think about the things the way that we used to think about. We can't think about structuring a project the way that we used to structure a project, you know. Um there are some things that we need to experiment with. There are some things that work because they work. And um so as an example, you know, if you're being asked to design and develop a website, then yeah, you might jump straight into the end result. You might you might skip Figma totally. that there's no need to create something in Figma or you know I don't know I'm just making stuff up but basically like you might charge for different aspects you and I know we've you know we're all part of the same community with flux and you know the future and all that kind of stuff where they talk about value based pricing um that's always kind of been the case however it's been more justifiable or sorry it's an easier thing because you know as much as those guys say value based pricing. Value based pricing. God is it a headache to determine the value and to dig into all that. Sometimes it's like [ __ ] I'm just I'm just give me 50 100 bucks an hour, whatever it is. Just that's just easier whatever. You know, people don't want the headache of figuring out value based pricing. But there's so much more um incentive to now because if you challenge papay the hour and you get something done in a couple of hours, it's really exemplified the fact that those who work efficiently are the ones that get um or sorry, those who work unefficiently get rewarded by charging by the hour because they just take ages to do something else would just do in a second. So, I don't know. It's it's a big old um mixing pot, but uh of opportunities uh and I think we've got a bit of time before we will figure it out, but yeah. Yeah. Cool. Anyway, um we'll see like what also maybe the audience says about this, but we're all trying things right now. Everybody's throwing throwing [ __ ] uh aka tokens at the wall trying to see what sticks. the current tokens at the wall. Before I sleep, I let my agents to work on something that will never see the light of the day. I don't want to give my tokens to anthropic. Sooner or later, it we might we might be laugh. We might be like, do you remember when we used to pay a hundred bucks per month for practically unlimited token? Do you remember those days? And we'd be like, H we hand over like $1,000. Someone posited the idea of a thousand a month um uh what do you call it? Uh subscription for Fable, you know. It's like Yeah, I feel like that's the direction. But if they are selling Yeah, if they're selling Fable like um employee that you can have a thousand bucks, you know, it's getting into that territory. yesterday was totally worth it. Um, I know the our project we're on, it's like, you know, one day I was battling with Opus, getting it to do things. It would miss little things here or there. And I tried Sonic 5, just quickly scraped Sonic 5. Opus Opus 4.8 was getting like 90% of the way there, but it was a bit of back and forth and whatever. Fable just like one shot at a bunch of stuff, and it was just like, holy [ __ ] Like, it took a little bit longer, but it got the job done. was longer. The the prompt was longer, but the back and forth between Opus 4.8 was 10 times long. Well, not maybe not 10 times longer, but like it was like five times longer to do it through Opus 4.8 than it was to even though the prompts took took to took a long time. Anyway, oh, where's he going? I look down. Oh, I'm adjusting my fan. Well, um, how is Yeah, your voice. Can you double check it in I I don't know if you're like if you can check it. I'm going to check it on YouTube as well. Do this. Is this louder? Is this better? No, it's not about the volume. It There is like um like kind of like walkie-talkie at the start and the end of the sentences that you say. There is like Okay. Hang on. Yeah. Right. I'm going to talk for a bit. Is it happening now? When I start and stop talking, can you hear the walkie-talkie bit? Yeah, you can. Yeah. I I thought it was cuz I had my fan on, but I've um I've turned my fan off now. I I'll mute you for a second and I will check uh YouTube. and stop talking. It comes through YouTube. Yeah, I know. Yeah, I don't I don't I don't doubt that it comes through YouTube. Um uh where are we? Sorry, I'm just saying nonsense. I'm I'm slowly getting used to a new setup and I'm running my audio through OBS. So, I'm I'm thinking instead of like going to my Twitter, maybe we go to my Instagram sometimes. Well, be quiet just for a second and we can discuss your Twitter Instagram thing. I mean, we get a lot of attention on Twitter, right? Just listening out now just for this crispiness that you talk about. I can kind of hear it, but I don't know. Um, I don't know the values on the compressor. We I'm struggling to hear that. That would be that would be noise that's coming through the microphone. Um yeah, it is it is some noise. Let me try something else then. It could be electric static from Right. Is that better at all? Yeah, I I do think it is better. Yeah. Check one, two. Yep, it is better. Okay. It was cuz I was charging something. Okay. Okay, cool. Uh do you want to do the intro real quick? Let the let let the people know what's uh what's what they've got to look forward to today. Yeah. So this week uh we have open claw on mobile. We have actually the most important news is our favorite model ever. Fable 5 is back. It's a bittersweet. Yes. A bittersweet because not for long until it will be gone again. We'll talk about that. Uh Meta is spending billions with a B on just AI tokens to use, not to develop. It's just internally use it. Uh Cursor has now an iOS app. Now you can program on your computer through an app. We'll talk about like different workflows how you can do this. Um, Ford in a very funny news rehired 300 plus engineers that they fired in the first place because of AI because AI couldn't do their job. And we'll talk about Nano Banana Light. I hope I haven't missed anything, but we have a few more news. Uh, all of this and more in today's episode. This is Command AI. I'm Kabarza. I'm Samuel. And this is a week in AI design and dev. What's do the fishing thing? What fishing thing? Yeah. Yeah. You the harness. The harness. Like come on. It's not It's not fishing. Well, it could be fishing. I was horse like h No, I was there. Yeah. Control. But yeah, so the the first news, let me actually just open this and just get it over with. I think this is just funny. Well, I've got some screenshots here, but I'm presumably you've got some memes. I I don't know about the memes, but this is a funny one. Yeah. Sorry. Yeah. Say it. Say it. Say it. Is it regarding this one? Open claw. No, no, no. It's regarding this one. So, it's like gone. Go in. Yeah. This is not This is a quick and funny one. Meta employees uh consumed 73 trillion AI tokens in a single month, which is about $2.6 billion a year on consuming tokens. That is just insane how much these companies are paying. So we see news from Tesla, from Uber, from now Meta. Every company is trying to be like hold on uh what why are we using these this many tokens and what are we actually doing with these? Uh we saw Spotify saying that they have been doing I don't know how many uh PRs or whatever. uh they are like apparently very productive with AI but people are like on what like what have you shipped we haven't seen anything so it's a moment again kind of uh the industry is waking up again and be like what are we spending this much money on I mean is this not crazy $2.6 6 billion dollars. Well, I mean, it's pocket change for Mark, ain't it really? I I mean, no, wait. Like, 2.6 is kind of still crazy. Like, even for meta, that that's a crazy cost just added out of nowhere. You know, they're engineers doing stuff previously. are they doing now this amount worth of you know money uh extra work I don't know waking up maybe Facebook Facebook can only go so far in it they're all building well I suppose they own WhatsApp Instagram uh they own so much so yeah that's a lot they bought Instagram for $1 billion and now they are spending 2.6 six billion a year on just tokens. Probably half of it to generate slop and the other half is to summarize the slop. Yeah, that's that's what we are doing. Yeah, it's kind of that's crazy. You did you did see that they were going to be monitoring the the people's screens and learning from how devs work and stuff. Yes. And they said it's Yeah. They said it's in your interest to not know why we are doing this or like what it Uh yeah, there was a hidden phone conversation with Mark. Voice recording with Mark, wasn't it? Um yeah, I don't know. But [ __ ] I Yeah. Oh yeah, [ __ ] Meta. [ __ ] Facebook. Um yeah, I don't doubt they're spending a lot. I know a lot of people who work at Meta to be fair, but they're on Yeah. various. Did you know that they have a brand uh what's it called? um tax. So they have to pay higher salaries to people to work