#57 GPT-5.6 Sol, SpaceXAI & Grok 4.5, Figma Acquires Bud, Cloudflare Monetization

Oh, that means I have to sit in the middle. That means Yeah. messes my composition then. This is really uncomfortable. Sit here. Sit here and I'll rotate that like that. Okay. Well, that's pretty much the same. Okay, cool. Uh, go live. We're live. What is add link? It It seems like I can add a URL. Oh. Uh, okay. The URL would be you like on Instagram apparently I can add an URL URL is uh YouTube uh YouTube and I will just link our channel. Okay. Hello Instagram. I already see people joining. Thank you so much for joining. So share, copy, paste. So, I'm just going to add the YouTube URL here and continue and confirm.

Cool. Now, Instagram allows me to add that. That's a bit strange. Maybe there are like too many people on like, you know, on our servers. Just move them somewhere. Anyway, let me put on my headphones so the voice won't be Say something. I can be your hero, baby. Yeah. Yeah. Sounds sounds all right. I can kiss away the pain. Yep. Yeah. You can sing. Of course. Of course. Right. We're live. We're alive. Um and we're one. This is There's something crazy going on. Yeah, there's some there's some streaming stuff going on. It's cutting out. So, I see some lags and it seems like Yeah. Hello. Hello. See? Yeah. I'm closing down a bunch of stuff.

Let us know in the chat what what you can hear and see because there's some Yeah. Crazy crazy lag. Um I'm just closing everything down. I can hear myself double and I can't mute the site even though it says muted. Do I not need to Is this not the Sorry if this is creating a lot of noise but all right so everybody we are just starting check check one figuring things check check out is that is that too loud is that too quiet what is Yeah, it bothers me a lot. I I can hear myself double double. No, you you should be fine. Yeah. I hope I can't hear myself again. Yeah, I'm just going to listen to it on live.

Well, I can hear you fine. I don't know if it's good everywhere. Maybe a little bit louder. Yeah. Nice. Okay. So, today Oh, I I should I should have uh written the script for the start, but we can just wing it as we often do. So, well, we kind of know what's up this week. So, this week it's a special week. It's one year of doing streaming together. Um, that that's that's crazy. Anyway, so this week we had the new chat, we had the new GPT models, especially the GPT 5.6 Soul. Uh we have Claude Flare giving us an update on how we can potentially make money from publishing uh on the web. XAI has officially officially rebranded what was officially rebranded to SpaceX.

Yes, it's it's giving me literally a stroke like it's not SpaceX to SpaceX AAI, it's XAI that is now called Space. We we'll talk about that. Figma acquires but a a vibe coding tool which I will show you. Uh Anthropic is doing some redesigns around Claude co-work and how Claude co-works on the phone. In our case, the lack of it doesn't really work the way we want it. Anyway, uh Meta has entered the uh the game again with a new model. Meta Muse. No, what is it called? It's Metam Muse, right? Muse Spark. Yeah, Muse Spark. Yes. And we have also Grock 54.5 uh TypeScript 7 and we have a a bit of a a discussion around if design is dead.

Uh because a pro designigner just quit their job after fable taking over. Uh, all of this and a bit more in today's episode. This is Command AI. I'm Kabarzel. And I'm Sam. And this is a week in design and dev. I think that's changed since we were young and subtle over a year ago. Just Yeah, just over a year ago. Yeah. I think it's design and web because we were kind of focused on the web side of stuff, right? But it's changed design and dev, of course. You've nailed it. I just I'm just an idiot. But anyway, design and what makes sense to you, design and dev or design and web? Uh design and web.

I I feel like dev is becoming I don't know the dev is in a way everywhere but at the same time everybody also can do it. So that's a that's an interesting topic in itself to talk about AI design and web. Well, we do I think web is better just because we do talk about web AI design. Yeah, we we Yeah. And and web encapsulates dev. Yes. And I'm a bit like sometimes I talk about it when I talk to people who are just starting something and I say, "Hey, with AI you can build an app." And they're like, "What do you mean like a an app like on my phone?" And I'm like, "Well, yes, but web apps is what I mean."

And a lot of people don't see web apps as apps, which is a bit unfortunate because web is the biggest platform if you want to see it that way because if you just build for Android, you are targeting, you know, a portion of people and then iOS the same thing. But with web, you are literally targeting everybody with an internet connection because they will use a browser to use whatever you have. So you're building a web app. Yeah, you have the biggest platform. Actually, a lot of people miss that point. Anyway, I think I think I think people normies will think, well, is it on the app store? Like as long as it's on the app store, it's an app.

Anything else is like not they don't see Facebook as an app. Even though Yeah. or like okay I don't know a dashboard of some description I don't know but yeah yeah either way a lot of these apps are web apps first and then they become native apps that's where you start and that's where you typically not always anyway uh should we get on with the news because this week a lot going on and we want to also uh yeah be on time. Yeah. So, what's up with the Cloudflare? Well, hopefully they're going to start your you wonderful people building your websites are finally going to get uh reciprocated in financially for your information and content that agents have stolen from you apparently.

And this is something we reported on a long long time. Maybe like maybe about a year ago. It was probably one of the first things we reported on uh this idea that uh Cloudflare want to gain back control of the data that agents have like taken from you. So this is the article announcing the monetization gateway charge for any resource behind Cloudflare via 402. And this is very very deceptive because they haven't really announced anything. They've already announced it, but they've just made some progress with it. This is very slow. Sorry, I was just showing you that. And was it not sharing the screen? Uh, I think you just didn't on the add. Bloody well did.

It's whatever. Well did. Um, it's working now. Yeah. Cool. Cool. So like uh this will give Cloudflare C Cloudflare customers the ability to charge for any asset protected by Cloudflare web pages data set API cutting off. So um is it for you as well guys? Let us knowing up for you because it does sound like that. Uh yeah, you're pretty much gone. melted for. Okay, we're just gonna wait for him to come back. I'm here. I can hear you and everything. You're I hear that that you can hear me, but then you're off again. Um, just testing my network speed. Don't know. Don't know. It seems like Can you hear me? Okay. now. So, uh yeah, I guess you're back.

Okay. Yeah, my my network speed was okay. Um sorry, you were back for a second, but then Yeah, because you cut off again. Let us know. Chat's going to have to let us know which side is it on. Yeah, because then the other person will have to take over. Um, I think you're back now. It seems like more stable right now. Okay. Um well either way Cloudflare will be giving you the ability to charge for your content and it's basically through a payment gateway that will settle in stable coins over X42 which is their whole Coinbase partnership uh crypto thing. So you'll get paid with crypto. Yeah. And I don't know. So um and they make a point here.

An agent does not need to look at ads to maintain a monthly subscript or need to maintain a monthly subscription. All the tools it wants to access. Um it reads pages or consumes a data feed once, takes what it needs and moves on across the web. AI crawlers already request content anywhere from a hundred to ten of thousands of times for every visitor they send back. So basically, you'd get you'd charge a few cents per web search build per call, 0.001 cent base fee plus 1 cent per megabyte for an upload endpoint, 99 cents per resolved support escalation. I guess that's paid only when work succeeds. Now it's um this is still a pie in the sky idea.

They don't give any dates. They don't give any indication as to like so let's say you're a crawler bot. Do you like this wallet thing here, right? Do you then have to just pay into a wallet uh through Cloudfare and then it just takes from that and otherwise your bot will be rejected from the from the thing. I guess um it's sort of a bit bit it's not quite 100% clear about how they're going to implement it nor have they given like I say any sort of time frames associated with it but they're demoing these kind of like ex like uh what do you call it like dashboards and things like that how it might work you can see that um if you get or post to generate image it will cost you $2 um if you get the news.

