#23 200 subs! 4.5 Haiku, Is Vibe Coding DEAD?! NVIDIA Spark, Weird AI Jobs

We're live. Awesome. Say hello to the world. Hello. Hello. Hello world. I'm going to click go live on Instagram as well. Oh, is that like a No, they're all there. They're all there. Um, yeah. That's not like a secondary thing I need to do, is it? Or you I have to do it. Yeah. Oh, wow. Okay. How strange. There you go. Hey, Sam. How are you doing? We have another Sam in the chat. Oh, okay. I think I thought you were gonna ask me how I'm doing. Yeah, we've been talking for an hour and I haven't asked you how have you been. Terribly. How have you been, Sam? Terribly. It's getting cold. It's getting awful.

This is This is normally where I disappear to sunnier sides of the planet. Where would you like to go? Cape Town. Cape Town. I I would have guessed it. Yeah. South Africa. Yeah. Just coming into the summertime now. It's spring, you know, baby lambs are being born, people being murdered. Just love it. Love Cape Town. Um, yeah, I was talking about Cape Town the other night and I was like, yeah, that if if I could just like teleport somewhere, definitely. What where where would you go? Uh, I don't know. I've been into Italy Italy a lot, so just kind of feel Italy. I love the weather and the food and yeah. Yeah, probably for my friend's literally talking to me about Italy right now over WhatsApp.

I haven't been in uh so many countries. Um but probably Spain as well. Yeah, Spain's pretty cool. Spain's pretty cool. Um what do you think of this backdrop? Am I in Am I in the Matrix right now? Yes. Metaverse. You're like metaverse. Well, you're in metaverse inside of metaverse. You see? I don't know whether it's Yeah, because of the, you know, bars at the bottom. We need need something different. But or the or exactly the same. It just continues and I'm just like a floating just like a floating head, you know? Maybe I should wear a green t-shirt and then I'd just be a floating head. Yeah. Uh, I was actually like downloading an asset.

I wanted to create a video with it, but I I couldn't. Uh, when I send it to you. Yeah. Um, like maybe I send it to you on Discord and you tell me if you know it. I was going to say put it here, but I can't download it from here. And yeah, we can't show it. We're not allowed to show it or what. You you could you maybe you can also set it as your background before we start. It's quite funny. Oh, I see. I got you. Let's uh let's do this. Yeah, that would match with mine also pretty good. I I think so. Yeah. Yeah. But your head is a bit a bit too big for that, you know, scale.

Mhm. No harm in trying. Okay. Right. Um I will duplicate. Oh, what did that do? I just pressed command D and something happened. Um copy paste. Do command AI, not command D. Yeah, exactly. And you ready? Oh, that didn't that didn't um have the impact I wanted it to. Why is it do that's See, look at this. See what I hate OBS. Yeah. Why is that there? Who knows this background, guys? Well, it's a matrix right now. I don't even know what anyone know this background. It's too light. It's too light for my um white balance. You're from the headquarter. There we go. That's We have Sam joining us from the headquarter. I need to I need to like whiten this a little bit.

How do I do that? Yeah. Oh, no. Turn it off. No. Here we go. It was pretty good. Yeah. No, it's there. That's what we want. White. Yeah. This is This is uh This is Web Flow's background. The thing is, yeah, a little bit off for my liking. Oh, hello. It's quite bright. They're in a bright white studio there, aren't they? Yeah. You got to have it really bright to, you know, have that quality. Anyway, are we are we are we gonna go? No interesting news this week, unfortunately. Well, obviously interesting AI news, but nothing. Yeah. Anything on your side? No, no, just uh just building. Well, I mean, my my exciting week has is part of the news.

It's the new model that's been released. So, yeah, I can my my week's events have been deferred until later on. Yeah. U mine has been quite busy with client work. So, yeah, we're doing some exciting stuff, but I can't show that. That's the most Yeah, it's so frustrating because of you know clients don't some clients don't want to share the work uh which makes sense uh before uh launching their product and yeah we are we are preparing a client for a huge launch. So it's um yeah that that's something that I will show in the next few weeks. Okay, good. I'm just trying to like participate. You were ready before I sent you that image and then I sent you that image and I I feel like I broke the show by like I'm looking.

So, I'm looking at you on the screen right now and I'm looking at the camera and now I'm looking at you on the screen. Is it better that I've I've moved my monitor like Yeah, it's pretty good. There's my monitor. No, that that actually looks you know a little bit of it is fine. Uh I do this by the way also sometime like a little bit but mine is you know black you yeah you can kind of like there's my monitor it kind of works kind of works it's like I'm working at my machine it's like I'm actually looking at something anyway we're going way off track right let's get the show on the road it's um it's neither funny nor nor professional how is everyone doing how is how is the chat doing over your end I haven't even got Um, we don't have a lot of comments.

We have Hi guys, we got Robert. Robert Petros Petras in the Haven't got my glasses on. Sorry if I mispronounced your name. Hey. Hey. Right. Hey, Robert. Hi. Cool. Yeah, let's do it. Let's get to business. Should we? Yeah, I'm good to go. Okay. So this week, Google dropped their new video generation tool. Claude introduced a cheap and fast model that is good at coding. Nvidia started shipping their um mini computer that is built for AI. And as a shocker, VIP coding might be dead already. This and a lot more in this week's AI. Uh I'm Kavarza. And I am Sam and this is a week in AI design and dev. Yeah. Cool. Um yeah, this week we we don't have a lot of like news like crazy ones like last week, but we have a quite a few like decent ones starting with maybe we should start with the Google VO.

All right. So this is a response to the one that um OpenAI released, right? kind of I feel like Yeah, probably. So, as a response to Sora 2, the video generation tool from OpenAI, Google released their new VO3.1 model that is pretty good and I made my first video with it. Sorry guys. I'm so sorry for what we are going to put you through. Um, let me bring it up. And oh my god, the future is here and it's being shaped by the incredible advancements in artificial intelligence. Yeah, that's I don't know why it feels so awkward, but it speaks better English than me. I mean, yeah, it sounds um it sounds nothing like you though, you know.

So, so talk talk me through that. How did how did you do that so quickly because you you literally signed up just before the the show? Yeah. So, did you provide it? So, first of all, I couldn't sign up with my normal email that is like a professional, you know, Google Workspace email. It which is paid. I'm paying like 400 bucks actually maybe 600 bucks for myself and my team but I couldn't use that. It asked me to use my private Gmail which I did. I signed up for the Google one. This is my first time hearing of it. Google one just like Apple one for Google one and the first months is free. I signed up and then I just gave it a screenshot of literally this like myself in this scenario and I said make a video about command AI and say this.

I meant something else and it said something else. So it didn't say what I wanted to say. Um I know it's 8 seconds maybe because what I had uh planned to say was maybe longer than 8 seconds but yeah it just through Google Gemini just sign up for Google Gemini and there you can create a image ask the AI just like you know chat GPT or create video. So what is nice it's all under one UI. It's in one place cuz it looks really good. I mean like you know just to flash it the future is here and it's like it to give it one image and that's what it comes up with. Yeah. Pretty pretty impressive.

Yeah. Um, yeah, it it also it preserved a lot of like the the stuff like my my shirt is kind of like the same, but it added some artifacts. I noticed that the plants are roughly the same, but it kind of I see that it is not my room. It has a lot of like resemblance of the room that I'm in, but it's not literally my So, it kind of like has some artifact. So, it's not like 100% correct, but it's really impressive. Really good. But I would still not use it. Not like I wouldn't feel like I when I say I wouldn't use it, I if I would want to say something, I would say it instead of doing it with that.

But I have a really cool use case for it. So last week, I don't know if you seen my Instagram or not. I mentioned it even after the show. I said like I'm going to shoot my friends. We went for uh laser tag. So just to play laser tag and something funny happened. I got in in a conversation with the laser tag guy. So when when you go there, regardless of like how good you are, there are always like a friend from the friend group who hasn't played laser tag. And for that reason, uh, the owner has to give all of us a rundown of how everything works and the rules and, you know, the introduction and he gets a vest, you know, you wear a vest with um, with the laser gun and with the sensors and he has to explain it all.