at Facebook at Meta because it's a how [ __ ] the brand is. So they pay above industry standards because people just don't want to work for them. It's the opposite of Apple. Apple actually apparently underpays to to a lot of uh people because people are just like crazy just take us in. We we want to have Apple in our resume. Yeah. Yeah. That's crazy. Yeah. I don't doubt that whatsoever. It's like h but it's Facebook in it, but they do pay a lot of money. So, yeah. Yeah. Yeah. Crazy. Well, um did you see the new openclaw iOS app? The designs, the abomination, they look horrendous. They look horrendous. Well, yeah. Get off there. Get off there. Let me show it. So, here's the Get off there. Get off. Okay. Yeah, I got a better one. I got a better one. Nice and clear now. So, these are the these are the screens. Although, I can't sideways scroll. I was hoping to be able to sideways scroll, but the app store is not. But yeah, this is the AI slop that's been produced. Um, some key highlights here. Oh, here we go. Kih heights being like the double header there. Uh telling you online about three times telling you you're online. Uh what else? What what's some what are some sins we can see here? I mean the awful color the awful red they've gone with. Yeah. The grouping of things are also off. The it's it's just like placement. Yeah. And the typography uh even the font and even like Yeah. Yeah. Those touch targets you imagine those touch targets they are tiny. Yeah. They are tiny. And I don't know like the shadow is also kind of like Yeah. You see the shadow. It's it's an old thing with the button, you know. Yeah. Um and the icons also do really look like AI. And I think this design maybe five years ago would be okay. But today why people are so angry about this uh is because like he seemingly has unlimited AI tokens and this is what they put out and just pe people are not happy with it. Like the logo they have like five different logos. like each platform. It's it's it's not even a logo. Let's what I was just preparing while you were speaking right now. Look at this. Look at this. Right. This is the website. Right. So, you got this font. You've got this little maltbot thing, whatever you call it. And then you go to the GitHub and then you got this this weird like thing. And then and then the icon there. And then going back to here, this thing, it's like they I'm not a designer. I'm not a I'm not saying I'm a good designer, but I don't think you like how much of a complete like people just don't think at all. Like, hang on. That would that would drive me insane to know that I've got a product where there's like five different logos all look completely different. I never know what, you know, if I talk about it, I never know what ones to use on my bloody on my bloody videos because I'm like, which one is it? I mean, this one is just and and actually they use this one, this this lobster dude in the in the Twitter. So, and I we'll talk about this in a second, but like it's just no just doesn't have a clue. Um, and I Yeah, go on. Wow. I was going to take it on to some comments I saw on Twitter. Yes, let let's do that because I think uh Okay, so for people who don't know what's up, this is Open Claw uh launching their mobile app on iOS and Android and the design community especially is furious and I think part of it why they are so furious is because it it's a multiple thing. one side these AI companies have been saying that you know you don't need design is dead like AI does the design and things like that uh on the other end I think it was the open claw uh creator who like Peter I think I I I searched for the tweet so I don't have it but I think he mentioned a soft a design software and saying like what's the point okay yeah and people were like really angry about it and I saw one of the replies was screenshots of the app and saying this is the point. It it's kind of like Yeah. On that though quickly like are people going to use it? Yeah. Like what like what's the what's the worst that going to happen or you you upset some people? Like it's not going to inherently stop people from using the app. Like there's there's a bit of an argument there to say that is is it that bad that people going to like oh no I'm not I mean there will be absolutely people who go I'm not going to use that cuz it looks like ass. It hurts my eyeballs when I try and use it. But there are going to be plenty of people who still use it. They're still going to make a not directly but they're going to still make a fair bit of money uh people using this app. But check this out. This is a tweet that I saw today. Open claw. Just gonna leave this here for the haters. And like it's a tweet here. I'm gonna open up in a new tab because we're gonna hit we're gonna go back to it. Colin. Look at him. Look at Colin here. Okay. Hi, I'm Colin. Hello, Colin. I did a lot of the open claw iOS UI. So, as one of the volunteers responsible, I accept my public flogging. I'm just a maintainer with limited time. Too much excitement about mobile agents. Roasts fair. Have a dialogue. tell me what's how bad it is and want to make it better. Come join me in the Discord so we can make it as better a community. Right? So, but looking if you think about this tweet like yeah guys stop [ __ ] on us. Here's one of our volunteers. [ __ ] on him instead. Uh this one kind of sums up throwing a volunteer under the bus like this is wild. Like it's like it's pretty I think this is pretty bad. Like I I I like the sentiment of this. It's like these are volunteers. They don't necessarily know uh what's good or what's bad. They're not being paid or anything like that. So I I I totally understand that. Um but and I also saw a tweet as well about someone being like, and I get this on my videos. Um I didn't know what an acronym meant on one of my latest videos, LRU or something like that. I just didn't know. So, I defined it and it was like last uh last reused uh I don't even I can't even remember right now. It's not even important, but I had someone in the comments say vibe coders. Do you know what I mean? Like calling me a vibe coder because I don't know what LRU policy means. And this comment that I saw against the sort of, you know, someone complaining about the design, it's like, oh, like uh again using the vibe code thing as a as a as a uh a reason or excuse or whatever. And it's like it's just because the the design looks [ __ ] doesn't mean the app is [ __ ] Basically, I've done a terrible job of explaining my point there, but it was just a comment that I saw where someone was using the front end as an excuse to blame Vibe Coders and also assume that the the the code was alit. It could be a solid app. The code could be absolutely phenomenal, but just the UI just looks [ __ ] And and one thing and another comment I saw actually was about someone saying Apple letting this app into the app store is wild because again people are locked into this idea that open claw is insecure and stuff like that. Um, actually it's just it's just a it's a remote control into your own computer. Like the app itself isn't flawed. Like the app itself or it might be, I don't know, but the app itself might be solid. The code itself might be solid. Doesn't represent the insecurities of open claw because obviously people just want to bang that drum over and over again. So a lot of a lot of misunderstanding floating around about this app, but there's no doubt about it. The UI looks so shite. Yeah. Yeah. At at the end of the day, this is I know it's not directly open an open AI project, but they did bring in Peter and yeah, it's seemingly having enough resources to hire a proper designer, not even like an open-source maintainer who might be I don't want to say like that was part of the deal though. That was part of the deal that it remains open source and it remains fine that that's fine but they they seem to have like enough resources and I don't know like the design is it it's really underwhelming. Uh some people started like mentioning uh some people started doing like redesigns and the redesigns were much much much better. Yeah. So on that quickly then I saw this video from design course. I don't know if you know this guy. these long heads. Yeah. And so he takes it and just I mean I'm not that impressed with his redesign, but he just he just takes it through and shows how you can, you know, more of a UX understanding about how you can kind of break things up. You don't need to have like online like 10 times and and whatever like so it's a nice little run through and and this is his kind of like his favorite thing. Cleaner, you know, I I like it already. It has better contrast. It may maybe the branding is not like too distinctive at this point, but on the left it's very old looking. You have that double thing and you have like extra space unused and it feels like the extra unused space is wasted. While on the right one, you don't feel like the extra space is wasted. Um, and