It will cost you 0.001. I mean, it might be quite cool. Certainly from our from our website perspective. We've got all of our uh you know, streams and clips and things like that on there with all the transcript. There's a lot of healthy information on our and that's the point, right? We wanted our we wanted our content to be searchable and indexable and all the rest of it. Why not frigin charge? It would be nice to say, well, Google can do it for free, but you know, um, agents can't, but then you get you run the risk of mirroring them and, you know, so yeah, that's kind of basically it. Um, it's a nice idea.

I don't know how effective it's going to be. Yeah. What What are your thoughts? I I like the idea, but I don't think it's going to happen. For something like this, it has to be industrywide for it to work. How will that work if I decide to not charge if agents are going to visit my website versus you know you charging? So in a way like either all providers is like the majority of providers like Cloudflare and AWS and everybody else needs to implement that uh to enforce it and also like people maybe need to have it turned on by default in order to enforce it otherwise it won't work. That's one side of the thing.

the other side. I don't want to sound pessimistic, but I can't ignore the fact that Zuckerberg and everybody else, they literally stole books and they didn't paid for them even though they could. It would have been not really expensive for them from what I understand. it wouldn't cost them as they have ton of money, but they just decided to even not pay for books that authors spent probably years working on and they just straight ahead they went on uh and stole them. So does that mean that all of the sudden the industry has a change of heart and they will start paying for articles that we write that are also almost certainly uh were written by AI completely or AI assisted.

I don't know if how how would that work? How will Claude Flare make sure the is that it literally like it it's gated, right? It's not I could go down the bookshop right now and steal a book, right? I can just put it in my bag and like walk out of a shop, right? There are ways to get books and things like that for free. Whereas this is saying the one and only way to get this article is through Cloudflare and without like it's just a blank box without it like as much So, how but how how can Claude flare make sure that somebody with computer use an agent that uses the computer use can't just steal it?

Can I Yeah. Can I like go scrape it in a not very efficient way but like scrape it with computer? Like what if I am anthropic and make a million uh instances of you know I just spin up a million uh computers and then each will have an claw. I know it's costly but it they can they can do it probably if it's worth it they will probably just do it. I have a feeling this isn't just for agents though. I think this is for everyone. So if anyone wants to come to our website, they got to pay the 0.001 cent. Interesting. That's interesting. That's like for the individual, they can crawl the internet for a dollar.

Whereas an agent who is constantly pounding a website, you know, tens of thousands of times, it it scales. Do you know what I mean? I know. I I kind of get the idea. So we are literally paying for well we used to pay for the megabytes that we download right nowadays pretty much every provider is unlimited but it kind of like uh in my mind it compares to the internet itself and this is maybe a new way that um Cloudflare is thinking about how how can we uh keep this go this thing going people doing research or like working on tutorials and like like very specific knowledge and still encourage them to share it. Uh previously they would share it and get paid by ads directly or indirectly.

Now that in incentive is gone, how can we still make sure that people will you know produce content? Yeah. Um interesting take. Um, I'm not sure if it's going to happen, especially that it has taken so long and not much has has been done. But if they can if if there's a company that can do this, it's Cloudflare. I mean, we don't believe in Amazon. Yeah. Yeah. Yeah. It doesn't mean to say that everyone's going to adopt it. It just might be that there's a few people who do utilize this this payment gateway thing, right? for those people. Yeah, I I I guess in I I compared it to the internet itself because like you are paying a provider to give you internet.

You are not expecting to put on a SIM card on your phone uh or like connect your modem to the to the wall and just have free internet. This is not what you expect. But somehow we started to expect we Google something and we get information for free. And that is like super valuable information that we get it for free. It's I don't know. It's kind of like a cultural thing as much as it is a technical thing. That's what I'm Okay, maybe this is what I'm trying to say. This is more cultural than technical. Yeah, maybe we'll figure it out. But it's definitely interesting. And uh I do want to see the the web to continue working the same way that it has with people producing content and putting it out there because this is how we are learning stuff.

Uh especially when it comes to specific stuff. Yeah. Yeah. But at the same time, it's just people have taken advantage of that. Like these AI companies, like, you know, Mark Zuckerberg, they've they've taken that honesty and and moral kind of, you know what I'm trying to say, right? They've selling you data. You you can't use the the word moral and Mark Zuckerberg in the same sentence. It's just just not how it works. Not my rules. Yeah. No, it's true. It's true. Yeah. There you go. Yeah. Let's see. Watch this space. Anyway, in other news anyway, how we don't know how we're going to tackle this one. Do we still know how we're going to tackle it?

Uh, let me actually pull up um artificial articial analysis.ai. Um, you can you can figure out um something to start with. Yeah. Any unless you've been living under a rock, GBT 5.6 Soul was released today or yesterday. It was teased for a bit and it is becoming mighty close to the great Fable for a third of the price of Fable. Um, so yeah, we want to take a little look at the sort of benchmark, show you that sort of stuff. But there's been a few, you know, shakeups around the whole chatbt ecosystem as well, which we want to get into, too. So, that being said, yeah, it seems like it's super close to Fable 5.

Uh it's a bit better than Grock 4.5 which we will talk about in a separate video. Uh you can uh stay with us uh to learn more about that. So, it's basically the second best. And it's a bit strange for OpenAI to make a big announcement like we have something amazing and it's still second best. Uh especially, you know, the underdog taking over so good that they can't even like announce it after and be better. Either way, this is one of those um weeks where you say this week we have a new model and it's this week's best model. That's not true. Not this week. Fable is still the king. Um but yes, GPT 5.6.

Yeah. Yes. Yes. Exactly. But if you care about cost and cost per intelligence especially, uh it's a very solid model. I haven't used it myself, but from the reviews that we've watched, everybody's like it's really good. I guess there is not too much to say about it. I think the general consensus is what I'm hearing is that Babel is without a doubt the most intelligent model, right? it can do or it can figure out some very very complicated uh tasks. However, uh GPT 5.6 Soul and we'll get into that that this is more of a workhorse. This is more something that will just it's the hard worker that will will get it done with brute force almost which is arguably you know a better or worse way to get things done.

So either way very very good model um can run for days. People have been running things for days on this by setting a goal and yeah I think it's gone I think gone of the days of like you know the these are these are new generations of models. These are new we've entered a new era of AI models for sure. But interesting interesting. So let's talk about that because I haven't used GPT 5.6, but from what I watched uh on Twitter, I've been like watching on YouTube, everybody's like, not everybody, but like from what I understand, GPD 5.6 Six. Soul is still the same generation as the previous one, but Fable feels like it's the next gen.

That That's how it is to me. Comparing Fable to um the other models except the GPD 5.6. Comparing Fable to the other models, it feels like it's it's a senior senior designer talking to almost I don't want to say a junior, but you know, it's next level. It's that good. Um, and yeah, Soul is not that good. You're literally just I mean, it's still it's well, yeah, it's not as good as Fable. It's not as intelligent as Fable, but you you've described exactly what it is. Basically, like the the fact that we've got a major version jump between uh Opus 4.8. I mean, to be fair, Fable didn't exist before. I guess it was Mythos, but like um it's 4 like Opus 4.8 is the same basically you're talking about the training that happened to create Opus 4 or whatever.

Four. It's Clawude 4, let's call it, right? Yeah. And we're still on those those models are still on the minor versions of tweaking that training set data of the fourth model. Fable is a new set of training data, the fifth model. So, oh yes, of course, we can't forget bloody sonet 5 as well. That's on that's on a new architecture, let's call it. Yeah. Whereas GPT 5.6, that's that's also that's on the fifth generation of GPT. It's still getting squeezing every bit of juice you can out of the fifth generation of GPT. So yeah, you are quite right and that's why Fable does come out on top when it talks about intelligence, but it doesn't mean to say that 5.6 isn't impressive and is I mean for a fraction of the price, you know, it's uh it's a good it's good.