So imagine he does this, I don't know, like 10 times a day. Now you might know where I'm going with this. So he has to repeat himself and explain the same thing over and over again like shoot here, shoot there. This is where the sensors are, don't do this, do that. Things like this. So I got in a conversation with him and we got to a point that it was very interesting the idea that we could use an AI to transform. So he can record himself and we can transform it using AI and make something like potentially interesting out of it. So if he just records himself with a camera and say it and then show it on a big TV, people might not be that, you know, interested to to watch that.

But we thought maybe with AI we could um this is where it gets a little bit tricky with these artifacts. But if we could reliably take that video and just add effects to it would be amazing. For example, he does the full explanation but then we turn him into Shrek or Darth Vader or like you know any other character. you know, copyrights is another issue because he would be using it in his shop, right? So, I don't know that's probably like commercial use because it's in his shop, right? But he doesn't publish it. Anyway, so I thought like with today's tech, what is really cool that I could like at least tackle this challenge uh of making a video of him in multiple themes, right?

And then create an app. And I have no idea on what platform, but I know with vibe coding and you know having you and having I could I could like get enough help like for you guys to help me at the right direction uh to point me at the right direction and then we could create an app that you he could have on his TV right in portrait mode and people could like choose one of the themes and get the full rundown of how things work in the theme that they want and maybe even with some logic like if they have some questions they would just tap on the screen and they would get you know a video respond uh from the guy in you know for that question.

So that is a use case like a use case for AI. And what is also an interesting point he thought every time that people would watch that video he would have to pay that what was it um the cost of like open AI Sora 2 API like 50 cents per second something like this. He asked me how much does it cost? I was like 50 cents per second and in his head it would cost 50 cents per second every time people would watch it. But obviously we know that's not how it is. But it's also interesting that a lot of people like normal people and this guy is like even a little bit techy. Maybe a bit outdated but still techie.

He didn't know how that would work, right? If it's if you own the video or you don't have the ownership of the video. There are still a lot of questions and we're it to me it feels like we're so so freaking early and just knowing that we can do these things. It got me really excited and he gave me like extra games by the way and I got free consulting session paid off. Yes, it did pay off and I'm sure he will give me many more games if I end up like creating something for him. But yeah, I I should I should try to tackle this. Uh it's really fun. This is like somewhere that I don't feel like, oh, AI, I'm not going to use it.

No, this is somewhere that I'm like actually genuinely uh interested in using AI and I think it's a good use case. Not bad. Why isn't you just recorded that intro anyway? I I have my own It gets boring. Yeah. What? Just to just to do to what? Well, who boring for who? for the people probably not paying attention. No one pays attention. I've I've seen it before like you know you watch a video explaining it's just not engaging. So why would this be more engaging because I I'm trying to take the I always so as a product kind of person I always just technology comes last right technology is the last thing to get at it.

Right. Why why would you think the the special effects aspect of this will make it less boring? Yes. Yes. That's that's the idea. What what what we could do is this is like over engineer engineering the out of it. We could have a camera analyze the faces of the people and if they don't pay attention be like you didn't pay attention you have to watch the entire group has to watch it over again because somebody didn't pay attention. Yeah. Yeah. Yeah. I mean at the end of the day Yeah. something like that is is the sort of stuff you find at theme parks, right? And they have an introduction. It's all like whatever. But they would have to go out and shoot that and um you know, pay for costumes, pay for special effects, pay for the filming, whereas yeah, it just brings it brings exciting opportunities like that to to the everyman, you know.

So yeah, I'm all for that. I'm um Did you know about VO3 before meeting with him then or what? Like was it just coincidence? No, I knew about uh the Sora 2. Okay, fine. We got a we got it. Uh when I go to places like this, I talk kind of like business with them and also distribution and kind of like social media. Yeah, let's talk business. So, it's like how do people know about you? like what how do your clients get aware of um you existing in the first place? I ask this question quite often. It's just genuinely interesting to me. And then obviously we talked about websites and web and then it got into AI because you know and then and then he talked about like his um iss like challenges with business.

One of them being like doing a lot of these introductions or like explainer things over and over again. And I don't know, maybe maybe the video from Google is not the answer. And maybe we need to buy him a robot from Elon Musk. Yeah, I think that's might be it. It might be it. saying that we didn't cover it, but there is another team of people building these robots and they released a promo video last week of it like folding clothes and like putting like they're actually releasing real footage and have you seen it? I don't know what the company is. I I saw the like the 3D um video like kind of like the 3D render.

I sent it to my Oh, no. This ain't a 3D render. This is literally they've got a house that they've they test everything in and they've got robots like walking around the house like doing actual things. It's wild. I'll find it. I'll find it. Um I bet you're a hoot and a holler when you're out trying to go play laser tag and you're though. So, how do people find you in your business? Just just get out there and have fun. Kabaza, you know. Yeah, it's um it's just typical me asking about that. Yeah. Wow. But I want to try this. But I noticed I can't kind of like these AI models, they don't edit your existing video.

They just take an image and create the video, but we need to edit the existing video. Yeah. Yeah. There'll be ways. There'll be ways like Yeah. Comfy UI I feel like can take inputs and then modify and whatever. But uh yeah and this you know this is why this is why I love or the sort of you're seeing the benefit I think of just exploring new technologies no matter what it is in what field because what I just love doing is I love listening to people's problems and then using my like database of like knowledge on tools and what they what they can do. I can sort of figure out something to do like something to solve their like to solve their challenge.

Do you know what I mean? Like you you know you you wouldn't have may not have been able to think about that if all you did was live in the world of web as an example. You know, not to say there's anything wrong with that. I'm just saying this is what I enjoy. I enjoy just like going into my brain and like like just thinking about like how can that be solved and what tool and how would I use it and stuff, you know? Yeah. And like honestly th this is this gets me also into another topic. We we should talk about vibe coding later on. We will talk about coding. We will talk about vibe coding later on.

But when you go to these places, any place anywhere you go you if you talk to the owners or even like the people who work at these places. So it's whether it's a bar or like you know laser tag like a theme park anything anything at all. Um you and you look at like the way they run operations and they have like things that they do over and over again and you know what I'm getting with this uh into like automations and things like this. A lot of these businesses use like super old software like really bad software software that doesn't have like proper API software that is just like hard to use and a lot of things that they do over and over again that could be done with automation and as building software is getting easier these people might like start doing that themsel or they you know as it gets easier more people can get into building software in whatever means that is vibe coding being one of them and then these problems can be solved and we have my point being we have a ton of them we have so much software to build that is to me near infinite amount of software that we need to build that until AGI it's you know this is my whole sales pitch at the moment this is where I'm sort of pushing ing my company to do is like building software used to cost a a lot a lot of money right to build bespoke that's why SAS was born that's why you know to have some sort of internal computer infrastructure was like so costly so AWS was like well we're just going to provide the infrastructure we're going to distribute all over the world and you you won't have to store your you know your your website or whatever it is your your um your infrastructure on premises, you can just store it in the cloud, everywhere's good, you know, and it it kind of spawned from there and and SAS said, "Look, why build your own piece of software when you can just rent it from us for a low low price of 3.99 or whatever it is, right?"

So, this is my proposition to say, "Right, the tables have turned now. It's now much simpler, much cheaper to build your own software like that the way that you want it." Because there was always a compromise with SAS. It was always like, well, I like most of what it does or but I don't use XY Z, you know, I don't use this, that, and the other or like I I doesn't do exactly what I want it to do, but it mostly gets there. And those are the conversations I'm having with my clients at the moment. It's like, you know, they they compromise on the software that they're having to use because it doesn't do exactly what they want or the UX is just terrible.

Yeah. That that all being said, it's like now it's like well now you can just build you the software as you want it. It doesn't cost you hundreds of thousands of dollars, which is what it used to cost. It it can cost far less than that. And sooner or later, you won't be asking people like me to build that software because you'll be building it yourself. You'll be prompting it. Right now, we're still in this middle period of like, well, you still need an engineering degree. We're going to get into some of that sort of stuff later on, actually. You still need an engineering degree to be able to vibe code. You can't People are learning that you can't just vibe code things as it stands right now.

But we totally will get there. So that's my business. That's my proposition. You know, just build your own software big or small. Problem comes of how you price that. You know, how do you price something that realistically took you a week to build? Truthfully, probably less. um how do you price that in reasonable rate to say well yeah still it's still going to cost you a few grand because there's a lot of expertise that go into that but it didn't take me 3 months to do but anyway um one only a quick one just to say that we hit 200 subscribers on um on YouTube which is really cool actually way more now because I wrote this down like beginning of the week or I think we're now on 230 So, uh, just wanted to say thanks to everyone for subscribing and watching our content and liking it and stuff cuz it's, yeah, uh, yeah, it's nice.