yeah, the tabs are easier to to see and click on. and a bunch of things like this. But I do think none of the icons are good like in both examples. Um like you can go with such high quality like really good icons and I feel like that's a big part of you know making it better and maybe with the colors I would probably not go with green. I don't know maybe it's green it's thing. Yeah. I mean, at the end of the day, it's V1. You know, they've got an opportunity now to take this and and make it better. But just like ju this this this argument or not this argument, but this kind of it's not a massive surprise that you've got a bunch of devs who are [ __ ] at design. This is not a new conversation, right? Like again, I think back like the developers just don't know about design. like an AI gives them superpowers to be able to at least throw something together without having to bring on a designer or have to spend time doing design. They're able to get get V1 up and then address it at a later time or actually get someone in or whatever. Similarly, like the flip the tables a little bit, ask a developer to sit down and review a designer's code. Like I bet I bet it's going to get the same. It's just because it's less visual like yeah you like no offense your code is probably shite. Yeah. It's like my design is shite because we just don't know these things right. So, uh, I'm not defending it at all, but I'm just saying like it's just so easy to stamp on something where if you take a moment to think about it, it's like, you know, people were just people, we'll get into it with regards to cursor app, which is next, and this idea of only releasing on iOS, but this is the industry, this is the um world we live in right now where it's like, get something out, MVP, get it out and get some feedback and then improve on it. The most important thing is to get something out because uh the the strategy of the waterfall strategy of having something in secret for months and months and months and months and months and then releasing it and then it falling flat on its face like you know that didn't work. The the agile strategy works. So yeah, anyway still to poke holes. Oh, that was the final thing I was going to say. Um, it was based on this the tweet where uh Open Claw threw Colin under the bus. Oops. Um, and I think, sorry, let me compose my thoughts for just a second because it was about them and I feel like Peter's like this a little bit. I don't know him at all. Gosh, like I'm literally just speculating right now, but he was a bit wingy when people were slating on open club and he was a bit like I'm just a guy um you know I did it for free and this and that and it's like I I thought of a quote if we didn't have criticism then we wouldn't have innovation like you need to be criticiz like it needs to be criticized like you can't just you know fall over and or or call it it was the word haters that really got me in this instance where it was like don't call them haters, call them people who care about their craft and they criticize it when it's done wrong. Take that criticism and do something about it. Like I hate this word haters because it's just it's almost like ignore the haters. Don't worry about what they say. They don't you know it shouldn't get to you. It's like no no it should get to you. These are people who care about the industry and they know what's up and it's not something to be ignored uh if you care about your tool. So that was the last thing I was going to say about it. I found it. I found the tweet. So here's why people uh especially the design community was so mad at Peter for that abomination of design that he dropped. Um, so here we have Andreas uh saying Apple acquires play. I don't know what that is. Uh, design mode in Xcode incoming. This is to do apparently to do like proper design like how you do it in web flow with vis in a visual way to do it for mobile application for iOS application and that that was the news and Peter uh replied this would have been amazing a few years ago but in the age of AI what's the point and a reply from Josh was this was hilarious. Lol. This is the point at this. That's good. This was very good. This was really good. So, Josh, good job. Good job. But again, being the [ __ ] he always plays devil's advocate and all this sort of stuff, but Peter doesn't have the comprehension to know like there's not enough information in the tweet. everyone else can see it, but I honestly think Peter doesn't he's he's and there's no I'm not I'm not um trying to be mean to him or anything like that. I just don't think he he's able to see from the outside what the outside sees. I think he's a I think he's a developer. I think he's a bit autistic, a bit developer autistic. Do you know that meanway? Yeah, but it doesn't mean that you should be ignorant of good design. Like it's so obvious how good design matters. And by the way, by the way, shout out to Josh. I have his course. I haven't done like too much in the course because I should do more. But like this guy, he knows his craft. He knows design and he he's not a designer or a developer. He I I would I would say he's a a design engineer. probably my favorite design engineer. Uh he has interface craft and look at this. This is amazing. Like this is absolutely amazing. Um he has uh this course where he shows how also to collaborate with AI um all of that. So he's not like an anti-AII guy. He's just like somebody who cares about like deeply he cares about design and engineering. Something that goes hand inand and this is where I'm I said this last week design is design and code like Figma sees design as a Figma sees code as a subset of design but I don't know it's I feel like they are the same thing. It's just code is a way of expressing design and you can be you can you can do it properly. You can express design communicate in a proper way and yeah anyway shout out to Josh. Yeah. So on that, you know, how he should see that design is important and stuff like that. And on the one hand, I'm like, yeah, he he kind of works um Peter now, he works in an industry where there's we really only care about a few things, right? It's design, it's development, and it's probably branding and marketing, right? Um we only really care we haven't got a a great number of things to really care about, right? So yes, you're right. In the few things we have to care about, design should be in one of those things. And and design matters to me as well. I like to I like to air on the side of caution of like I I do care about design, but I also want to be a realist and want wants a design to go as far as say maybe the budget or the scope of the project or the the business goals allow. That being said, I saw a video today from a friend of mine. when he was on a podcast and he said without brand like you can have the best product in the world but without good branding it no one like it will fail. Yeah. And and whilst I agree I think a marketer will say something different. A marketer will say without a good without good marketing a a [ __ ] product will do really well if it's got good marketing. If you've got good product and [ __ ] marketing, then it won't work. My point in all this is that you as a designer, you're of course you think everyone should care about design, but the truth is people do not care about the same thing. They think they're the thing that someone who who um I don't know, [ __ ] plumbing, a plumber, they think, oh my like everyone needs to know how to do plumbing. Like like the things I get called out for when there's a the all I've got to do is tighten a bolt under the plump. Like everyone needs to know about plumbing. It's just inherently how we are. But it's just not true. We want people to care about the thing. Yeah. Excuse you. You know what that was? It's It was your [ __ ] Um so I know what you're saying. Yes, he should care about design, but realistically, and I've worked with enough autistic developers to know that they do not give a [ __ ] Yeah. In the age of AI, there's more excuse than ever to really just give it an extra push and be like, just do a good job on this one. Do you know what I mean? But point is, they just Yeah. Yeah. It's kind of the old saying, Apple showed that design matters and Amazon showed that design doesn't matter. Exactly. Exactly. You know, it and it all comes down to again why I think I'm a realist. It's all comes down to the business goals and what's what's effective. Like Amazon made it work. You know, it looks like [ __ ] You know, there's a better ways to do loads of things and it's just like it it it would be a wasted effort to make sure that's designed well, but you know, at least it works. Uh let's move on to the cursor app because we could blast through this as well. So in the name of building apps, cursor also released an app for commanding their agents over in cursor and I thought it was quite a nice little It looks a lot by the way it looks a lot smoother. I don't