What you got up on the screen there? Um this is like comparing cost to intelligence. Uh and it seems like zoom in. Yeah. So fable is the most expensive. Uh and GPT 5.6 has one uh it's like scoring at 59. Um Claude Fable 5 is at 60 but costs way more. It's more than double. Yeah. Almost triple. But that that's the thing like if the task is worth it to you, uh is it worth going with the the top best model. I think it's quickly becoming in a I hate to say this, but in a way a luxury thing where if you have the the money, you can get the better engineer. Uh, and if you don't have the money for it, you get the one that is maybe not as good, but still can get uh, done.

Uh either way, we're we're um moving in the direction of everything getting better, but not as fast as I personally would love to see it because I would love to see um GPT beating Fable 5 by a big margin because that would mean that access to better models. Like at the end of the day, I don't care who I pay the money to, if it's this one or the other company. All I want is like better and better models uh as a consumer, right? So, it's a bit of let down that it's not as good. Yes, it's cheaper. It's pretty good, but I I don't want it to be comparably good. I want it to be, you know, the difference between Fable how Fable 5 blew us all blew all the other models out of the water.

I wanted Soul to also do the same with um Fable 5. I I really want to see this because even though I I feel like everybody got a favorite and you root for, for example, GPT to not be better than your favorite Fable Five, which is stupid. Like just I'll just say it. It's stupid. like it's in your best interest for the Chinese company tomorrow to come out with something that is better than Fable 5 because competition is good for us consumers um at the end of the day paying uh to these token dealers. Um saying that if I add my thing to the stage uh come on. Oh, I need to remove you. You need Yeah, you need to take mine now.

Yeah. Just to break down um this this soul nonsense if you haven't heard of anything. If you haven't heard this yet, right? So, they've released three new models. They've kind of copied anthropic here. Yeah. It used to be GPT 5.5 and then you just choose the thinking level. Well, actually, you've got GPT 5.5. You had nano, mini, all that kind of stuff. I think they've removed that and now it's Terra, Luna, Terror, and Soul. So, Moon, Earth, and Sun. The Moon is the smallest one. Luna is the smallest model. That's your haiku. That's for haiku. Terror is your middle weight. That's your sonet. That's the the mediumsiz model. And then soul is the big boy, right?

So, that's how they've broken it up. And the reason why I say about the uh Chinese model is if I search maybe it's here in this one they actually do compare Luna which is the smallest one um matches or exceeds the intelligence of GLM 5.2. fancy that a a frontier lab like boasting or no sorry they're not boasting because this is artificial intelligence's thing imagine being a Chinese lab being compared to I know it's the smallest model I know that but even still that's pretty impressive isn't it really that GM 5.2 two is up there. So, I am I am pretty stoked on that model. Um, it's just about working it getting it all in your workflow and making it into something like when do you send when do you send things off to GLM 5.2?

Do you know? I I'm I'm not uh on that level quite yet, but I can imagine setups where you you have something very expensive like Fable 5 that is really good at giving prompts to other sub aents and then sending those sub uh creating those sub other sub aents not with anthropic models but let's say with um GLM that can you can you do that actually is something. So, someone actually commented one of my videos about it, right? That you'd have to create the plan with you you would use um open code probably. So, you couldn't use your Claude Max subscription to do it nicely, but you can set up sub agents to use different models and whatever.

If you're inside of Clawude code, what you would need to do is get the get Fable to write up a detailed implementation plan in a markdown or HTML format and then you would get, you know, you'd get GP uh GLM 5.2 to to work. I see. If you are using APIs, you could build your own tool theoretically where you you create this harness yourself where um on the top you have the the best model and then you allow it to create sub agents uh and you have like strict rules for them. That that's actually really interesting. Um that's an interesting approach. Maybe somebody with uh much more money can can do this or people are probably already doing this.

Yeah, I mean like someone asked for a video on this and it might be something that I put together, but um yeah, it's totally something you can do. I would probably go down the cheap ass route and be like coding plan just have clawed code open in one window and open code open with the other one with GLM and just have them on to the same directory. So, I'd just do it like that sort of thing. But, uh, yeah, you could totally build your own harness to to be able to again, you can define sub agents in open code and then tell it what model that sub aent should use. So, yeah. Yeah, that that that's very by the way, I never Yeah, once upon a time you could have clawed code inside of open code.

This is before all of this nonsense with the programmatic usage came into play. You can literally run claude code max subscription alongside inside of um open code. However, open code do have their own coding plan. So that could also be an option. Yeah, I never cared about the token usage uh until this last like week where I hit my weekly limit both for Fable and just the general weekly limit. Uh, I hit the 100%. Like I I I had no claw for like five hours. Oh my god. I was just working like a peasant. Like I was working with my own fingers. Like I was I was clicking things and I was like typing. Yeah.

Um, it was horrible. It but I I want to make a video on this. Um, I mentioned it a little bit earlier and I know this is maybe a different topic, but the idea of scaling maybe I was thinking like can I create 500 websites in an hour challenge? It's like a business model. It's an idea. A business model. Can I like if I list 500 websites that are shitty and I have um like a proper uh model, when I say model, I mean like a plan, not like an AI model, like a proper plan and ready components. Can I create 500 websites in one hour? This is where you care about tokens. And why would you build 500?

I'll explain it in my video, but the idea is just a business plan. Maybe I can send 500 offers, like 500 prototype websites that I can create for people that might need it and I just generate the website, the plan, and the offer and send it away. So, this is where you would care about tokens. This this is what I why I'm mentioning this. And with all that being said, I would like us as we said in the earlier in the call to have our own uh benchmark. So if you guys have an idea for if you what we have ideas it's not that we don't have ideas for benchmarks but if there is something that you want to see to make it more tangible for you um let us know because we we will try to make a benchmark where it becomes easy for you to see here's how fable 5 comparing to whatever typ 6 when it comes out that would be interesting to see So I for me it's design.

I want I want to compare the design work. Cool. Yeah. Yeah. I have nothing bad. A lot of chatting. Yeah. Uh let's go to the next one. Uh XAI rebrands SpaceX AI with a new logo. Uh let's yeah tackle this. Before I tackle this uh from our comments on Instagram, uh what are you paying for? What what is your subscription uh currently looking like? Uh Claude code $200 only because of Fable. I'll downgrade it to $100. I will do the same this Sunday. Saying that, actually saying that, I don't know. Uh, I discovered I have a Kimmy plan today, but I won't be continuing that. Um, I have a yearly GLM plan, uh, which I want to upgrade to the middleweight, the pro version.

I upgra, uh, the month of the cheap one just to test it out and then upgraded. But yeah, but that's mostly because I've got a ton of referrals. But I do want to incorporate GM 5.2 a lot more into my workflow. So I do have that. Uh okay. But yeah. Yeah. Cloud code mostly cloud code. Okay. The answer for me is also I'd be tempted to get the $20 a month GPT plan. To be honest, I I still have the GBT GP. I have the GPT plan, the 20 bucks, and I have the 200 bucks from Claude. And I have the Hicksfield subscription as well. I'm thinking what other AI tools I'm paying for. Technically, I'm paying for Figma for four, five seats.

Uh, and they have AI credits with them as well. And we do use those AI credits as well. Um, for remove background and things like so this is like my AI subscription as of now. Uh, okay. Let's talk about this one is going to be uh to be short but maybe give me a second. I uh please I have I have like um I have a tweet that I want to show. No. Uh the the tweet was listed but the bot didn't add it. So, I'll just go grab it from the list that I have it. Okay. Um, crazy news has just dropped when we get to uh Facebook Muse. Interesting. Yeah. Yeah. I would love to see that.