It means a lot. And I don't know, I can't remember how quickly it took me to grow my, you know, my personal channel or I don't know how long it took you to get to 200 subs, but it just feels like, yeah, it's building up, this, this community is building up a bit of momentum and I just want to say thank you for it and uh, sky's the limit, you know. Yeah, you are early in this and yeah, one day you'd be like, "Oh, I remember when they said 200 and now they are at 200 million or something." 200 million. Yeah. Yeah. I was saying that like you know um I wanted to I wanted to mention the 200 thing but at the same time it's an opportunity just to kind of reflect as well because we started out just kind of being very because I can see the development and I'm I'm really enjoying kind of where it is right now whereas we used to just do the news and my focus was to get out the news and then move on to the next one.

Whereas now it's like the news is simply a a jumping off point for a discussion and that's what I really like doing you know and we did this without you know having a show like we we used to meet on uh Fridays so like we used to meet on Fridays and talk about business talk about like sometimes just life sometimes about you know what's going on and then we would just nerd out. Yeah. Geek out. Yeah. Yeah. Because what was the original name? Was it Geek Out Live or something like that or Yeah. I Yeah. Or V about maybe I don't know. But like Yeah. But then AI started peace out. Yeah. Um because then I AI started to get really big and I started to be like well there's a because we you know we obviously want to talk about things that we're interested in which is or sorry that's relevant to our online personas you know because I mean to be fair I am my online persona.

I faking love coding. I will I I what was I was at a party on Friday and I was literally there with I was the last person there me and my friends just drinking lounging on a sofa. What were we talking about? Friaking AI and code. It's like I'm at a party. Do you know what I mean? It's almost as bad as you going to like laser tag and talking to about business opportunity. Yeah. I live and brea I genuinely love and I think we've actually got a comment from someone saying like they've subscribed because they see our genuine interest in what we talk about. We're not we're not just regurgitating stuff for for likes and views and and you know we're not just we're not hopping on a on a trend for the sake of that you know I for better or worse I c I just cannot operate in that way.

I need to have a love and there are some downsides that come with that in that it it it just becomes hard to progress in a lot of things because it's like well what if I'm not interested in it. Some people can they can literally just follow the numbers and just you know for lack of a better word sell their soul to like do something but I can't I think that genu that genuine that genuinity genuine that genuine nature I think comes across. So um I know what you're trying to say. Yeah. Uh anyway, so yeah, happy 200. Um and here's to the next 200, I guess. Oh, and also, yeah, if you are if you if you want to sponsor the show, if you want to help us grow, I know we're not offering a tremendous amount, but you know, my subscribers over on Twitter, hear about this, if Hello Twitter.

Uh Kabaza's audience on Twitter as well um get streamed to all these different platforms. And so we are we are somewhat further further a field than just that 200. But obviously we the the core audience here is on YouTube. So yeah anyway sponsorships welcome if you want to chat to us and support the show. Cool. Yeah. Onwards onwards and upwards. Yeah. So so we we kind of alluded to this sort of stuff. A recent report from the duh crust data suggests that vibe coding is dying. Or is it? I guess I should share my screen, right? Um, and it has already been shared. So, we we we kind of just did touch on this sort of thing.

Here's the graph that they produced and it's this is I don't know about you, but this is a suspicious drop for lovable almost like something happened. either they weren't they started not reporting on certain data or they stopped a certain marketing spend or they fired a certain marketing team or something. It feels like this drop that's happened is probably more true to um their numbers than what was way up here. You can't really see my m you can't see my mouse at all, can you? But I'm on the I'm on the hill. I'm on the high bit anyway. So on the hill you see 35 million uh these are like 35 million users on the site monthly, right?

Yeah. Yeah. And now it's settled to around 20. So you know which is a crazy number because we know like there are roughly 45 million developers out there which is also a huge number. It's not a small number. Uh there are probably more designers by the way probably more designers. Well, when I say more designers, I don't mean like necessarily like Figma designers. I mean like in on a broader sense, anybody who would doodle something on the computer. That was probably because of the the um design joy thing. That was such a great marketing because that's the thing. Love are great at marketing. They're not the best tool, right? They just look nicer and they just their their personal online personality is great.

But that design joy battle that they did, you know, beating web flow, I think that really opened the doors to a big audience, you know. I agree. Um, so yeah, the trend shows that lovable is dying. Is that true? Well, I think that as I said, like I think their lovable is dying is just the clickbait. They are then this guy, it's this guy's article, by the way. Um, and he basically goes he talks more about the broad idea of of vibe coding and whether vibe coding is dying. So, um, I thought it'd be quite fun to kind of sort of speculate, well, what is vibe coding? What is it good for? Why do why do these people think it's it's dying?

Exactly. Um, because yeah, obviously it's a very exciting prospect for people who don't know how to code to now suddenly be able to code, right? Or or not even code, sorry, they don't care about coding. That's that's the key point in all this. They don't care about coding. They just they just want their product. And I think that is such an attractive proposition for anyone. Like it's an attractive proposition for me. Like I as a as a hardcore coder, I love to code, but I'm able to flex my my UX uh mindset. Um you know, how best to work. I'm building a tool for someone at the moment and it's just a simple Google Drive integration.

And I vibed the first part of it. Well, I say vibe. Where is where do you draw the line? Because at the end of the day, I planned it. I used claw code. No, I used taskmaster to plan out the feature. But ultimately going through phase by phase. I was like, look, I'm just trusting you right now because I literally didn't know how to build such a feature. Yeah, I definitely vied it from that perspective, but you know, the UX that it came up with wasn't optimal. You know, there was a couple of different ways to do things. We already had an integrations page. It was like, why am I being shown this? Long story short, I was able to flex my UX design skills.

And you ask me quite often, like, what do I do whilst I'm waiting for my AI? I'm using the app mostly and clicking around finding bugs and and and thinking from a UX perspective what's being built and does this make sense like I'm able to flex those muscles whilst I'm coding you know so I don't know where I was going with all that but ultimately like the I I do see a future where writing code is just is not is not the outcome. saying that what's his name? Code Garden or something like this did release a video saying that he he's he's done with like AI coding and that he wants to take a break from it because he's just he's not getting fulfillment out of the AI doing the work and that is totally fine.

Like we're all we all like what we like, you know. Um, but I do think coding will be relegated to hobbyists as AI gets better, as the tooling gets better, as our methodologies and our workflows start to solidify and get better. You know, you got spec driven development at the moment and um, context engineering and all this kind of stuff. Um, yeah, we I I wouldn't 100% agree with that. By the way, I'm not with like coding becoming something for hobbyists. Uh, I believe like driving a car would become will become something for hobbyists. Uh, because ultimately there is like no need I I believe like for driving a car. uh in I don't know what whatever time like in 10 years but think of the goal of you I think you're describing exactly the same thing the goal is exactly the same it's something in the middle to get to the outcome idea to to product yes yes and no that's that's where I would disagree that it seems like with coding it might seem like that um we just want a outcome like a product as an outcome And however we get there, it doesn't matter like if you code it or build it with web flow which is like you know a UI for code or you just vibe it.

It shouldn't matter. But I think there is like more into it. That's where I would disagree. So for the first part of like building products that are mostly userfacing um I agree that it's like going to die like this this way of writing code is going to disappear but we still have to develop um a lot like for the infrastructure and those will continue to use like coders like imagine from CPU architecture like the things that teams at ARM do I I don't know much about them but I know enough that these these things won't be like replaced by AI will be assisted yes but I don't think these are things because these are so low level and when I say lowle I mean programming um sense of things lowlevel so also complex And they are the like the mistakes that you might make will be they they uh they are going to be far more important.

So you cannot have like mistakes like um because like on that architecture level you have to optimize for the smallest things and there are from what I understand a lot of work to be done. We are nowhere near you know being perfect there. So we need to optimize those like le lower levels and we are talking about AI taking over coding. We are talking about the upmost level which is right now we are talking mostly about front end which is like the most exposed you know version of coding and also the most amount of code for it exists uh publicly for AI to learn from it and the deeper we go the the more important the work gets and also the less code is available and the more professional and like very um what is like you have to be an expert in that area to to do meaningful work.