know whether we can that it works. I can then read the code. This has taken ages that screenshots and when I'm ready merge it from my phone. I usually do these extremely long voice notes where so it's probably as you expect. You can just control cursor through your app. You can control your cloud agents which if you use cursor you'll be aware of how cloud agents work but you can also control the agent on your machine as well which is really cool. We've already got this in claw code. Um and what else was it? Um, yeah, we've already got this in cord code and it's just a really clean app and to celebrate it, they're given 75% off of Composer 2.5 through to July 5th, which is in two days. So, all that to be said, yeah, all that to be said, we had a good we had a good chat before the show, didn't we, about how we're running remote agents and because I watched um Theo's latest video with him talking about Linux and in that video, he doesn't really talk much about Linux at all to be honest, but he shows that he's got a plethora of like uh MacBooks, Mac minis, laptops, and whatever. in Claude. It's the Claude Cloud thing. Well, let me show you. Let me show you actually rather than because rather than getting you imagine it, but he's got all of these different computers which allow which have different agents on it. So, he's got a Mac Mini with open claw on it. He's got a Linux box with clawed code. Uh is it Clawude Code? I think it's Claw Code on his Linux box. Um he's got a bunch of agents he's able to do things. Um, and he demonstrated that simple things like mpm install happens in seconds on a Linux box rather than multiple seconds like 30 seconds on a on a Mac. I think it's got something to do with the permissions of uh Mac OS and it needs to do something to the files. Long story short, he kind of prefers to do things on on the laptop. Literally, I watched that this morning. I woke up. I was like, "Shit, I reckon I could figure out something cuz I've got this bad boy. This is my open crawl laptop that mostly sits around doing not a lot. And I was in a situation last last night with my mate where we were hanging around in an office quite late at night and we both wanted to go home but we were waiting for my agent to finish on the laptop that I'm streaming from right now and I couldn't close it. Like it was already midshot or whatever and I couldn't really close it because I was the agent would stop. So I thought, well, what is there a way that I can just send this off? Because you can use Claude Cloud, which I think Lou is talking about here, where um you've got managed agents and and um Claude.ai as well. Let me show you that real quick. Yeah, you can use Claude code in the cloud. Yeah. So this thing here, so you can just set up a new cla uh is this code? Yeah, this is it. And then I can set up a new session and it can pull from a a repo that I've got or whatever. So I can do that. Um, but that will you I think correct me if I'm wrong Lou Lou, but I think that uses credits on your overage your usage credits whereas I was like well what I've got a computer here sure I can just run it on there. So what I tend to do should I just share my screen and see if we can live demo it? Yeah, I uh not that one. I want to screen. So, this this is currently this is currently my laptop that I'm working off right now. And if I find it because it's sharing, but Double check. Here we go. Right. Yeah. So, this is the this is the laptop and what I can do this is claw code like I'm I'm in that machine right now and I've got tail scale. I don't know whether you guys know what tail scale is. Let's exit completely. Tail zoom in. Sure. There's not not a lot to show you just yet, but um yeah, Tail Scale allows you whatever machine you've got Tail Scale running on and and Theo has this. whatever machine you've got tail scale on, you can just like hop into that machine. You can even open up uh the screen sharing uh sorry screen screen screen sharing app in uh Mac, just the native screen sharing app and it treats it as if they're on the local network together. So I could be in a different country but it's as if these two computers on the same Wi-Fi network. So, what it gives you is a URL. Um, if I go to home agent here, all I've got to do is SSH in into that laptop. In fact, I need to go Samuel Gregory. And now I'm in that computer. I could have done a screen sharing like if you're more UI focused, you could screen share into it. But I like the terminal. And I can just navigate into my sites folder, go into the website that we were on last night and run Claude. And now I can just say like um uh you can see here this was something you told me about just before the call. This RC active. You can have this active for every single session you write uh that you um automatically automatically. So if you just go to config here guys and you go to where do we actually we just searched for it didn't we just search remote you can enable remote control for all sessions if you don't want to do that for all sessions because that might be overkill you can you can have it on default but then you can just go slash remote control y so I could have just I last night I could have just done this and then picked it up on my phone and then just did everything on my phone shut my laptop got on the train and just did everything through the Claude app on my on my phone. So, I probably should might be able to get it right now. There you go. So, it's on there now. It's on there. Yeah. And I was Yeah, it's super useful. I've been using the remote control for quite some time, but I figured out uh three days ago that I can do it by default for every session because I was uh I I was getting annoyed by not having it for every session and I had to do it manually and I checked and I could anyway uh and since then I'm also I mentioned I'm um connected I'm logged in in my versal on this machine. So, if I'm doing stuff with my phone that requires me to to check them visually, uh, I just ask the agent to push whatever that is to Versel. Sometimes it is a screenshot. Sometimes it does, for example, I'm editing a video file or like I'm creating an image by the MCP, whatever. Uh, I'm pushing those things to the Versal so I can check them on my phone. It's it it's just that simple. It's and it all happens automatically and maybe I will create a skill out of it saying a skill for how I prefer things to go universal when I'm on the phone. Yeah, but it's it's pretty cool. Just to jump back a little bit on the old um uh cursor app. So, what you can do on the cursor app, you can it will give you a video. You can ask it to give you a video recording of the what what's going on. You could ask it to send you a screenshot because it will use Chrome, take a screenshot in Chrome. I think you can even record, you know, so you might try that. Yeah. Yeah, you can you you can definitely do that. Yeah, I know. I have like tried uh to do like video. I have to try to like I I do a bunch of things like this. All I'm saying that through the terminal in the claw code app in the app in your phone you don't have media like in the ter the terminal doesn't you know show you media the way you do it you just push it to your versal you have a project to like and then you click on the link and you have the the video embedded there the HTML there the the photo doesn't matter whatever that you want it's it's there and it and it's really fast I I mean these types of uh changes and like pushing to versal directly it happens in literally seconds. But I'm I'm thinking quickly right? So I'm going to do something um load the website and send me a screenshot of the homepage. So this should but how does it that computer and it will load it in. I mean we might be able to even see it in action to be honest. And I can see Let's see. Yeah, but but how will it send the screenshot? It through my app. No, that's the thing that the app doesn't Let's see. Uh browser. Well, let's just see. Anyway, yeah, let's see how it's uh Let me pick in the extension. I really need to name my browsers. Oh, yeah. Um so, um the Yeah, the the what was I going to say? The last thing. Yeah. So the cursor app that you can also draw on the image that it creates which is quite cool and it will um and then you can send it feedback. Oh change this or do this or do that. So that's quite a nice thing about the cursor app. But you mentioned as well that you I've sed in like a big nerd. You can also screen share right and do that. You said about using dispatch which has computer control which can start a remote session for you like that as well. So there's another there's another way to do it. Yes. Yes. Exactly. Okay. So it's it's it's screenshotted it. But can you open the screenshot in the app? Yeah. Yep. Really? How? I can see it. Yep. It's done it the last time. Okay. Interesting. Oh, I see. I have it too. I can swear I didn't have this while while I was using the I was generating images and videos. Yeah. Yeah. Yeah. I