Okay. Crazy. Let me Well, apparently I've just I've just got news here. Anthropic wants you to pay up Claude Fable 5. Claude subscribers must soon pay usage based fees to access uh Anthropic's best consumer AI model. To be honest, I think we sort of saw this coming. However, that as we reported last week, they are looking to bring it to subscriptions, but in the meantime, I think they're going to from the 12th, they're going to say extra usage. The only reason they do this and they can get away with it is because Open AI is not beating them at quality. Yeah. Yeah. That's the article from Wired. Yeah. If if OpenAI was better, um Claude would have to, you know, provide their best model, continue it.

Um, as we were talking about GBC, why don't we move on to Codeex rebranding? That feels like a natural segue into that. Codeex rebranding. Yeah. Yeah. And then we'll talk about the SpaceX AI rebranding. Yeah. Feels like a much better segue. Um, so in amongst all of the rebranding that's happening with everyone and the and the naming conventions, if you open up Codeex, Codeex has completely rebranded to ChatCBT now. So it's now Codeex doesn't exist anymore. Um, and you have to tell us what you think about it because if I share my screen, I also have a screenshot of it. I took a screenshot before. Yeah. Uh before closing it. Use it though. I want to use it.

Yeah. Yeah. But like I want to show this the screen. You see? Yeah. So when you update your codeex app, it tells you codeex is now the chat GPT app which is super confusing. It says chat GPT codeex and then you will see also chat GPT work which all exists also on a separate app. Now you can show us how that looks. So this is a new instance. This is described and there's a few little bits to take note. It says chat up here. It says chat up in the top left side of my screen. Um but it's still a codec. It still says codeex down in my dock. So, you've got this and you can if we if we chat chat now exists as a sidebar like a like a Yeah, like a popup.

I guess I suppose we could do this and make it into its own little window. Oh, you can't even see it anymore, can you? Um, but yeah, it's a little popup. And come on now. And we can do this is a new feature which is great. And you can guess what it is. is called co-work but Chachi PT's version. So you can do tasks and things on your on your computer, get things done. For the normies, they can build apps um build more advanced apps than you can just bu in the through the chat experience. But then to get to codeex, this is how you get through to that. You go to chattex and this is where you can code.

So, it's slightly, you know, I was saying before the show that like I think codeex should be a separate app. I don't think anyone coming in here wanting to use chatbt, downloading the chatbt app and then getting introduced to work, this is all they need. as as questionable as this little popup chat is and then the primary focus being um chatbt work. I don't think many of those users are going to go into codeex. So I think that just belongs as a separate app otherwise it's just going to confuse them. I think any that's just my opinion. Um but yeah it's it's all of a bit of a mess but they have followed again copied claude code here anthropic in that that's a whole unified experience.

I think code should be completely separate from these people. I I think they just they should not be concerned with the coding aspects of these things. But yeah so if you've got chatbt app installed and you've got codeex app installed well congratulations you've got two chatbt apps installed. Yeah, this is how it looks like by the way. Like there are it it's confusing and on the chat GPT app you can still do like image gen. I haven't tried that with the old codec now also called chat GPT app. I haven't used it. Um another thing that they copied over from claude is the the model thing. Now, if you go from extra high to ultra, it does the purple thing that exactly clot does.

It's just it's just like copying their homework. But I get it. Like if something is good, copy it. Whatever. Uh but it's really confusing because they are both chippy and it seems like maybe they want to get rid of one of them and I hope they will. Uh maybe they should get rid of the old Chachi PT app and just go with the with this new version. Um I I quite like it. Uh I don't mind that the chat being like a popup here, but again it's also very This is not very friendly to people who just want to install an app on their Mac and start chatting with it. it it it's not as friendly as clawed uh desktop app where we have like our criticism towards that as well because it was busy and you know we we will talk about how they are cleaning that up but in the process of uh openai cleaning their stuff up they are creating more mess but maybe maybe they will maybe this is a process where they will merge these into one app which makes more sense to have chat coding.

I that that's what I would do. That's what I would like to see to have all of it in one place. Um do do something for me. Yeah. Click on your name in the bottom left. Which on which app? Yep. That that's it. Yeah. And then go show pet down. Oh, hello. This is also another copy of Claude. Yeah. Why are they just cop? Yeah. Okay, I guess I I can grab it and put it somewhere. Yeah, cute. I think it just gives you I think it just lets you know when something's done or you know uh some really Yeah, I think so. I don't know. Maybe maybe just ask a question or something. Uh by the way, do you have access to the new models?

I have terror. I don't have soul. Or maybe let's have a look. No, I don't have soul. I have sold I have sold Terra and Luna. Yeah. Yeah. I have the 20 bucks. Oh, there you go. I will um I will do something with it. I will do some I I will do the design test that I always do uh and see how it does with Figma MCP and all of that. But I guess I have to install install all of that. Yeah. Yeah. So, yeah. Anyway, very confusing. Very confusing. Let's move on to Claude co-work for web because I think that'll be a quick one. Uh okay. So that is we we can also talk about this literally here.

So that's why I thought speaking of redesigns now we have let me make it full screen so it's easier to see. Now we have another redesign. Co-work used to be here in the middle. Now it's just home versus code. In code, you can obviously write code and in home you have co-work versus chat. So the tab is now here. And if you use co-work can use your computer, use your files and you know create a website for you and do research and things like that. What's really cool it's it's it now exists also in the web interface as well as on the mobile app interface but with our testing the mobile interface could not access my exact desktop.

So every time I'm setting up a chat here, I can use it as as as uh as usual with co-work. But if I start a co-work on my phone, it creates its own computer environment and it cannot access my computer here. If I want to do that as of today, then I have to go into dispatch which is this is now the the mess that Claude has made. So each one is creating a because these are just companies like going like full speed throwing at the wall and seeing what what sticks and changing things all the time. Yeah. In my ideal world, co-work and dispatch would be one tool uh from which I I can I can choose if I want to create the instance uh on my local computer or I can do it on a on the cl in the cloud in the cloud with a Linux whatever computer uh and then the same uh options should exist on the mobile phone.

Maybe this is the direction they are going both if uh if I'm trying to read what Anthropic and Open AI are doing. Um it seems like they are trying to unify this experience and it seems like the industry is shaping because previously you you had no idea what they are trying to do. Now it seems like they know, okay, people want to chat. They want to spin up agents to do research and they want to get done on their computer. So, let's figure out a UI around that. I wonder what the percentage split is because you are I think you are right like the chat was obviously groundbreaking and a lot of people still use the chat.

I can see your history there, right? And um it's it's it's wild to think that uh like people still obviously use it, but now we've moved on to getting stuff done, which is what co-workers about. So it really raises the question about the new chatbt work being the primary focus of the app and the chat being secondary. Whereas I feel like only subtle, but I feel like Claude's co-work is secondary to the chat, which is kind of counterintuitive from a Claude standpoint. But I do I yeah, I just wonder what the actual split is. Whether whether people are moving away from chat and going into co-work or whether co-work or work, whatever you want to call it, is still an ambiguous topic for people to grasp.

Do you know what I mean? Yeah. for the normies people still don't know what it means uh when when you say claude can do something on your computer um and I assume for a lot of people it is because they don't work with files a lot of norm they probably like people who work in companies they they have specific software either installed directly in their on their machine or like really old uh enterprise level web apps uh that are very specific things. So, it's kind of like hard for them to grasp. Okay. Can co-work do this, which brings the you know the the whole thing with computer use and how how they can actually operate an app on your computer, but they are not really amazing at it.