So I think that will stay uh for a long time. Yeah, I guess I could have saved you an explanation there and saying yeah I'm strictly talking about software development. Of course there's going to need that AI is not going to take away from code in general. I'm talking about like software engineering and you'll always get those people who want to go and tinker and get you know in the in the in the context of of uh software engineering there will be people who still want to get down as close to the the bare metal it's not because it's still abstracted but but saying all that we're talking about vibe coding and is it is it dying right and like I think that um I don't think it's dying I I I think it's just the hype's just dying off.

Like people get excited, people want to try things. Like there's this wave of people that get really not offended, but they get really um like uppety about the fact that people are jumping on a new and exciting tool. It's like what do you expect? Like people want to see if it's right for them. People want to see if it does this and that and the other. So you're going to get an influx of users that explore this stuff. And sometimes this just takes a little while to figure it out. like you're not going to I think about you know if you got kids um which you do you know your your kids are probably going to want oh dad can you buy me this paint set or dad can you buy me this science kit and you're like they're going to you're going to get it for them and they're going to pick it up and they're going to play with it once and then they're never going to see it again because it's just not for them.

But you you've got to allow them to explore that, you know. So, and and sometimes they'll do it'll take them three months and then they'll move on to the next thing. You know, this is this is kind of what we do. I don't know what people kind of expect with all this sort of stuff. So, I think that people have thought, okay, I can build tools. Realistically, I think they're probably coming to the terms with the fact that it can only go as far as prototyping and even then your app is at risk of probably getting hacked or something like that. Like I said, it's still a big responsibility. They lovable have sold you the dream of building whatever you want but the reality of actually getting there is slightly more involved than this.

Yeah, this is an this is like an awesome uh thing to note and to mention that the idea of building an app, it's different from like actually building an app and maintaining it and maintaining and like the idea of like building a prototype and like getting a prototype is far like like further away than actually shipping something. Because when you have a prototype like we built a you built a prototype for you know a potential client and you did it in probably a few days right but we we we're thinking like that the actual app is going to take probably what six months minimum right I don't know I'm I'm I'm quite conf I'm quietly confident but you know it won't take a day you know it won't take a few days you know it's you know at least takes significantly more like order of magnitude more.

That's the point. It's order of magnitude more uh from prototype to final product even though the prototype might resemble like 90% of the product at the first glance. Like you look at it and you're like it's 90% there but that just looks like that that that extra 10% that you don't see is much more difficult to handle. And also there are like a lot of more like edge cases and like special use cases that you have to cover and that might require like rethinking the entire thing. So yeah, building apps is not that easy. No, that's the point. And going back to that UX piece as well, you know, you mentioned about mostly designers jumping on lovable and that's fair.

Um, but you you have an idea for an app and then and then you start to vibe it out and you suddenly realize the the nuances or intricacies of go that goes in particularly from a UX perspective like yeah, it does what I want to do, but it just it just doesn't work the way I want it to do. You know, there's there's so much that you need to think about. It's so much more involved than the dream of whatever. So if you know what you're doing, it took me as a developer a while to get my rhythm with vibe coding. And now lo and behold, I'm kicking projects off in replet, but then very quickly moving on to I get the starting point done in replet and then I take it into clawed code because that's when that's where I feel like it's done its purpose.

That's where I feel like I can get the most amount of it plus be efficient with my with my money because replica costs a lot of money. But um yeah, it's not that the dream. It doesn't it it skips the journey to get there. As with all of this stuff does like Apple are fantastic at that. They sell they sell you the dream of like having all of your devices working, calling friends and family and you know doing all this sort of stuff. They sell you they sell you the dream or they sell you the vision and but they actually but the realistic nature I mean we're talking about products here so it's it's not like you have to work to like call your family or anything but it's this idea it's simple marketing sell them the dream and then that's the best from a marketing perspective that makes sense right yeah it makes sense you know um because it can happen it can happen and that's we're going to get back into the YouTube rabbit hole but a lot of people on YouTube are selling you the dream.

Like they're they're they're they're um inspiring you with the dream, especially when you talk about money, you know, money influencers and stuff like that. They're selling you this dream, but the 0.001% of people who might even just sniff at that success metric that they're painting, we don't we don't re we think it's going to be us. It's going to be us that is going to be make it. But we don't realize that so many people will will will fail at that, you know, so sell them the dream. It's the easiest thing to market. Um, but anyway, back to the back to the vibe coding thing, you know, and that graph. Exactly. You know, I I it's probably makes sense, but like it's it's a bit clickbaity in that it's not necessarily dying, you know, it's just less people are finding that it's relevant for them.

they're not building the tools that they want to build. Whether that's they just can't get it to work for them or, you know, the strategic angle of that product isn't quite there. It doesn't work for them or or like they just, you know, I thought it might help me, but it doesn't, you know. So, uh there is another graph that I shared uh another tweet. Maybe you can also open that. Where did you send it? It's just in the docs. Uh uh the Anton. Yeah. And I want to mention something. So we spoke like a so this is about lovable dying or not dying. Well well not that's not actually the way I want. It's more about vibe coding in general if it's dying or not.

And lovable is riding this wave really well. It doesn't matter if if the news is good or bad. they are just good at spinning it the way that they want and kind of like ride the wave and on social media and get the views. So they shared also this other chart that shows their new paying customers that is just going up. But this is just their new paying customers that doesn't show any churn churn churn being you know users leaving. We we don't see that. We just know that there are more new paying customers, but we don't know how many are leaving. So, yeah, you know, I I would take this with a grain of salt.

I'm not saying they are lying, but uh it's a private company. We don't see much of like what is actually happening inside. Yeah. Yeah. It's all It's all just drama. Yeah, it is drama. It is drama. Um, yeah. Um, I'm I'm trying to I'm trying to turn this conversation into something that's more like less clickbaity and more just relevant to my perspective on on um, vibe coding and coding in general. I just think that yeah, I think people are coming to that realization that it's not really working out the way they intended. And there's nothing really wrong with that, you know. I think I to be honest, I'm finding a way to get off of vibe coding.

I want to try and replicate replets, you know, how they build apps and then I'll be all in on um all in on claw code or whatever, you know, next month it'll be uh 4.6 GM 4.6. Yeah. Um from for for from what I see um this is like contributing uh to the diff world in a positive way. So regardless of people staying or not staying with lovable the fact that more people come to lovable try to build a product I think is a net positive. Yeah, even if they get frustrated, they see okay like building an app is actually difficult because there are so many like as you said like UX and other points that people don't see and once you get into it you get you will have some appreciation for how difficult it is to build apps and also for for the ones who you know decide that it is not for them they will gain that appreciation and they leave that That's fine.

That's a net positive. And for the ones who come in and they started, you know, doing it and they fall in love with all that details and all that things that you need to build which it because it's not easy. So if it's not easy, it's probably you know sometimes it's worthwhile trying it and um I think a lot of people will graduate from lovable. This is something we wanted to talk about and they they become developers and that is also a net positive. So all in all together this is net positive for the dev world and I'm really happy about that. Yeah. And also you know those people who did gain that appreciation it might be not only gaining an appreciation but you might be gaining a client because they're like I tried vi vibe code in this thing and I couldn't do it so can you help me please?

You know so I can't fix your vibe coded app. Yeah. uh you know it's a deeper it's a deeper subject alto together but I think for whatever reason a lot of what we see people are trying to sniff out drama sniff out negativity from something that's just not negative there's nothing negative there's nothing wrong with it you know um people just want to sh people just want to shoot up don't shoot up is that punch up with I don't know they just want to punch down punch punch up punch down similar sort of thing like they just want to they want to punch up to try and be better than that person or that company, whatever, all punch, you know, punching down is actually terrible, but anyway, whatever.

I don't know where we're going with this, but um yeah, interesting interesting bit of drama that's going on. Um, more power to you if you're a vibe coder. No judgment from me. Um, and if you struggled and give me a shout, you just or us a shout, you know, we'll sort your app out. Yeah. I I I mean I mentioned this before. A client of mine, they delivered kind of like a prototype of the website that they want they wanted us to design and build for them in Webflow. They delivered it us to us in Lovable. So they used Lovable to get kind of like a it it was a prototype. It was kind of like a wireframe.