have it too. I just checked. I just checked. I have it too. I was generating images and videos and it was like I can't send you uh the assets here through the terminal. And I was like okay just put them on my versal and give me the link. And that was like two seconds later I had it. So yeah. So, if you do it through this, if you do it through this app, then you've got computer use. You could even ask it to send it to you on Discord. That's really fun because Yeah. Yeah. Because I I asked it to do something. Uh, and it did take screenshots for itself to analyze it. And I can see those screenshots on my phone. So, that's really really So, I'm I'm creating a web app from my phone. which is like technically connected to my uh terminal on my computer. That's really really cool. So I think the bigger picture here, sorry I'm facing my phone. I think the bigger picture here as well is that more and more people are thinking about how they can be I mean how they can be more remote with their agents and how they can keep them running and and utilizing their tokens and and yeah just you know I don't know what this means for the human psyche because it's like well if you're out and about if you're at a wedding you shouldn't be thinking about how you should need to utilize your tokens or anything like that. Like if you're out and about with your friends, please be out and about with your friends and not worrying about sending off your tokens. Like be present, be whatever. But there are little occasions here or there. And certainly when local AI becomes a bigger thing, like you might want to send off your local AI to do it and you are completely isolated and free to kind of, you know, keep building or do whatever it is that you want um with these remote systems. And when I was making my lunch just now, I was thinking I think I think the the two computer is going to be the norm. Like everyone's going to be looking for two computers like one for remote and one for home or they just leave it completely at home. Yeah, I don't know. Apple is really happy about that. Either way, um yeah, I I do gener generally really like the uh development through my like using my phone because I'm as you know like throughout the week I'm very often on the road um for short times and it's kind of like really bothering me to do nothing like half an hour of drive do nothing while I can set up an agent and if I need to I can in between like 10 15 minutes after I can send up another prompt and uh continue working. It's quite it's quite useful. So yeah, I might have to swallow the costs of you know, let's say I'm on the tube and I'm not about to maybe I don't have my laptop with me and I idea strikes and I'm like [ __ ] I know I want to kind of send something off. I might have to use a remote session that pulls down a git repo and etc etc. But if I have the capacity to like yeah set off an agent on Claude code like Claude Max or what with my Max subscription then I'll I'll kind of do that. But it's I'm not the same like I I don't I don't need it as often is what I'm saying. I don't need to be able to be kicking off remote sessions as often. But it did occur to me yesterday and then Theo's video dropped and I was like I reckon I could do something here. I reckon I could do something. And sure enough it took two seconds. Yeah. You can literally even just go to the gu. Like I would go to the garden uh to my garden and still like have my phone to just to continue the the next um the next task or just check in because in theory I've got the website loaded up. You can see that there. Maybe maybe work that out. Right. This is the screenshot has given me. I could in theory um save this image, draw on it and then upload it as a reference to make you can you can even do a screenshot on your iPhone. Do a screenshot because that will im that that's what I do that that will immediately uh give you the tools to draw on it. You can do it and then click on that button at the top instead of saving it uh say copy and delete it. So you don't add it to your gallery, you just copy it to your clipboard automatically button, which I do a lot. Yeah. Yeah. And you paste it. So yeah. Yeah. So it's uh it's an interesting interesting world of uh opportunity I think. And Yeah. But making websites from our phone. Yeah, man. Because you know, we've got nothing better to do. Yeah. Anyway, what's next? What we got going on next? Okay. Ford. Um, this is a short sorry actually last little point I'll add about the um cursor thing. So I saw a comment I might even have it uh in my browser. I saw a comment here and it will go back to something we spoke about earlier about MVP. Um, god damn it. It's not this. I'm just okay. It's probably on this tab to be honest. I did see a comment. I see a lot of comments actually. Um where people are really It's always iOS first, right? Obviously, these are Android people. Uh very funny. Uh where is it? There's quite a few um people. Why always iOS? Right. And there's a reason for this. I think you and I know the reason when you build for iOS, you've got one operating system. You've broadly got only a few different um screen sizes to contend with, right? They all perform relatively the same. The the bar barrier to entry with getting an iOS, something working on iOS is so much lower than it is to get it on Android. With Android, you've got thousands of devices to get that app up and running for. So, that's the reason why we get things running on iOS first because it's literally it's just so easy. It's not it's not like a preference thing. You you nail so much of the market by focusing on iOS first and then getting the Android up and running. Like, what what so diff I don't understand what's so difficult to understand about that. like how um for better or worse how versatile Android phones are. You get ones that big, you get ones [ __ ] this big. Do you know what I mean? Like you foldables. Do you know? Well, Apple will have foldables soon as well. But at the end of the day, it's not yet, but even with uh their current uh setup, they have better SDK. So, they have always like had better SDK. These are like tools for developers making their lives easier when it comes to even porting an iPhone app for iPad and vice versa. It has been always easier on iOS. So app developers do it for iOS first. That's not everything though. Um, there is another reason and Oh my. Sorry, I've got I've got a call about 3310. Yeah, it doesn't have any iOS. Barry's dead. So there is another reason though and that is uh iOS users are potentially better uh paying clients for the apps potentially. Potentially I mean no like for they they they pay for iPhone they can afford an iPhone so they are probably you can get some pretty expensive Android phones. Yeah, but like technically if you like statistically if you buy an iPhone you probably will have a bigger budget. Maybe maybe people stretch to iPhone. They call it what do they call it? Something luxury, consumer luxury or something like that. Um maybe consumer luxury where people are they believe by having an iPhone. A lot of people do this where they'll buy a Louis Vuitton handbag or they'll buy an iPhone or they'll buy something that makes them feel like they're quote unquote rich. But actually that's all they they buy they buy the this expensive sports car but they can barely afford the insurance is what I'm trying to say is Yeah. Yeah. There are even studies on it. I I saw one that they they did like identical uh Tinder profiles with someone having an iPhone versus the one having a Samsung. And that was not even a bad Samsung. It was like the the highest flagship of whatever. And they had I don't remember exactly, but it was like three times like 300% more matches, whatever. It's like it's just this perceived brand thing that you know Apple has ma managed to create that their stuff are better or like if you have it you are better you I guess you're an alpha male. Yeah. Um anyway that I'll we'll leave it there. Do you want to check do you want to talk about Ford? Yeah that that one is also a quick one. Let me open it up. I don't like to I I like don't like talking about um money and rich and poor. It makes me feel uncomfortable just cuz like I think it's so easy to say the wrong thing. So because I think no one's taking any offense to that because I'm too rich. No, I don't want to I don't want to feel I don't want to don't want to come across or put anyone down, you know. Yeah, but this this is like a dumb thing. I'm not saying this is a good thing, but this is how it is. Uh anyway, so this is also a quick one and pretty funny. Uh Ford rehires more than 300 human engineers after they fired them because they thought AI can do their job and then they figured AI could not actually do their job. So now they are rehiring them. And the memes, as you can imagine, have been really good. You could not live with your own failure. And where did that bring you? Back to me. So yeah, those