Yeah, I know why you are laughing. I just looked at it's 6 p.m. Yep. Sounds about right. Like clockwork. Anyway, uh speaking of should we talk about the XAI thing as well? Exactly. Yeah. Yeah. Yeah. Okay. So, speaking of more redesigns, we have probably the most um comp complex. No, what is the word like complicated? No, not complicated. It's like confusing. Speaking of redesigns, we have here the most confusing one, SpaceX AI. So space X is still called SpaceX but XAI that was acquired by SpaceX is now called SpaceX AI. Kind of makes sense but at the same time it is uh a bit confusing. So we have also this funny why we have this funny tweet that says speaks SpaceX AI is not SpaceX.

SpaceX AAI is a rebrand of XAI which owns X X the app not SpaceX. X is subsidiary of XAI not SpaceX AAI and SpaceX AI is a subsidiary of SpaceX. So the main company is called Space X and now Space X yeah has SpaceX AI which makes sense. Who's a daddy? But the rebranding like going from XAI to SpaceX AAI, I don't know. It it sounded a bit um confusing. But I have also something to say about the the logo. So this is the old logo XAI. Oh my god, look how rubbish that is. This is the old logo. Uh and the new logo is just that changed and put on, you know, it matches with the Space X logo.

So there is an AI under it. Uh and this shader. I really like this shader. Look at this. I I like this. I like this. Yeah, I really like this. Uh but the logo, some people have been saying that this looks a bit familiar. Yeah. There's only so many ways you can skin a cat, but you know. Yeah. Why would you skin a cat? I I don't know that uh saying. I don't know that saying. I'm learning a lot of sayings from you. Baby Jesus Christ. Baby Jesus. I mean, that's that's not a British thing. I think that's a me thing, but you know, I don't know. Yeah, I'm I'm Yeah, I'm learning things from you.

Either way, uh, another rebrand. I guess now that I, uh, I read it, it does, uh, kind of like make sense. Elon Musk, I don't understand why he loves understand. Yeah, he loves it. He loves it. I still call it Twitter, though. Well, the the the whole idea is that SpaceX is the main company and now SpaceX or has a subdivision that is SpaceX AI? So, does SpaceX own XAI and X or does SpaceX own XAI and XAI own X? Yeah, interesting. I don't know how they are exactly structuring this. Either way, it owns the whole thing. Very uh yeah, very rich. Um anyway, yeah, on that note though, Grock, who is Oh god, we just opened up a whole new can of worms, isn't it?

What is Grock? Grock is XAI's model who worked with Cursa on the new Grock 4.5 model. And it's actually apparently not a bad model at all. Is this on your screen now? Oh, there we go. Yes. So, we have Grock 4.5 compared to the other best models and it's respectfully good and it it is a surprise. It is a surprise especially because they were so much behind. They were a joke, you know, they were a meme, but now it seems like they're catching up. Everybody is catching up and this begs the question, how are they going to win? But what's interesting is that they have a lot of um infrastructure and they own cursor and this might have have to do something with the data that they have from cursor uh which will make them even more and more powerful with the time.

But either way, we have a new player in town. not not really new, but they're back in town and they're not really, if I share my screen, they're not really going for intelligence here. They're they're kind of going for speed and cost, which kind of makes sense because you got composer, which is Curs's model. Um, but it is their smartest model obviously, and trained alongside Cursor. So this is like what we see with the benchmarks with the deep street one which is the closed source uh you know arguably more reliable one. You got Fable Max at 70 GBT 5.5 be interesting to know where 5.6 sits there. Uh but just below in between Opus 4.8 and GLM 5.2 again GLM you know mingling amongst the the the big boys there uh sits right there.

And again, it's a it's a pricing thing with regards to I mean, this was apparently one prompt, this little 3D thing, which is quite cool. Yeah, it's Yeah, it's it's pretty cool. Uh f like 4.2 4.2 times fewer tokens than Opus Max. Um 80 tokens per second, which I didn't think that was that great, but apparently so. I was getting 40 tokens per second on local AI and I thought that was I mean it was good but like but this is smart and doing 80. Yeah. So 2 million in and 6 million out. So pretty good. Um yeah I think it's I think it's now finally a bit of a contender. You can actually use it for free using their tool Grock Build uh which you can you know install with this here.

I've actually got it installed and it's quite a nice little quite a nice little interface to be honest. Uh I'm going to install it as well. Yeah. So you can see my terminal, right? So, I've got it installed. And if I And what what's funny is that they've done what they've done is that you can actually type in agent to start it, which I think um and nobody else uses agent. Exactly. Exactly. That's what I'm thinking. Obviously, you can type gro as well, but you can type agent. So, now I have access to this. So, um, what what were we going to do? Um, build a direct clone of Streamyard. Yeah. Feature and make a marketing website.

And make a marketing website for it. There we go. Feature par. Let's uh because I'm in another project right now, I'm gonna have to copy move my project into it. See what it can do. We'll have to check in at the end obviously. Yeah. But um here at my place it's still downloading. Yeah. Um I wanted to to say something about Grock being fast and cheap and being also like intelligent enough. It's it's kind of funny because if you if you want to do design work, you should use Fable. But I'm just saying if you have some agentic work to do uh and you want to use a model that does anything you ask it to and doesn't like ask too many questions and just does it and there is no you know too much sense of moral you probably want to go with this one.

Yeah I guess so. Yeah. Um I did try and generate a sexy dragon though uh because you can generate images and video from within Grock build. Mhm. It refused. So, sorry guys. Show you a sexy sexy dragon. Did I did um generate a bug on a mo motorbike though, but I can't find where it is. I've lost it unfortunately. There we go. It's going to build stage cast. This is So, we're using Streamyard right now. Um it's going pretty quick, I have to say. Well, they have all the infrastructure. This is quite good. And this is picking up on my like agents and my skill all my SCA agent MD files and skills and things like that.

So it's able to because it that's the thing. Claude is the anomaly here that has its own clawude folder with its own clawude MD and its own skills. It's so annoying because if everything just used agents.mmd and used the agents agents folder with all the skills in there, then you could just float in between all of these tools. But a Um, Anttopic really are the apple of of uh what do you even call it? Apple of AI. Yeah, this is this is pretty quick. Uh, by the way, are you using the free tier? Yeah, this is free. This is free. But it's free for seven. But you signed up for the $30 but with seven days or No.

However, I did have to sign in and I'm sure it's knocked me into that free trial tier because Yeah, it is trial because Yeah. Yeah. So, there is a because for me exactly it doesn't allow me to do anything unless I subscribe to the seven uh now I need to tell an agent to remind me in seven days to you know cancel to cancel it. Uh oh. subscription. I I have way too many sub I have a weekly call uh with uh my teammate where we go through a list of subscriptions and I try to cancel them and it's not really easy. They make it really difficult uh often. So, well, should we check in on this in a bit?

It's it's moving very fast. This is moving very very very fast. And suppose maybe I'm used to a bit of Fable, which is quite slow, but we're just three minutes in and we've already generated a bunch of files. Hey, what if this is like perfect? What if this is exactly what we need? Well, we've already paid for the year now, which is annoying, but you know, like 300 300 bucks for Streamyard. 350 bucks. Like ad revenue just about pays for this. Just about. Yeah. I I don't know like um I I I wouldn't create something like this with VIP coding because like Streamyard I love how every week where when we sign in it shows us a new feature that they have added.