They were like, "Oh, we thought like we want to have like three boxes to explain this part of the the copy." And the copy was also written by the AI from Lovable and that was not final, but it was something that they could start with. It was funny. They could use Reum as well, but you know, people used Lovable because they are just so known. So, well, yeah, I say marketing probably led them to lovable and really probably didn't even know was there. I bet that was better for you as well because it's like cool, we can see what we need to build. Much easier to to slap down a fixed price on something when you can actually see what it is you're building.

You know, you still might get it wrong, but it's, you know, people expect you to fix price something. You literally have no idea what they want to build. You know what I mean? So anyway, yeah. Um, let's move on to the next one. faster. Hang on, let me start that again. A faster and cheaper model that is great at coding is here. It's Haiku 4.5 from Claude and it might become my default model for all my coding needs. It genuinely has. I've been using this all week. Well, I say all week. It launched pretty much yesterday. I've been using it all day yesterday. Um, and it's pretty it's really really faking fast. And it's better than Sonnet was like five months ago.

Let's get That's crazy. So, okay, just just to understand Claude they they Claude have set they have haiku and what was the name of Opus? So the idea is that the Opus is the frontier model like the best and most expensive model right seemingly theoretically. Um and then we have Sonnet which is like the medium range model and then we have Haiku that is the light fast cheap model but now the light fast cheap model is better than the other two. The light fast and cheap model is better than what Sonic was six months ago, five months ago. So, HiQ now is 33.3% on um software engineering um well sweet bench verified whereas sonet 4 which is what we were using dayto-day was is 72.7 so it's better than what sonet 4 is sonet 4.5 has since been released which is now better than opus and um well yeah previous sonet versions so basically what we're saying here is that we've got a and the speed is just incredible.

It it's really really freaking good. But the interesting thing is here it's not as good as GBT5 codecs. Um it's better than GBT5 on the Sweetbench verified and much better than Gemini 2.5 Pro. It the these are comparable to Frontier models whereas this is their fast and cheap model. They're not comparing it to realistically this should compare it to GPT5 Mini or uh Gemini 2.5 Flash but it's not. It's comparing it to the other companies Frontier models. So it's it's a really really good model to have. Um and and yeah, we we used to praise Sonic 4 when Sonic remember when Opus 4 and Sonic 4 came first came out. It was like whoa. Now Haiku 4.5 which is their again cheap fast model is better than that again.

So here we go. Yeah, 5 months ago, Claude Sonic 4 was state-of-the-art. Today, Haiku 4.5 matches its coding performance at onethird of the cost. Now, it has increased in price um by like a dollar, which is, you know, bit of a shame to see because you want you want stuff to get cheaper. It's $1 in,5 out where it used to be eight 70 cents in and then $3 out or something like that or $4 out. So it is, you know, um I want to go backwards actually. It is more expensive than what it was before, but it is a really good model. And um this means so not a lot of people have this, but um oh actually this is to do with um agents.

So what I've done is I have put all of my sub aents as Haiku. So sub aents, if you don't know what they are, they're kind of specialists that run in their own context window. So I have a TypeScript specialist that goes and fixes all my TypeScript errors. They don't need a lot of they get told what to do. They don't need a lot of thinking power. They just need to churn out whatever it is they're good at. So to have and that was the first thing I did before even reading the suggestions before even reading any of the reviews I'm like okay cool whack all my sub aents as haiku um and yeah so and because of that speed as well not a lot of people have it but you've got the the clawed chrome uh plugin so it's great it's just great at like quick summarization data it's just a really freaking good model to be fair um for those sorts of tasks.

Now, what I want is I want to be able to um what they used to have before Sonnet just became the best is you used to have Opus plan mode that would degrade to Sonnet degrade to Sonnet uh whenever it is you implemented stuff because you that's what you kind of want. You want the big boys to do all the thinking, the planning, the heavy grunt work, and then you want to delegate to the the runs, you know, the guys who can just don't need who just do the leg work. Do the leg work, right? And that's what I loved. I would do the plan mode, Opus, and then go to Sonnet. They've removed that since Sonnet just became the de facto.

Um, I want that back now, right? I want I want plan mode, Sonnet, and then Haiku. And then when I do the work, it goes to Haiku. So we can just churn it out and not use a lot of um tokens and whatever. So what I've done, I've built my own slash command called plan. So when I go slash plan, you can choose your model that that plan mode runs in. So I leave I leave my default model as haiku slash plan do this that and the other and that gets set as sonet. Then that's sonet 4.5. So it will run that and then when I actually do the work or when I don't run the plan work, sorry when I don't run plan command.

So, I've kind of rebuilt that thing. Um, that that system that I used. Um, but yeah, it's it's it's just great. It's it just it's just a testament to how far we've come really, you know, in just five months, what was the groundbreaking frontier model, state-of-the-art model is now being beaten by little cheap thing, you know. Yeah. Saying all that, I I've canled my claw code subscription. Um, I'm going to commit to GLM 4.6 six next month. Um, but this definitely made it slightly more tempting to stay on core code, but we we'll see how we get on with with 4.6. I'm going to go use it in cursor to build a gap animation. A gap animation.

Yeah. See? Yeah. Like Gap's easy, right? You don't need you don't need a lot of thinking for that. You don't need a lot of like it's not you know what you want to build. It's not like it needs to take into account a massive code base. You're just working in one file, right? Yeah, it's one file like roughly 300 lines of code. It's not doing exactly what I want. So, I want to try this new model. Uh, and I checked in cursor. Um, it's available. So, if anybody is wondering, that's that's where I have it and it's just simply available. Yeah. Yeah. Um, let's move on to Nvidia Spark. Did you want to take this one or should I or um Well, we Yeah, you So, okay, let me start and then you can go with like the the one that you had as well.

So, I don't know what this mini superco computer does, but I just know it's for AI and I guess it's for local AI and I kind of want to get it. Can you explain what it does? For four grand. For four grand. You sure to get it? Yeah, it was three grand by the way when they announced it. Uh, but still I Yes, if I had the money I would look how cool it is. It's so tiny and it's so powerful for how tiny it is. I'm not so sure about this gold this gold pebble. Like it's like a stone wall. I'm not so sure about that. But so, um, so I'm going to be totally frank like this only fell on my radar two, three weeks ago, something like that, maybe a little bit more.

Um, and this is a tiny uh consumerrade PC announced by Nvidia because so I I and again I don't know the history of Nvidia. Do they make PCs normally? There's a custom OS. You can see like actually this is running on a Mac. So I've I just I just don't even know anymore. Is this a Mac? It's that's running on a Mac. Yeah, it seems like it. Yeah. So is it Yeah. See this is this is where my limited com knowledge comes into comes into play. So Nvidia basically have released this tiny little computer. Um 128 gigs of unified memory. It's basically built specifically for AI right local AI and it is quite pricey like four grand is quite a lot of money but but what can it run?

This is like the main question like can it run haiku 4.5? Absolutely. Like this is this is um I watched a a review of it um a like a genuine review by I'm going to get him up because he's only I say he's only a small channel. He's bigger than us. But like uh you know, shout out to him. He's doing some great work. Um Behan Bowen, I think that's how you say his name. But this uh try to clean up my camera. Can you hear that? By the way, I'm gonna um I'm going to Yeah. So he was he was running comfy UI which is for generating video. It's like a nodebased thing that he was he was running on it and um it was doing fine like it was doing really really I want to see the LLM usage because I think he was getting into the I want to see his token usage because again here we go you can just see it live.

He's got another he's got another comp competitive product on the on the left here. Um, so it's I think it's it's marginally faster this one. The G is it GeForce or GMX, whatever it's called? I don't know. So, you can see the results here. Like, yeah, I don't know. I don't know what to make of this speed because I think we're used to, you know, obviously hitting Chachi BT and it just just flying out. Like, are we supposed to be that impressed with this? Is this is this really is this really worth like you know locally running your AI? There is argument to say that there is because I don't know I I sort of see a future we talk about you know AI being in this bubble and not earning any money and this and that.

I sort of see a future where it's like these services are just heavily degraded or they've got to like they've got to they've got to um scale down with them or they've got to do something where like 4.67 tokens a second. So that's to me that's not that impressive. Yeah. But from a local LLM perspective like I you know people chime in where they want to like I think this is for the size that it is as well for the amount of wattage that it consumes. I think this is this is this is fine, you know. Yeah, this is pretty cool. Can you chain them? I don't know actually. Like run them in parallel. That um that would be cool.