engineers walking back into the to the headquarter. And so basically what is really funny is like the meme saying like those engineers that now know they even AI cannot beat them. How they will beat them. So it is just a fever that the entire industry is going through which is let's fire everybody because AI can do it all and then finding out that actually you need even with AI you need somebody who knows uh what they are doing in order to do anything. though. Um, so what was the first thing I said when you showed me this post? You I don't know like Poly Market. Yeah. Like so viewers, if you didn't see our thing last week on Poly Market, like I had my skepticism about the fact that this is on Polyarket and I don't like the fact they called them 300 veteran human engineers. Gosh. So, I have just done a little quick Google and yeah, like Guardian have reported on it as well. Return of the Greybeards they call it. Um, so yeah, it looks like it's a it's definitely a thing that Yeah. thing is as well, I wonder if I wonder if there's more to it and that like, you know, maybe they got admin or a marketing to to do their engineering, you know, or whether they just because you still like we like as we've said a million times like you still need someone to manage the agent who knows what they're doing, you know? Yeah. You might be able to five 10x the work, you know, but you still need engineers. Yeah. Yeah. Yeah. Um, yeah. Should we move on to Sonnet? Yes. Wow. I mean, not a lot to be said about Sonnet, to be honest. Wait, wait. Before we go on, uh, somebody uh, hi. See, I I was playing around now. Right. Um, I'm sorry. Uh, say when? When? Say when. Is that louder? Is that better? We'll wait 10 20 seconds for that to come through. Um, so kind of overshadowed by Fable, but Sonic 5 was kind of released without a lot of fanfare, to be honest. Um, and I don't really get it at this point. I don't really understand it at this point. It's sort of like the unloved middle child. And uh, yeah, we'll we'll we'll see about that. Um, benching, you know, better than Sonic 4.6, uh, but not as enough, not, you know, not better than, uh, Sonic 4.8. There was what? Opus 4.8. Sorry. There was something it did better on. Uh, knowledge work is doing better than Opus 4.8, hate, which is quite interesting only marginally, but it's sort of just in this real weird sticking point of like, well, what are you what are you using Sonic for? You're not I know you're not using Sonic. I'm asking the audience like what what are you using Sonic for? Like um I was very much one of those people who really did play around with the different models. I would get I would I would use Opus plan mode where I um I get Opus to plan it and then send it off to Sonic so I can sort of save on tokens. And I think that's fine on like hobby projects and something you're just casually doing, but honestly when I'm on client projects there's there's no room for error, you know? So, you want the best model and so I'm on Opus High pretty much all the time. If I wanted to like do something quick, then I would use Haiku. Like if I say, "Oh, go make this change or like exactly this change or push this to GitHub or get to check something on GitHub where it's really just there's no you don't need any sort of thinking. I'll use Haiku." But when I'm working, Opus is my boy. Saying that, coding is a very very demanding task. And I don't doubt there is this middle point. Maybe it is churning through Excel documents or whatever. Maybe it is computer use that that Sonet's better for or more efficient for um or something like that. But all that to be said, Sonet 5 is here. Um, whatever. Have you used Sonic 5? No. Like for what? Even exactly. Start the dev server. Fable got that. Fable's got that. It might take it 10 seconds. I know. Probably a lot longer than that. Like seconds. Now npm rundev and build. I'm doing them manually. So, I' I've been learning a lot in the past year. It's It's significantly better, though, because you get to actually see the logs a lot nicer, right? It's not like hidden or anything like that. So, something get an error, it's there, copy and paste it, you know. Um, but we did get the return of the the the king. The return of the king. Babel has returned. It's not a fish. It's a horse for a week. For one week. Uh four four days. Is it? It's going to It's going to disappear. Uh sorry, I'm I'm not prepared. Here we go. Fable five. Fable five is here. I love these cute little animations they've done. Oh, no. You can't see it. Hang on. No, there it is. Can you hear that? Yeah. It's my first time listen watching this, listening also to the music. I didn't know to the sound. I didn't know it had sound. There you go. Anyway, so 15 million views. Yeah. Jesus Christ. What? Where you seen that? Oh, yeah. There. There you go. Yeah. Um, Fable I've been using it the last few days and is lit. It is literally god tier. Like I'm it well we haven't we haven't covered it. GBC 56 was released as well under the same um um well it's not been released yet actually. And I think that do you remember when I said to you that the best thing the best marketing strategy you've got is when you get your own government to tell you you can't release it. Well, that's exactly what they did with GBT 5.6 as well. So, we'll no doubt see GBT 5.6 get released under the same sort of thing. But Fable is back and it was truly one of those Oh, until July 7th there. Look, uh, Fable can use up to 50%. Oh, I misunderstood this. I thought they're only giving you 50% of your budget, but you might have seen that there's actually a separate fable bar that tells you how much you can how much fable you can actually consume. But I saw this firsthand yesterday where the day before yesterday, I was doing working on this project and it would take me, you know, I was wrestling with Opus for a little bit. I was, you know, getting the work done, but it took a little bit of nudging here or there. Much better than Sonic 5. Sonic 5 came and went pretty quickly doing this work. I thought it might be at least a little bit quicker, but it wasn't. Well, no, it was quicker, but it just didn't do the work, mate. One shot at it with Fable. Absolutely one shot at it with Fable. Like, it just did it. It just did the work. and just took a while to do it, but I can imagine the back and forth I would have had with Opus is more than worth the wait to to get it done with um with Fable. Um yeah, I have not yet had a moment where I'm like wanting to hair my pull my to pull my hair out. What am I? Yeah. When when you speak too many mixed languages. Um, I have never I I have yet to have that moment with Fable. Well, where where with other models I'm like, why don't you get it? The icons, they they they need to be aligned perfectly. What are you not understanding here? Yeah, it's so simple. Like, why do like why do I need to even explain this to you? Where Fable is like, yep, that that's a task. It's done. It's just it's just done. Um and yeah, I'm Are you done? Are have you used up your complete um h have you like used up FA Fable completely or do you still have some credits left? I've got some juice left in the tank. But you to answer your question, yes, I did. And I immediately upgraded to the $200 a month plan because I wanted more Fable. So, for people who have not got Fable or the the $100 tier um subscription, I saw a tweet. I don't know if that's true, but it pro I I think it is. You can buy the $100 tier, you get Fable for that amount, and then you can upgrade to the $200 and then it resets your usage and then you have I think it did. I think it did. So I I've got extra fable then. Yeah. So yeah, I used the $100 and then it just reset back to zero. But um but there you go. Saying that there are some uh safeguards in place. They've increased the number of safeguards uh basically. So if you do if you are coding because I thought actually quick story time I thought it was just going to get released uh in the chat app because uh in their product release they said um sorry because they teased they didn't tease it but they announced that they will be releasing it and they said for coding and research tasks it will fall back to Opus 4.8 eight. But again, I was with my friend yesterday, Susie, and she was like, "Oh, co Fable was back coding." And I couldn't see it in Clawude Code. I updated it. I still couldn't see it when I went to go change my model. But I typed slashmodel fable and it kicked it into gear. So now, now we're back in business. But saying that the safeguards are in place, but you should you should basically let them know if you're trying to do something. So, I'm guessing a few people might run into any uh might run into blockers. I personally haven't yet. Um the I was going to show something else. Oh, and also there's a word on this availability. So, it's available until the 7th, but we aim to restore Fable as standard part of our subscriptions as soon as capacity allows. It's