I really like we kind of sometimes do but yeah I get it. I get it. I can if we can well o um even OBS is open source. You could theoretically just take that, clone it, and build, I don't know, Streamyard on top of that. Maybe. Maybe we just Yeah, I guess we just need the Well, no, because I need your feed. We need a place where our joint feeds uh you know, meet at the same point and obviously the local saving the local files. I by the way with the subscription you get access to Super Grock. Okay. So what's that then? How is that different? I I don't know. It's just the app. It's just Do you have the app as well?

I wonder how that differs because surely Yeah. Is there another version of Grock 4.5? I'm loving how fast this is. Right. Should we move on and then we'll get back to this? Uh, yes. Let me like give me one second. I'm not sure if I should. Yeah, just because I mentioned when you sign up, this is what it looks like. You have super grock uh and you have imagine. So, you can create photo and videos and you have the chat experience. But now you have also the build mode directly here which you no actually it's just it's just a tab that says you you can use it in the terminal which now I have it see I just installed it and now I can build me a billion dollar company make no mistakes.

Yeah. Oh I thought you I thought you used your voice. No, I will. No, I I should have used it. No, I I don't want to use up my free tier uh for that dumb idea. Who wants a billion dollar company? I want to create a clone of something. I'll just figure it out. Maybe a clone of my website. Um Babak is saying, "I'll do push-ups for every time you guys uh say SpaceX." Well, we're done now. We're done now. But our audience like you are I don't know. You are getting buffed because we're saying uh certain words way too much. Well, let's uh let's move on to Muse Spark 1.1. Luckerberg's latest attempt, latest stab at AI, which I guess means llama's no more and he's given up on the open source dream of local AI.

Uh, but some news did just come in whilst we were on the whilst we were on the the the we the live the live the stream. Yeah, I share my screen right now, but I will get to that. I'm not going to shoot you at it. So, today will you hear? Yeah. Today, we're releasing Muse Spark 1.1, a strong agentic encoding model at a very low price. Guess why that is, boys and girls? is available through our new meta model API and in meta AI. Muse spark 111 is strongest at agent performance tool used in computer use. It does well on long running tasks with 1 million context window. Uh can delegate execution to sub aents running in parallel and is trained to use computer interfaces on desktop, mobile, and browser.

So this is what they're comparing it to. Apparently better than um Opus 4.8 eight max better than GPT 5.5 on uh high on MCP Atlas on job bench manage last exam finance agent it's it's apparently well yeah it's pretty modest I mean it's losing to a couple of these things here but uh losing quite badly I would say on deep but anyway so they haven't given up no they haven't given up meta model API allows developers to build Muse Spark for the first time are focused in delivering strong agentic mo aentic multimodel uh models very low cost more to come soon. Jinx what I think they released this because they released this the same day as uh Grock 4.5 came out.

Oh man. Oh man. Billionaires. Yeah. I I don't Hang on. And there's an advert. And there's an advert for like developers on X. So that's I think that's all we've got. Here we go. Input 125 425 out. It's quite not bad. Not bad. Um I'm I haven't seen anything more than that. Oh, here we go. Mark Zuckerberg. Have you heard of Met? Um yeah. So, the news I got in fact was a tweet from um Oh, I hate X. It's just Here we go. So, from Marcus Marcus Brown Lee, can you read that? Yeah, I know the guy. Yeah. Instagram decided to roll out new Muse AI. So, this is an Instagram now. uh a feature that lets users create AI images based on people's Instagram photos.

They opted in everyone with a public account. So you're opted in. So anyone can just now create using your face on Instagram. Maybe I can just share my screen. I'm trying to see if that's available. No, not yet. Maybe not in the EU. Maybe not. This is the uh this is the announcement. Yeah. Opted in everyone with a public account. To opt out, click your profile picture. Three bars top right. Scrolling. Sharing and reuse. Toggle off for posts and reels. How crazy is that? And that is so Mark Zuckerberg and exactly why you're being offered Muse 1.1 for cheap because they're probably training off of your ass. So yeah, I think I think Mark Zuckerbo is screwed at this point.

Like he can't do anything. He's the most untrustworthy, slimy excuse for a human being that the world has ever seen. And there's stuff that we've not talked about that and that hasn't gained a lot of notoriety that he has done that's despicable. Like created fake companies that um spy on traffic. Uh there's a VPN service that they're offering for free that was spying on people's traffic and stuff like that. He's done so much that I feel like the the media hasn't reported on at all. He's still going. He's still just he's just a horrendous human being. So that's that's that. Uh yeah. Um also Instagram is now officially opening up uh the index its indexability.

So now your posts are uh readable to all agents and all from browsers and yeah we'll take who owns X. What who who owns X? Elon I thought you were going to say SpaceX and that was going to hit our SpaceX. Yeah. Technically it is owned by by the guy who owns SpaceX. Yeah. And now owned by SpaceX SpaceX AI, which is a subsidiary of SpaceX. Who knows? Who the bloody world knows? Anyway, uh not one for the designers, but certainly one for the devs. Typescript 7 has been announced, and it's got gained a 10x faster native port of Typescript. They basically rebuilt it in Go, which has been an ongoing project for many, many years.

Type 67 is available via MPM. And for those that Here we go. There's a little thing here. Um, for those that don't know, if you're building if you're building with AI agents, it is no doubt using TypeScript. It's the code that's building a lot of this stuff. for this is it's perfect for AI agents because it helps them understand how all of the files are linked together certain types you know whether if you pass something into the function whether it should be a number or a string the agent can know all that so TypeScript has really become the one of the most important languages uh on the planet for sure I ran some tests over my side and just to put it a little bit in perspective I've got a little um uh Uh Jimmy Cricket here.

If I stop screen. So I ran it on a project and this is this is what I was seeing. So the the normal TypeScript what I was running on this project our builds took 66 669 milliseconds which is about 6 seconds 6 and a half seconds. Um 7 seconds compile time using 208 megabytes of memory. Pull run build 1,949 milliseconds with with uh TypeScript 7 125 milliseconds from 669 0.092 seconds from 070 half the amount of memory consumed uh probably about a third less. So you're seeing a five times speed up there in the uh I guess the wall clock average. I don't know what that is. 7.6 six sec times on the compile time.

Uh two and a half uh two times less obviously uh memory cons consumption and one overall one and a half times speed. So this is a really small code base and uh the obviously these times are small tiny but when you've got a bigger project and you've got to build and do all this sort of these are noticeable. So, TypeScript 7 is a pretty big deal and amazing what you can achieve just from using a different Yeah, I don't understand Go or any or Rust or anything like that. I think it's the compiler of the language. I don't know what it was using before. Maybe it was using Rust and now they've moved it to Go, but it's the thing that compiles everything and how much speed you can get.

So, yeah, really big deal. Um, and I guess most people won't care. They'll just benefit from it because our agents will start downloading the latest version of of Typescript. Although, I do think it's um a release candidate. Oh, is it? We're on 0.02 now, so maybe not so much. So, yeah, Typescript 7 is in the house. Babel, uh, Pro Designer. Um, let me get that ready. Copy link. Um, maybe I want to find the Dylan. Dylan Field, the CEO. Okay, got it. All right. So, here we have Gal Sheer. Uh, you probably have seen him. Um, he's a designer with quite some following and he just posted something that went super super viral, like 5.2 million views.

It's over. I'm quitting design. A client of mine just created a logo with Fable 5 and the result left me speechless. Uh it understood the brand story, value uh values, audience, strategy and turned all of it into a smart minimal symbol. A genuinely brilliant concept. The kind of idea that uh captures everything at once. Something I honestly don't think I would have come up with myself. And it didn't just nail the idea. It executed the design pixel perfectly. So I raised the white flag. My skepticism about AI's ability to do great design is officially gone. There, I said it. AI beat me at design. Now that AI finally took my job, I can peacefully quit and dedicate my life to studying the only thing it may never achieve, human consciousness and the pathway to God.