Well, um Oh, he's getting 64 here. Look, he's getting 64 with um OSS. That's um yeah, GBT OSS. So, I didn't I haven't I should have skipped through to his honest thoughts here, but um yeah, go watch that video if you if you want a bit more of a a genuine deep dive into it. But, um yeah, I I kind of want one, too. I don't know whether I want this one, but I you know, there's been times where I just kind of I would love just a little box just to know that I'm not hitting any server and that I'm just I'm coding completely off the grid. Do you know what I mean? off offline.

Um, where was it? There was something recently, maybe my internet down or something and I just thought, man, I just I'm like stuck right now. Like, you know, I mean, I could handr write the code, but we're doing that. Yeah. Um but yeah, so I there's something in my gut that tells me that one day we will wish we had bought something like this sooner rather than later because Chachbt goes down. All these things start like degraded service. We don't get they're not pumping as much money or whatever. So people resorting to buying machines like this and then but there's a shortage for some reason and it's like we should have we should have bought them when they were readily available.

Bowen here he walked into Best Buy to buy it, right? He was surprised that he thought there was going to be cues around the wall around the, you know, corner or whatever. He was able just to walk in, buy it, walk out, all done. That that's fun. Yeah. So maybe maybe we're maybe we're living in in times where that's even possible, but you know, maybe it won't be too long before something happens and we're wishing we had walked into Best Buy and bought it when it was freely available. Yeah, but for a grand I I kind of want to have it because it's so cute and I I like the hardware, but I don't know if I could use something that slow.

Another interesting like hardware news that also happened um 100 100 tokens per second isn't too bad for for GBT OSS is too bad but anyway we we had another hardware news that is kind of kind of I'm trying to make it related M5 just dropped Apple just dropped a new MacBook with a just base M5 and they're advertising it for precisely that for ondevice AI. So what I would love to see is like a comparison to see how big of a difference that would make. Maybe not necessarily with the base M5. Maybe something like M5 Pro or like M5 Max when they drop that. I don't know, maybe later this month. But that would be really cool because if I could run um like a model on my just my on my Mac without any extra extra hardware that would be really cool.

Another thing that I find super fun and well like a cool thing about a hardware like this you know a separate thing is it it kind of has like a summary of all human knowledge if you think about it. Oh yeah, exactly right. Yeah, it's it's crazy to think about like all the books, all the knowledge. Yes, it will hallucinate. Yes, it's not like perfect, but let's let's not buy the idea that these companies are trying to sell us. Just look at the tech and how impressive it is that we have a model, a mathematical model of all human knowledge packed in a machine like this. And if you could go back in time and you could bring this with you and somehow, you know, be able to turn it on with how crazy it would be, you know, have all that knowledge.

It's crazy. We like we take it for granted totally because again, you you take the calculator as a simple as the simplest example. Like that was unfathomable that we could do such advanced mathematics just at any moment's notice. You know, normally you Well, I say there was probably once upon a time where you need a pen and paper. You need a freaking abacus, you know. Yeah. Um, you know, uh, just to say though, like I am I'm running, um, how do I stop screen sharing, present screen sharing. Um, I'm running Alama right now, right, which is which is this is completely offline. I just asked it to write a 500word story. Let's go. Let's go.

Um uh write uh sexy SQL um 3,00 words and this is local. This is completely local, right? This is locally. So it's doing um it's doing some thinking here. Is it an M1 machine that you have? This is M1. This is M1 Max, right? So M1 Max. And what software is that that you're using? This is Look at it go. Oh yeah, that that's quite fast. Yeah, that that's that's actually really good. And I could use it. It does make me want to try GLM. I think GLM is open source. Like can I download um how do I how do I do this? I don't use normally, but you you know um with cursor you've got an Do you have that as well?

I didn't know. Cursor settings. Uh network. No, I'm sure there's an alarm kind. Yeah, you have a long list, but I don't see your llama. Okay. Uh models, view all models. Is it under API keys though? Is it under API keys? Well, you can add, you know, whatever model that you need. You can add a custom model. You can add as Bedrock, Azure, OpenAI. I'm sure there's a way to pipe cursor into you know maybe I'll look into that and do a video on it. But can you do open router uh to cursor? You can do open Yeah, you can do open router. You have to use the open AI um thing. You just put I think you just put your AP your open router API key in in the open AI key spot I think.

you know, um, where are we? Stop screen share. Present screen share. Uh, window cursor. Boom. Boom. Boom. So, here you've got Open AI. Can you see that or is it too small? It's quite small. But yeah, guys, if you're on Instagram and if you can't see it properly or on Twitter, just hop on on YouTube. We are live on YouTube as well. There you can go full screen mode all glorious and so in API keys this is where you put your open router API key and then I think you I think you override it here but I I wonder if there's again a way to override this to a local host with a specific port and then that's how you get it into but there must be there must some tool that allows you to pipe your requests to autoama but you know um can you like run cursor locally any like would you do that then locally right yeah yeah yeah so any kind of AI request would armor so maybe I can play around with that and see if I can create yeah that would be really cool because yeah we Again, we're talking about the Nvidia Spark here, and you, you know, you want something more fast, but I've got the M1 Max here, which is probably being outdone by the M5 already.

Maybe not the I think the Pro. I think it outdoes the Pro. Maybe not the Max. But the point is, it won't be long before we've got some pretty um good enough um you know, machines that can um are good enough. the world's information in our pocket or whether we're coding or something like that. So, yeah, I did have a look on Amazon, but I can't see it on Amazon or anything like that. So, I think we're still out of luck when it comes to the UK. I think it's just a US thing. Wrong wrong link. But yeah, Nvidia Spark, pretty cool. Pretty cool. Nvidia Spark. Uh, I'm excited about it. I'm excited. I will watch a bunch of videos and I will watch some videos on M5 as well.

This is like the M5. And do you see what they did here? Yeah. Yeah. With the five. So, well, here's the thing. Yeah. I told you, right? If you're in Europe in your Maybe not the MacBook, but the iPad. If you're in Europe, you don't get a plug socket. You don't get a plug. I think I think it's a green thing. I think it's like a um e-waste thing. Yeah. Not not even a cable. Do I get a cable? You get a cable. You get a cable. Okay. But like I have a 120 or like 140 watt charger for my MacBook. Uh like I have the M1 Pro, the 16 version. So it's a it's also a really big charger that I don't bring anywhere with myself, by the way, because it's so big and it has one port.

Yeah. Yeah. Um here's the thing. Too sad. You're not too sad. Honestly, I get I I honestly if they would if they would I know it's Apple and they they don't do that, but they if they would charge me less and not, you know, not include the charger. I'd be fine because go on Amazon, there are a bunch of like uh these like newer version like what's it called? Like they they have a techn the technology gallium nitrate. That Yes, exactly. the the charger are smaller and I have two of them actually. Yeah. Like around the house they're smaller and they they do the job and they have like multiple ports. Anyway, so M5, all I wanted to say here M5 was draft and it feels so wrong.

Yeah, I know. Yeah. And then you've got M3 Ultra. You you got M3 Ultra in that lineup as the best model is like the best one is M3 Ultra. Yeah. And then yeah, M5, M4 Pro and Max and then M3 Ultra. They they started really good with these uh chips and now it's not that it's bad, but it got confusing and it's like I don't know. They could have done a better job. I I would I I wouldn't I wouldn't worry about it too much. Yeah. Me neither. What? So, actually just on that real quick is like I So, I've got an M4 M1 M4 Pro laptop. M1 Pro or Max Pro. Okay. And I've had that for years and years and years and years, right?

And always wondered like what doing what we do, doing what as coders do, like what was I missing out on, you know, for the max. Always tempted to go the max. always, you know, I know you you've got the max, right? You're always No, I have the Pro. I have the Okay. I thought you said Oh, you got 16 in Pro. Yeah, 16 in. I have 32 GB uh of RAM, though. Yeah. Yeah. So, you know, as as the M1 computers start to get very old now, I treated myself. I say treat myself. I just bought the best laptop you could have bought, MacBook you could have bought. um uh M1 obviously and it cost like a grand something like that, right?

Bought it off a guy down the pub. It was very funny. And but you got it secondhand, of course. Yeah. I only bought it like a few months ago, right? Yeah. And I will be honest, I not doing what we do, right? Coders, designers, vibe coders, not really not really noticing anything at all. No difference at all. So if you are ever ever thinking, do I get the M1? Oh, sorry, sorry, base model, the Pro or the Max, I don't think we will never need the Max. Come at me in the comments otherwise, but we do not need the Max at all. It might be nice, but even in even now the M1 like I know I know we we YouTubers have a habit of not thinking about the future but I we are in the future right now and I'm using the best of the best and my M1 Pro is perfectly capable do performing the same as as the M1 Max.