basically because they just don't know. They're obviously going to anticipate a lot of high demand. Here we go. We anticipate high demand for Fable 5. Difficult to predict. So hopefully we will get it back. But it does raise the question, would you pay more for Fable? I mean, I'm already paying more. I I I doubled my Claude subscription just to get Fable just to get because it was just it was so it was so nice just getting it and getting it to do something. It just doing it. No wrestling around. It just did what it asked. I highly recommend you like to the audience to use Fable connected to the Figma MCP or paper MCP whichever you like and try to do some UI design with it. It's genuinely so good. It's it it has such a good understanding of layouts and probably like spatial understanding and it it just produces at worst like the UI is useful usable like at worst like the worst result that I got from it is 10 times better than the result that I got from the Figma agents that I don't know which model we're using. So it's scary good for UI design and I'm at the point where I'm like I really want Fable to become a mainstream thing where I can do like more design with it. I should continue. So here we have also Lou saying uh same I upgraded last night and yes it did reset when I upgraded to the $200. So there you go. Do this. Buy the 100 first and then once you uh used all your credits, then upgrade to the $200 and you're going to eat through it quite fast. Um I'm at 88% and now I'm in this state. I'm like, what can I do exactly with 12% because I don't want the job to get, you know, half done. So, I'm trying to figure out probably um to give it some tasks to do it. Yeah, that's it. Uh you want me to continue? I can't hear you. No, no, I I can read your lips. Uh you're saying, can you hear me? Or like do you hear me? No, you just say just one second so we can guess what you're saying. No, I don't hear me now. Oh, you can hear me now. You can hear me now. Okay. My microphone was about to run out of battery, so I've changed it to cabled usage. Um Okay. Yeah. I um saying that whilst you were talking as well. So, I've you might have seen on my latest video that I've gone for a new kind of design of my thumbnails. It kind of looks a little bit like Pornhub. I've just had second thoughts on it, but it still looks cool. It's a yellow and black thing. Um, but I got Fable. I got Opus and I got Fable to design. What did I just Should I just show you? Shall I show you? I mean, it's not, you know, don't get me wrong, guys. This isn't like [ __ ] It's not going to blow your absolute tits off. Uh, saying that, I will change it back to the um Oh, I I see that. Yeah. Uh, I see. Right. I stop screen. But why yellow? They call me well cuz fable said so. Ah. Uh here we go. I mean you had that uh previous this is what this is what um Opus came up with. Sorry it's not sharing this. What you got? Uh this is what uh Opus came up with. Come on. Right. And I I you know I didn't I told it I gave it a basically I just wanted just to come up with something. I didn't want to like go for any kind of like round the houses and like oh it means this and that and the other. I just said just just [ __ ] do something. Here's my website. Just go do something. So this was like think I didn't mind it. I didn't mind it. Um read and then this was fable. So, it's not and I don't I don't like like there's too much here too much, but I did like I did like the yellow and whatever. So, it's a bit better. And then these are the two thumbnails I made with this. So, this is the one that looks a bit Pornhubby to be fair. But the reason why I showed you this is that if I'm going to go down this route, I might get Fable now to redesign my personal website. Um, which is pretty basic to be honest. and just apply that design again whilst we've got Fable cuz um you know it's it's only here for another couple of days. Yeah. Yeah. And hopefully they will figure out a way to to keep providing it. Yeah. Yeah. Now I I get it. Yeah. You should you should ask Fable to redesign it. Yeah. Yeah. So yeah, make the most. Um, but I don't think there's anything to more to be said. It's fable. Go and use it. Like it's it's it's genuinely a step above. Like it's just it's it's a fantastic model. It it it doesn't disappoint. It doesn't really disappoint. Yeah. I'm curious like I'm curious to know what 56 how 56 comes out. That's all. Yeah. Let's stay let's stay on Claude as we're on a roll with Claude. Uh we've got Claude code. um hiding proxy requests and tracking your usage and also Nana Banana light coming up. But um basically Claude Claude have been a bit sneaky the last few months. Now they've had some issues with the old Chinese open-source labs uh stealing and and and uh what would you even call it? Training off of Opus or Claude models really. So there was an investigation done uh by a guy who is this chap I will show you uh the where he took a look at some really interesting things and I won't bore you with it but um it's it's an interesting technique called steganography where it's It's it's a technique to use to hide data in plain sight. So, uh he noticed that there are a few checks going on. First of all, your time zone, whether it's, you know, in in these sorts of time zones over in China, but it would these are the two things it's trying to change. It removes uh or changes sorry the apostrophe on today's and changes dashes to slashes. And these are the indicators that Anthropic can use to know that their models are being used uh by employees of these open-source um uh labs. That's crazy. Yeah. Really, really interesting. So like they're checking the base URL cuz we'll get into it but I change the base URL quite a lot and I use a lot of my GLM subscriptions, my mini map subscriptions and things like that. So first of all it's just checking that then it will check the What is the base URL? The base URL is it's so it should be api.anthropic.com but you can override this and and you can and this isn't illegal. you can override it because you might have a company where you're running Claude off of Bedrock which is AWS's infrastructure. So you'll change the URL to run on Bedrock but you can also change this to point to Miniax M3. That's how you do it because I haven't done that's how you do it. Yeah, you just you just set this you set this in your settings and it will just and Claude will accept it and just use that. It's not there's nothing wrong with that. But then it goes into these additional checks to check if the time zone matches. Um it also checks uh so yeah see so it's it's checking for those Moonshot which is um GLM and all this kind of stuff. Alibaba which is Quen um but also just does a few more little checks here or there. And yeah, um the the issue being and and by the way uh confirmed this this is an because uh someone reported on it here uh breaking anthropic has been hidden spyware. Uh this is an experiment we launched in March was meant to prevent account abuse and authorized unauthorized resellers and protect against distillation. So he's confirmed it. They were absolutely doing this. resellers. I is there like a I I I heard like there is a black market for tokens. So yeah, like can you buy clawed subscription for cheaper? How does that work? Well, it would just be a case of giving people your API key, right? Giving them access to an API key and then they've got access because the thing is uh some But who's paying for that? Well, I mean, imagine again, it's like you might find a I might get the $200 a month plan, but might actually be okay for two junior developers. So, it's like I'll give them the API key. So, I don't think there's nothing that's probably against terms of service, but I don't know. But it just all maybe it's just better better for them to understand the landscape of things. So, um yeah. Yeah. Um yeah, the issue being is that uh that the the fact of the matter is that they're putting tracking uh telemetry and things like that in the API because obviously before it reaches the model, it goes to anthropics API and that's where a lot of the data collection is and a lot of the quality of life things like memory and stuff like that are collected in that middle API layer. And when we're sending off, you know, I'm sure you you you do it. You'll give it an API key to put in the EMV file or something like that. Like I try not to do that, but we are and a lot of people are chatting personal information and things like that that you don't you want to feel you want to know you want to know what anthropic are tracking. If they're going to track, you want to know what they're tracking. And I don't think this was revealed, you know, I don't think this they were honest about this. And yeah, you know that that this is this is um what's the was it Cambridge Analytica, not to the same extent, but it's like you know, you say one thing, but you're actually doing another. I think that's where a lot of the the annoyances come