Good luck everyone. So, this tweet got quite a bit of attention and some people called it like a joke or rage bait. Um, but apparently a lot of people took it seriously. Like a lot of people that I respect in the community, they it seems like they are actually taking this as a serious um as a serious tweet. Zero chance AI is coming close to your brand launch edits and sound design. zero. Yeah. Uh maybe motion designers have another year or two, but it's still a dead end. Uh so we'll talk about how you feel about this. There is another tweet I want to share with you uh which is from Dylan. This is the CEO founder of he's also the founder, right?

He's either way he's the CEO and founders of founder of Figma. So that's kind of like a big deal. He literally replied to that tweet. Uh, it's a long one. I'm not sure if I want to read the whole thing right here for you, but we will for sure link it. But the idea is that um he he thought that he this might be actually um a bait but decided to answer it. And like just a summary, I'll read the summary. In summary, I've never been more confident in the role of design and impact design can have. I wish designers felt the same confidence. This is the moment to be more bold, to take more creative risk, to double down on the power of design.

Everyone is on their own journey and there are lots of fascinating ways to move through life. Uh so if Gal is serious about quitting design, then I wish him the best. um in his adventure ahead. But I hope if others follow the follow follow they do it but but I hope if others follow they do it because there are other things they are so excited about spending time on uh versus fear of AI. You can also read the rest like the the start of the tweet, but it to me feels like he's talking about how design matters, but he's not completely capturing what um the the issue is, which is yes, design still matters, but what if who does the design is changing?

Uh what if you can you know instead of like actually um hiring a designer and you know spending the time that you would give feedback to them what if you would spend the same amount to give feedback to an AI you know in that case you're not losing time you might even gain time uh because you don't search for a designer you don't chitchat you just you know get the give them the work for them to get the work done. So I feel like Dylan is not seeing what the issue is completely. Yes, design is really important, but it feels like if an AI agent can do the job of my designer, I probably will give my designer the same AI to do the the work with.

But what if he fails at capturing that um transition? What if he fails at that transition? Uh what if designers like look what we did. We we we are working in the terminal day in day out. Uh what if people not that it's too complex but maybe they they don't like it. Maybe it seems too complex to them. Or maybe maybe they don't love the idea of giving an AI a task to do. Maybe they maybe you want yourself to do the thing. Why do you make uh bread yourself at home? Sometimes you know you can go out and buy bread. Maybe you enjoy the process. But that means you cannot make bread at home and make it financially viable.

So if my designer in my team if he fails at you know getting along with AI in a way maybe he doesn't like this new world then what will happen I guess they will be left like designers or developers in this case as well they will be left behind because a big part of this is about speed AI can well it's getting there can do bigger chunk of work faster. I'm not saying it's true for literally I I know I have like I'm saying this with a big grain take this with a big grain of salt but there are things that AI can do really fast and my whole saying here is what the whole um talk here is about what if people fail at adopting that um yeah uh I feel like there there are a lot of people who will be left out.

A good example is today I got uh a message from somebody who wants to do React development I I guess as freelance like why would I hire somebody and not just do it with AI? Um, I'm not saying this is the case for everyone, but it's it's becoming so difficult to hire people who are not AI enabled. That's my point. Mhm. I think that's the point. Yeah. And that's that's the norm. Like you need to be AI enabled. Um, but you also need to know the foundation. So you just need to come at you need to come to the table with with already a deep knowledge. you know, you need to come to the table with an already mid-level um knowledge of your and by the way, everyone knows you go to university and you you learn the theory, you learn the facts, you learn the things, but it's the stuff it's the real world everyday knowledge that you you don't come with.

And I think that's the point is that okay, do they do they know the theory? Do they know understand the principles and now they just need to apply that in the real world? Um, yeah. I don't know. This sounds I don't think this is a joke. I I think this is serious, but I also think that this person is potentially a bit um insecure. You know, they even say it themselves. Um I think anyone with anyone with passion and and whatever will see the opportunity as exactly that an opportunity right you can do so much more with design AI sorry but as I mentioned earlier I said I'm building a system around the idea of how can I create 500 websites in an hour but that's the kind of like the goal that I have.

I'm not saying this is like replacing design. It is a new way of thinking. It is a paradigm shift. It's it's shifting the whole thing. You know how there there was a time where this whole notion of Figma doing design and then doing build didn't exist. That that's something that happened in between. There was a time where you design was done in code and then we had this era we still have where we're doing design in an intermediate mode and then we iterate in code. I I think that is collapsing like a such a speed that is changing. If you are a designer and your whole world is doing UIUX manually in Figma, I'm really sorry, but like you you got to get going.

I hate to sound like this AI pro like you know scaring people. It's not about that. It the industry is changing like I see my clients I do AI. I do UI, UX and development for my clients. I see this whole industry is changing at such a speed and it's all about AI and the only thing you need to do is to get in start trying it's not as complex as it seems like and you have to update yourself um love it or hate it the groundwork nowadays is done with AI so you no more do the the the massive work of you know building things uh bit by bit. The big part of the work is done with AI and our job is changing uh from doing the thing to orchestrating the plan uh looking how the plan looking over the agents how they execute the plan and then review it.

I I feel like that's that's kind of like the is the general gist of how things are happening. But yeah, I don't know if people if people are ready for this. Um, that's not what I see. I see a lot of people still still doing things uh the the old way. Yeah. Yeah. It's a waste of most people's time and money if you're not, you know, and don't get me wrong, I don't want I don't necessarily want an AI design. I mean, if the design is good, the design is good. Don't necessarily want an AI design. We just need to whatever it takes to get to where we need. I think there's an expectation of, you know, certainly from a speed perspective.

You know, design is a bit of a weird one though. Design is a bit of a weird one. I think you just to Yeah, I think uh the the the best of the best will still want handdesigned stuff. Yeah. I I feel I I know I know personally people who have I don't know if they still work at those companies but they have worked at companies uh doing Figma prototyping. I that that that's kind of like very old like why would you do Figma prototyping now? specifically like Figma prototyping. We're working I'm I'm working on a project right now where we've just showed the client a Figma prototype. I do think it helps. I do think it helps.

Um like granted, if they hadn't done the prototype, we probably could have, you know, whatever. Um just gone with the static design and I could have just used my agent to build it. Um however, no. I'm saying like why not do the prototyping in code? That's what I'm saying right now. That's what I'm saying. Yes, we could have done that. However, I feel like there's a bit of a um well, I mean, I suppose we can throw away we can throw away the code um like afterwards. It doesn't take much to actually do it. Uh but yeah, you're right. There's there's a communication standpoint or there's a they call it political. Do you know what I mean?

like this is the this is the design process and there could be not worry but there could be some um how do you say uh assumptions made because they're seeing it in code. I can imagine them being like but we've already got it like what what like people people will might freak out or whatever because it's like well hang on like we've already got it in code like surely this is done and this has happened and this is actually quite separate that the chance of that happening are slim but this has actually happened to you to me is when people um see the front end's done right because it's taken you no time at all to build the the front end and click around do the prototype type they think the app is done because the visual size is done.

It's like well hang on just a moment. Yeah. And I I got really stressed out on a on a on a project a few months ago where it's like just chill out because you've just you've just seen the front end. I need to wire all of this up. None of this is really working. We're not sending any API requests or anything yet. I'm getting feedback on the design or I'm getting feedback on new features that they want and this and that. And it's like, you know, it's a nice way to kind of pace the project if you can show them a uh a prototype. That all being said, every client is unique and you've just got to play it carefully.