We we never we never need it. So, I definitely think the Pro I wouldn't go as down I wouldn't I would I would hesitate to go down to the base model. Um I kind of say this in my M1 H sorry M4 Air video that I made. You would you could get away with it if you're tight on cash and you know you wanted something to get the work done, but you would I think you would inevitably upgrade to a pro, but never a max. Never a max. Yeah, I agree. So I would say if you are a student and if you don't do anything heavy with like any heavy software, I wouldn't even call Figma heavy by the way.

Uh anything in the Yeah, I think Figma sluggish dude. Yeah, but but it's essentially it's just a RAM eater Chrome. What is Figma? Figma is just the Chrome app. Mhm. Uh it's based on chromium and it's you know a bunch of like frames that you have. It's all eating your RAM. So I would say if you are a student and if you are a designer and if you are on a budget, don't go for the 13 in. Go for the 15 in and then go for something maybe with 16 gigabytes of RAM. And I wouldn't honestly go for 24 gigabytes. Yeah. Because I think this is the point of like diminishing return. Yeah. I mean these are in dollar.

in the Europe. So if I go to the um like German version of the site, these are like significantly more expensive. Uh I would I would honestly buy this as a student. But this is the Air, right? Yeah. Yes. The Air. But if you are if you want to be to do something more futurep proof and I have the M1 Pro and I bought this in 2022 and it's still it it's still like a a working horse. It does everything I want it to do and if it lags which does sometimes because I open 500 tabs and I have like literally three instance of Figma with each having like five plus tabs, right? So, I am like an extremely heavy user with three monitors.

When it lags, all it needs is a restart, nothing else. And it's fine. It's really fine. Uh, even my battery is dead. My battery is been saying that I should go for a service for over an year over since over a year ago. It's telling me to go and change it. I haven't done it because, you know, it's doing just fine. I have a thing about this. I have a thing about power. But anyway, yeah. So, if you want something powerful um and future proof, I mentioned my machine to say that it's future proof. I've been using it for over three years and it's just completely fine. I would go for one of these two and I wouldn't go honestly for the 14 in because the screen is too small for me.

16 in. I've been thinking about this as well just the last couple of days like if you travel, if you move around a lot, go for the 14 inch. Yeah. But if you if you are mostly working from home, you connect to a monitor, like go for this, definitely go for the 16 in. Um, yeah, you don't have to, but like I think you'll really see the benefit of going smaller if you travel a lot because everyone's everyone I know's backs with in like Yeah, this is heavy and this is coming by the way from a a lifelong PC user and I have next like I my setup is I even have a let let me actually like geek out a bit about this.

I have this controller. So this is a switch a switch for my PC and Mac. So I can with a press of a button change I all my peripherals from Mac to my PC and I use both of them and I've been a PC user for years and something that really changed the way I work was that this MacBook Pro the 16 like the the 16 part makes a big difference that I can unplug it from everything even from power even with this like relatively dead battery and I can reliably get work done. Not just do some email and like moving a few pixels here and there. No, like actual work because the trackpad is that good, the keyboard is that good, the multitasking is that good that I can reliably do this.

I could not like do that with a Windows laptop. I could not do like reliable actual work. Um, with some Windows laptops you could get. No, not really. And at that point when you are talking about that level, you're over like 3 4K um USD for a Windows laptop. So get a Mac if you don't already. I I think this is like the the best computer or like the best tech gadget that I have ever purchased this MacBook. Yeah, it's it's crazy. I don't know how Well, we're on M1's, right? I did the maths on, you know, yearly costs, you know, or even monthly costs even like four years, I think, is is a sensible amount of time.

I think we're at four years now or five years, right? Our max cost us something like a few hundred a year. You know, when you break it down like that, it's like holy crap. Like these are value quite literally machines. Yes. The fact if you think about paying I think it was something like £30 for me a month. If you think about you can get a computer for30 a month like a a really good computer for 30. It's just that's crazy. like it's almost like they should be worth a lot more but obviously no pay and the max just a small note on that is for somebody and this is very wrong but for somebody for who for some for whatever reason wants to play games on their Mac don't do that but if you if you want to buy a Mac it has to be a Mac and you want to sometime play some games on it then you need the max or if you do something with obviously with a lot of like 3D or a lot of video and or even video like I'm I'm not seeing a much benefit at all doing video.

I granted I'm I'm using like a and B- roll like I'm not using tons and tons of layers and effects and that so there there is that aspect but like yeah think of that how heavy you are editing but I would also add music production as well they're quite greedy with all that kind of stuff but like yeah our world not even not even um not even worth considering a max really unless you're a yeah you just won't you just will not see the benefit like you might think you're getting the benefit, but you just you just do not see it. You cannot see it. The thing I was Sorry, sorry. No, no, go, go.

I was going to say about the power thing, right? Yeah. So, are you connected to USBC? Yes. Uh, and I'm running like one of the monitors through the USBC. Yeah. Because you got the 16 inch, you probably What is your It's 167 watt power, right? 140 watt power. Yeah. Yeah, that's one. Yeah. like you're probably killing your battery running it through USBC because it's not able to provide enough power that it's consuming. So, because you can only you can only charge a maximum 100 watts through USBC. So, you might consider like I mean you're you're screwed now because your battery needs replacing or whatever, but like it the the Mag Safe because it can charge at the 440 watts.

Interesting point. Yeah. Yeah. you're like that. I I used to where is it? I've got remember remember dongle life, right? I used to I used to put my power um into a dongle. I just wanted this really clean setup where I could just have one cable. And so I had a dongle where my power my USBC power cable went into that and then the dongle went into my my computer. that dongle killed the, you know, it wasn't like a direct throughput and I think I was probably charging it at like 30 watts or something like this because it was just killing it. You could never have that M through. But yeah, thinking about like it trying to work and charge at the same time, whereas Macs are really good at like, you know, holding holding their charge and and not eating the battery or using the battery if you're plugged in.

So, but if your monitors and things like that are constantly drawing power, it's constantly having to charge it back up because it's eating away at the battery. Yeah. You know, but anyway. Yeah. Um Yeah. So, that's a good point. That's a good point. But it tell it is telling me that it's fully charged and the power source is power adap adapter. But, um you're making a good point. Uh can't we see like what is the wattage that the USB If you go if you go to the Apple sign at the top left and hold alt um and hold alt. Yeah, you should see one of them change. About should change to system information. Yeah. And then go to power.

Oh, I I I I used to come here from somewhere else. Yeah. Scroll right down. It should see uh it looks like 52. Which one is your lo up up up up up up up up up up up up up up up up up up upup up up up up up up up up up up up up up up up up up up up up up up the one right at the top that's indented. Yeah, see it says wattage 52. Interesting. Yeah. So mine says 140. Yours said 52. Whoa. Yeah. So 50. So on idle these Macs use like around 30. So I can very well imagine like going over 52. That means the battery has been helping.

Why are you only charging at 52 watts? Because I'm also running it through a dongle because of the what I just showed. You're doing You're doing the same thing that Yeah, I was doing. It's killing your It's killing your power. Yeah. Yeah. Now I see. Okay. Okay. Now I see. Your your computer is only charging at 52 watts. So So guys, if you if you do this now, you can also check this. Yeah. Okay. Good point. Yeah. Anyway, we nerded out for a moment there, didn't we? Yeah. I love this. I love this. Yeah. So, that that's cool. Let's um let's wind down with uh weird AI jobs. Um by by the way, this this is a discussion.

Actually, I just came across this this interesting video. Let me let me get on the screen. Um copy link. Make sure I'm muted. Yeah, I am muted. Good. Good. Could you imagine working alongside your CEO day in day out, watching their every move in order to clone in order to clone them with their digital twin? Or uh you are the next founding member vibe coder for a company that has yet to raise any funding. This is about weird AI jobs and how dystopian they're getting, basically. So, I came across this video. Where's it gone? You want to show it? Yeah. Oh, there we go. There we go. Came across this video by Joshua Fluke, and he's looking at this weird job application where a CEO literally just wants you to follow him round.