from. Uh so just a short note on reading your environment file N file like clock doesn't do this by default right but can you tell your agent to do it or like can it do it and tell you I didn't do it? Yeah. No it it should it should respect by default if you go and deployed code it should respect your git ignore file. So you have in there node modules, you have in there your EMV files, right? So that would respect if you don't have a git ignore, then it there's probably some training data to say don't look in it, but I've had it look in like EMV file. Yeah, I I remember like I had an instance it was telling me uh I'm not going to look at it. I'm creating a duplicate of the env file and I'll show you like where to go and where to put your stuff. And then I had some difficulties and it was like I'm basically the model being done with me like I just did it for you like Yeah. Okay. Yeah. But like you needed to give it the API key for it to put it in there, right? Yeah. Yeah. So it's by that point it's passed, you know, it's it's in the model. But yeah, so yeah, flip of a coin. I've had it ignore it. I've had it look into it. you know, it looked into it last night because I was like trying to figure out some stuff. Um, I don't think I even prompted it to look at it and it did. So, it's definitely it's definitely going going through their thing. Um, but yeah. Yeah. Last one. Anyway, last one. Nano Banana Light. I don't have a lot to say about it, but we're going to do a little uh test. Um, believe it or not, Gemini and Google are still pushing new models and they're actually okay. Like I think their multimedia models are pretty good. So they've just released Nano Banana 2 Light. So to to to confirm what we have, we have Nano Banana Flash, Nano Banana Pro, and Nano Banana Light. And truth be told, they have those are the three kind of categories of models. That's this that's their Opus, Sonnet, and Fable. It's the Light, Flash, and um uh Pro, you know. Um and these are the super fast ones. Also, sorry, I'm not sharing my screen. Ball bags. Uh so two also Gemini Omni Flash. Now Omni is their video model. So again, this is flash, so it's above light. Now, we already had Nano Banana light, but now we've got Nano Banana 2 light. Yeah, quite a confuser, so um yeah, introducing Nanobanana 2, our fastest, most costefficient image model Nanobanana family yet. Built for high throughput speed and scale. Uh Naban 2 light is available today in all of these and also uh Gemini Omni Flash is available to developers uh all of the same areas Google AI Studio. So pretty cool. So I thought as something fun I thought we could give it a go. Um I don't know. Let's go on. Let's let's take my latest thumbnail. Oh, what should we do to that? How should we change this? set my laptops on fire. Yeah, sure. Okay, let's see. That fits well with Europe. Tops on F. So, this is quite challenging actually because uh Yeah, they are. Uh they're out of focus. They're blurred. And let's see how it does it. This ain't Flash. This ain't Flash. Oh, wait. You're You're not sharing your the Oh, look at that. Well, that's good. Anyway, so I'm just going to refresh and I'm going to start again. Uh, where do I do it? Here. This is what I was showing you. Okay. What should we do to this thumbnail? And we said set them on fire. Set set them on fire. So let's how how can we I have tried this and it just did it in one shot. So um edit the image so the laptops are on fire. How do we change the model? Um, yeah. How how on earth do we change the model on this freaking thing? Models image. Here we go. Right. Nanobard 2 light. Let's give that a go. Run. See how fast it is. Oh, okay. Good. That was fast. Yeah, that was really fast. 5 seconds and you can see it's kind of it's not really done anything like it. It's quite fast. Like if you have like it's kept the arrows. They were quite tricky. Well, that's the point. Oh, I don't know what that is. Little glow thing there. Is that in the original? Oh, it is. It's the It's the It's the shine on the laptop which is not understood. Um Your face is red because you're also like it's quite hot. It's quite hot, isn't it? Yeah. Yeah. Yeah. Yeah. Um now, okay. So, now can we Yeah. Should we download that and then uh do a new session and try on the fire? How do we How do we How do we do a new thing? Ah, here we go. Okay. Man, I do not know how to use this whatsoever. Uh, already in a new chat, but I want to I want to change tools. Maybe you need to. How the hell? I just see the thinking level that you can change, but I don't know anything else. How do we I'm on a new chat. How do we just change? I was going to basically animate the fire. Ah, let's do that. Start. Okay. Um, no, because we want to choose the the specific version of Omni, don't we? Yeah. Oh, here we go. Here we go. We just select that and we go video omni. Okay. So, it's this one. Okay. Animate the fire and make me scream. So, this should be uh this should be like fairly fast, but I don't like I don't know if you used omni before. Oh no, I've got to uh link right. Close your eyes everyone just in case it reveals an API key. Please provide the dialogue appropriate time for you. This is why this is why you should test your app. Huh? Let's have a little look. Animate the fire and make me scream. It says token count failed. Okay. Well, maybe not be able to do this. Oh no. Yeah. Okay. Failure. But anyway, whatever. Boring. Long story short, Google is still cooking. And I think these models are more to do with like again apps, things that need like quick like you're if you're developing an app that needs to edit images at scale and you know you need cost is a factor, speed is a factor. These are new ones there. So, yeah, it's quite we've come a long way. Like we I mean we we played around with Nano Banana when it first came out and it was like, "Oh my god, this is amazing." But I don't do a lot of image editing or anything like that anymore to be honest. Do you? Um well, we had this project and for that we did a lot of image generation and video generation, but I don't think the light models would be helpful. Um, no, no, because that's like a client project, right? You want maximum maximum power. Yeah. So, I don't know. Should we leave it there? Yeah. Anything more to say about anything? Are you concerned? Are you happy? Are you sad? Uh, no, not really. None. Well, I'm I'm thinking like my fable will be gone and yeah, I have only 12%. But um I think I'm pretty good with Opus. Um it's just design that Fable was really good at. Uh but we'll see how it will be. I'm I'm going to experiment with these opportunities like these business opportunities. So we'll we'll see how that develops. Yeah, it's definitely it was definitely something that's been done before like building something first and approaching companies say, "Hey, look, I've built something. Do you want to buy off of me?" It's definitely and now it's easier than ever, you know? Yeah. I'm thinking like, what if I could do this at scale? I could definitely do this at scale, but because it's not really really cheap. I mean, technically you could have like three different $200 subscriptions that would be like $600. It's not too crazy to think if you could like run proper crown jobs like you could probably do in a week you could you could do several hundreds if not into the thousands of landing pages and then just automatically send them. But before me trying to do anything like that, I want to know that I can get reliable good data on like finding businesses like at that middle point where I know their website is really bad, but I know also they are making I don't know a few grands at least a month that they can afford like a two grand website and maybe they are not too big of a company to uh to burn my chance of like selling them, I don't know, a 50k website by just pitching them something too too small. So maybe it's about finding that uh that data source. I don't know. Uh I'll I'll try to figure that out and report back. Yeah. Yeah. Cool. All right, guys. That's all we got time for. Like, subscribe, do all that stuff. Yo, Josh is here. Yo, Josh. Hey. But today we today we figure that uh X is like people on X actually are not bots. We are we are getting comments from from Twitter. So we are indeed when we just finishing up. Um Andrew was talking about going on Instagram and I was thinking about going on Instagram as well. Yeah. Mix it up. Maybe we'll upgrade. Maybe we upgrade. Yeah. Well, I've I've got to go and meet a friend now, so I've got to shoot. So, sorry. I've got to go. I mean, you can stay on while you can chat to Josh. No, I have to go as well. But yes, um Josh will we would love to have you. Josh is doing some really cool stuff with like the industry leaders when it comes to design. Very cool. So, yeah. Cool. All right. Yeah. So, catch us next week. Same time, same place. You know what's up. A week in AO and Dev. You've been Sam. I've been Sam. I have been cover my