So, the option is there to build a a prototype. I do think there might be a time and a place for it. However, I do agree with you that we need to be pushed to create a prototype um with code a lot more for a more immersive and uh accurate faster like the the work that you would do in Figma manually connecting those buttons creating those stats in Figma if yes like if you don't do it with an agent you have to do it manually. Well then, well, that's what I'm saying. Like, why don't you just do it with an agent? Could that not be another Well, at that point, why why don't you just do it in code?

You know, if it's an AI. Yeah. Um because the Figma prototype is the speed thing, the speed thing doesn't hold water is what I'm saying in that instance. It's not about the speed. It's it's just again, if the communication works better that you do it through an a uh through the prototype, then do it. and then use your agent to make all the the links and whatever. Um, I'd be interested to know on the project that I'm working on whether the designer did actually use an agent or whether he actually hand coded uh piped all of the links together and stuff like that because yeah, I would totally expect to be able to do that as an agent.

But yeah, I'm sort of like I'm I'm in total 100% agreement with you. However, I can't help but remember some of the projects where they see it in code and then it's like they start their mind starts running away with them. Then it's like hang on like just don't get too excited. This isn't this isn't a working thing. This is this is this is all make believe right now. So yeah. Yeah. You you need a you need a button that says this is a prototype. Yeah. Yeah. Yeah. Take it with a pinch of salt. I mean, this is a big problem. Whether it's a prototype, whether it's a coded prototype, whether it is a flat design, like I always find I'm having to be like, "Look, just chill out.

This is a work in progress. Don't focus on the copy." Inevitably, they focus on the copy. Don't worry about the fine the imagery. It was just stock imagery. They worry about the stock imagery. Like, no matter what you say to them, they'll just start picking holes. And I think that's probably just human nature, which is annoying. Yeah. Anyway, do you want to see our StreamYard clone built by Grock 4.5 in uh 10 minutes and 9 seconds? Show me. Yeah, I have also something to show. Yeah. Is it related to what we've just spoken about or are we It's related to a clone. No. Gro. Okay. Cool. Cool. Cool. Cool. Cool. Cool. All right. So, let me just find it.

You wanted a landing page. Here is your landing page built by Grock 4.5. It does look like a claw design. It's just this just shows how how prevalent this is on the internet, isn't it? This is this is, by the way, this is when I say like building 500 websites in an hour. This is the look that I'm going with. I'm not going with something crazy. This is this I know we hate it because we've been seeing it over and over and over again, but this is kind of like the winning layout that and this is the layout that 95% of the websites should have and they don't have it. like just look at the old ugly websites, they they they can totally use this layout and it's readable.

It's usable. It's not award-winning. It's not beautiful, but it's Yeah. not distinctive. I mean, it's not ugly. It's when I say it's not beautiful. It's just not amazing. Anyway, so so yeah. Uh classic. Uh oh, we do have separate pages. Look, we do have separate pages. I didn't know that. Nice. So, yeah, we got separate pages. Whatever. More importantly, we open up the dashboard here. We have a dashboard. Oh, where I have we've got some um destinations. We've got a brand kit. We've got recordings. So, I've There is something Yeah, there is something in the dashboard that tells me this is built with AI. Yeah. Uh can you see what that is? Like I I I'm not I'm not focusing on the design right now.

Like this is about functionality. Okay. So, sure. We've got Oh, does work. And and oh, this is what we haven't tested. Check the private messages. Don't reveal this URL. Just check the private message. Click on that. See, up on the screen right now, guys. Oh, it's not working. Well, because it is like on your local machine. How would that work? What's happening? Uh, it says this page isn't working. is currently uh unable to handle this request. 5002 interest. Let me just let me just double check the URL because um I would love for this to work. Interesting. Well, actually, I'm I'm going to say like we could clone this. How would we do this with the the API keys?

I I'm not sure about that. Do we need like API keys from X to do streaming there? Because if we rebuild this entire uh Streamyard thing, then anytime we need a specific, you know, thing to add to it will be just a prompt away. And then we can also do my Twitter and your Twitter and we can do like, you know, more channels. Yeah. that that would be cool. So for those who are joining the stream and have no idea what we are talking about, we're using a software to do this stream and Sam just prompted Grock to rebuild this software that cost 300 bucks a year. Uh potentially more because we want to add more channels.

It will cost way more. And just 10 minutes later, we have a prototype working. It's an MVP. I think if this was live, the the the issue is here is actually more so that I'm just trying to set up tail scale. So, um what is tail scale? Is that like a tunnel? Exactly. Yeah, I think I used it. So, again, for those watching the stream, I kind of have to connect to his computer, but the software is on his computer. So tail scale uh I guess it creates a connection that like from the outside world you can access uh whatever software that is running from on in his dev server. Uh so I can connect so he becomes so you become a host right in a way it's like that's how it works you host some files.

Yeah, but we could we could put this on Versel, but this is just Yeah, this is just for the demo for the demo sake. So, we're just we're just getting somewhere. I'm getting Grock to uh fix it. So, um meanwhile you do that. Maybe I can show the design that Grock did for me. Yeah. So, this is my website, my current website that we are like redesigning. Uh if you notice we have like a hero section. We have this section with our past work. We are showing it off and we have like this section. And I asked Grock literally here I said like create a clone redesign version of my my website homepage use webgl and creative coding kabarza.com.

This is a very minimal um prompt. I wouldn't prompt like this normally, but yeah, that's what I did. And here's the result. Let me reload. So, it's Oh, it's hard to see. I don't know. It's not It's not amazing, but I guess yeah, I get the idea. So, it it didn't use my I I mean, I didn't specify anything. Okay. So, I'm not complaining about why didn't it do this and that, but it did use 3JS and Gap. Um, and yeah, I guess this is not the this design, but it is it is what AI does with a oneshot prompt. It's it's not too bad u for a oneshot prompt. Not for me, but I I guess this this could be usable for I don't know a smaller um client or somebody who wants just a simple website.

Simple. Yeah, it it's Yeah, it is what it is. Do you want to try that link again? Sure. Oh, yeah. It is working. I don't see myself, but I can. Did you put Did you type your name in? Yeah. Yeah. Yeah. Oh. Um, you are you're in the green room. Oh, now it's asking for Wait, give me a second. Actually, I'm going to allow another camera just for fun. You're in the green room. Guess status backstage. Stay on this page. That I'm in the green room. No, but it seems like you should leave me. Uh, you should let me in. That's how it's selected. Yeah, I can't see guests. Let me Oh, sick. Yeah. Okay.

Well, maybe maybe we can't I can't see. Whoops. Enter. You show the link. What link? That's fine. That the invite link. No, no, no. People can't get it to it from there. Oh, anyway, I see my camera. Okay. Well, yeah. I wonder where the my first studio guests. Um, yeah, I don't get any notification. But it's getting close. Yeah. I mean, for 10 minutes, you know, there's a lot more that we would have to do to um Oh, and I did a test whilst you were talking. I did a test and it records to your local machine and all the rest of it. I can download the local recordings. It's uh it's pretty good. And you get me.

I got to go. Um so, uh do we have everything? I think we have everything. Give me a second. Um, I wonder why. Let me see if I can join twice to this app. Your name blah blah blah blah blah. Allow. Enter. You're in the green room. This is what he would have seen, but obviously his own thing. But then over this side, we don't get any. We don't get any indication that someone's in the green room. Anyway, I'm back to say bye. Uh I I could hear you the entire time. Yes, that was accurate. That was what I was seeing, but I couldn't get in. Fair enough. Anyway, so all right, cool. That's all we got time for.

Tune in next week. Like and subscribe on YouTube and anything else. You've been Sam. I've been Sam and I've been Kar. Keep on vibing, baby.