You need to be like, you need to be this expert at just agents, AI, all things AI. You just shadow him in in meetings. Um, and just basically, I mean, to be honest, I kind of low-key think this would be an amazing job, but he I think it's this dystopian idea of what he wants to do with this information. So, you'll build agents to respond to emails. This this this CEO just seemingly wants to completely remove himself from the equation of doing his job. But but is that not like the job of like a business person? Like as a business man, you want to um the the tasks that you just repeat, don't you want to like potentially give them to somebody who can do it just as good as you and maybe even better?

Yeah. And then move on. Well, of obviously he's identified the things that he does repetitively that he just doesn't need like Yeah, absolutely. Like you you systemize as much as possible. I just think that it's an interesting and by the way, I haven't said whether this is right or wrong. I think Loki, I would love this job just to like build systems and things like that to replace him. But I think he's going a bit I think this guy is going a little bit beyond um like the what's what how how what what do I mean like he just wants to disappear like he doesn't want he doesn't necessarily want to improve on anything. He basically just doesn't want to do his job anymore.

Um and there is there is room for a CEO. I do think there is room for a CEO. there's a room for a CEO to set the vision of the company, you know, to steer it in a direction. And if he's if he's basically replacing himself with all of that, you know, he he won't be clued into the company at all because I do think there's an element of like knowing how things get done in order to then improve them or or guide or or whatever. But um yeah, he he's just, you know, um this is the success metrics here. Five production agents live by day 30 30% reduction in VP calendar load time two external media features or conference demos by by month six.

Um and interestingly, you know, we we mentioned a couple of times in this show about like us being early and stuff like that. I I'm often surprised I'm often surprised at how some people who you wouldn't expect them to be like all in on AI or whatever. I don't know. I don't know. It makes me question whether we are actually early. Do you know what I mean? Yeah. Um so this like the my main issue with this is context like there is no AI no amount of like AI that can match even yes remotely the context of a human being uh in in this specific need like as a CEO as a as even like a even if you're like leading a small team you need to have such a huge context and also memory of the past and also you know what you want to do in the future.

The the context is just not there because with AI and automation you do these smaller things like you can reply to emails but having all that context and do it. I don't see any AI doing this but like the direction is very interesting. It is I yeah and I mean I want this with my team. I want so I have a project manager on on my team and I want her to be the second me. I want her to to reply to to clients the way that I would reply and potentially better, right? I want to replace myself. And I saw this also on a Mr. Beast video where Mr. Beast mentioned this that uh with a bunch of his friends that you know are in his videos.

He did this. He trained them to be second him. So they he he trusts that if somebody goes to them and they ask them a question, they answer the way that Mr. Beast would. And if they show him a piece of work that they've done, they would note on it the way Mr. beast himself would do because they spent such a significant amount of time with him that they would know what he would reply to a certain type of content and that is that shadowing exactly that the CEO is looking for right I think that you always thing is you're always going to account uh encounter new information or sorry not new information sorry new challenges right and again like identifying repeatable systems and processes and and whatever.

Like I answer the same email like every single day. Everyone asks me the same questions. Okay, build a process for it. Um I do the same thing every single week and it takes an hour of my time. Okay, build an automation for it. Something like this. I I totally get where where you can identify the same thing that you can tell someone by the way when someone asks you this say that that's fine but it will never it will never be able to think for you like it might be able to give you a best estimate but I don't think as a CEO I could ever trust that AI is ever going to do the right thing all of the time based on previous information.

like I just I just don't trust that it's ever going to be able to do that because again AI is derivative. It's based on previous information. That's where I think this falls over that they're never going to innovate. They're never going to progress. I hope that it's only doing this on repeatable systems. I hope that part of this application process is like identify things that I do over and over and over again because though I totally commend that, but it's where the innovation happens is where I'm um I'm hesitant to uh offset to a complete AI. I don't I I don't want a robot doing my job like Yeah. and and I often yeah encounter new things a lot or at least it's the same thing but with a new perspective you know or or a slight twist you know um but then there's this last one here which is a an expert level vibe coder basically which goes back to what we were saying earlier if we go to here they want four years software engineering experience so and and strong Python nodes typescript postgress MongoDB knowledge all this kind of stuff.

They're essentially are what what I interpret this as is that they want a developer who knows what they're doing, but they want the efficiencies of vibe coding, whatever that I don't know whether they understand vibe coding correctly, but they just want an efficient AI coder. um or so probably someone like me to be honest I probably be a good candidate for this because like I'm totally into coding but I'm I'm not afraid to use AI and I I sort of feel like I know where the efficiencies are gained and where the efficiencies are lost with AI coding. So we're seeing more and more of these. I hope that this is becomes a se series or something of just weird AI jobs.

But um yeah, it would be it's it' be interesting to keep note of. If you see anything, bring it to the table is what I'm saying. Yeah, we have a comment saying so how someone remember that someone else emailed him or text him if AI responds to that someone else. It's kind of the AI also need to Yeah, the AI should probably like give you a summary of what it had done at the end of the day. Like a lot of managers do this by the way like to to the upper management they they they give a summary of what is done so they know where is everyone at or it's just a case of like if someone asks then you step in.

Do you know what I mean? Or the AI is tuned to sort of flag when it's like I don't know. Like you got to tell an AI just because AI always wants to please you. you know, it's a very very eager gentleman wants to please you all the time. So, you got to tell it like just if you're not sure, flag it. Don't just and then maybe the CEO can step in or something like that. Um, it goes goes back to what you were saying last week. What were you saying last week? It was about plan mode in cursor and you were saying that you wanted you didn't want it to do the work and you wanted it to tell you step by step.

And you know, you don't need plan mode to do that. You just need to tell the AI what you want it to do for you. But for for whatever reason, we don't think to do that, right? If you don't want it to do something or if you do want it to do something, you've just got to say it, you know, rather than just assume that it's going to just mind readad, you know. So, are you saying that AI is making us better communicators? I think I think better communicators are going to be the standout prompt engineers. I know people hate that terminology, but I think better communicators that already they're, you know, they're so used to knowing where people fail and where they like, a you know, it goes back to the GBT hybrid thinking mode like tell it to think or tell it to back off.

Tell it to don't tell it to not think as hard and then it will go to GPT light, GP5 light, right? I I'm used to working with freelancers. I'm used to like kind of telling them like, don't spend too long on this. this is not this isn't going to make or break this project, you know, this little this this whatever it is you're doing. So, um yeah, hopefully people will get better at communicating because they're going to be frustrated that AI is not doing what they want to do. So, there is a level of micromanagement in it. So, you tell AI to you know how to spend the time and what to focus on and Yeah.

Yeah. And they will never complain about micromanagement. Yeah. They're happy to do it. Yeah, the the more general you are with the prompt, the the worse uh of response you get. We we know this since day one. So you have to be specific. And when we say be a better communicator in this case is seeing um the whole picture, reason for yourself and not do let the AI to do reasoning for you. And um and yeah, with that you can you can get something more specific out of it. Yeah, totally. Um anything else from your end? Uh no, but I'm trying to make another video. How where how's it where is it up to? Uh it's just taking time.

How long have we got how long do we need to store for? Uh, oh my. So, I I don't know. It's um Let me Let me see if I can upload it and then we play this new one. It's It's so racist. It's racist. What? It's so uh Yeah. Did you you didn't intend it to be racist though? No, it's it it gave me an accent. Uh even though it doesn't towards you. Yeah, it's racist towards me because it gave me an accent without me uh you know specifying anything. Just assumed my accent. Cheek of it. The absolute cheek of it. Well, whilst we're waiting, scan that QR code, give it that will link to our YouTube.

Let's get to 300. Let's get to 300 before the next week. That'll be amazing. So, are you ready for this? This is This is the No, let's let's watch the first one first because we have a bunch of like new people also. Let's watch the the first video I made and then we watch the second one. Okay. The future is here and it's being shaped by the incredible advancements in artificial intelligence. So I gave the same image back to Gemini and I said I was more specific about what to say. I gave it like what I wanted to say in quotes and I did it. But let's watch it. Be sure to follow us on Instagram and YouTube @ cmdai show and join us next week for all the news in AI design and dev.

Diabolical, right? Well, that's all we got time for. Make sure to follow us on Instagram and YouTube at Command AI Show. Join us next week for all the news in AI design and dev. I've been Sam and I've been Kavarza. Keep on. Peace out.