65·04.09.2026·01:49:34

#65 Webflow Conf, Fable 5.1, GPT-6 Astra, OpenClaw 2.0

This week, dive into the biggest news from the Webflow Conference, the groundbreaking GPT-6 Astra, and Anthropic’s Fable 5.1! We break down Webflow Source, new AI models, benchmarks, and why real-world usage trumps everything. Plus, our candid thoughts on Webflow’s future and the importance of open-source in the AI era. Don't miss our take on why pricing could make or break Webflow's new direction! — TIMESTAMPS: 00:00 Intro — Unlock the full potential of your online presence with Kabarza and Samuel—experts in web design and development (respectively), powered by cutting-edge AI solutions. We blend creative design with advanced tech to deliver smart, high-impact websites that stand out. Ready to elevate your business? Contact us today and see what AI-driven innovation can do for you! LINKS & RESOURCES: Website: https://cmdaishow.com Check out Kabarza's amazing work: https://kabarza.com Visit Samuel's website for more: https://samuelgregory.co.uk 📷 Follow on Instagram: https://www.instagram.com/cmdaishow — HASHTAGS: #ai #podcast #aidesign #aidevelopment #vibecoding #webdesign #webdevelopment #ainews #webnews #designnews #devnews

Transcript
And showtime. Hello there. Yes. Yes. Yes. That's what I was doing as well. I was just I was just having a little flick. What the hell? Oh, wow. Web flow. Web flow maxing. I'm web flow maxing. Do you like my Web Flow swag? And do you um do you like Web Flow per chance? Do you use Web Flow per chance? Sometimes you dabble dabble a little bit Web Flow. Yeah, I like Web Flow even like this. Even still after the conference after they even after they abandoned me. Yeah. Yeah. Wait, is it visible? Yeah, kind of. Like you see the Web Flow logo. I don't want to pour this on my laptop. That would be very unfortunate, wouldn't it? But that would mean I could get an M6 one. Okay. I'm just doing a quick sound check. That sounds good. Everything sounds good. Sounds good. Okay, let's Maybe I could come down a little bit. Wait, actually, there we go. I have to click on go live on Instagram. I haven't done Oh, yes. You've left him in the dirt. Hello, Instagram. Hello there. Look, look what I got for you. Just for you. Just for you. This is for all of you. Uh, so we are going to talk about web I guess today, right? You ready? Yeah, I'm well excited. Yeah, we have web flow and we have more. So, let let's actually get it cranking. We don't have too many different news, but we have a few big ones. Yeah, probably some of the biggest actually. Yeah, these are some of the biggest of the year definitely for our community. These are the biggest news of the year. Definitely. Sure. So let's with that being said actually let me separate my Instagram view so I can see your beautiful comments uh in one view and then I have my notion. All right. So let's let's get it cranking. So this week we have Astra 6 which we don't have access to yet but but they are making up for it. We'll talk about that. We have we just had Webflow comp. I'm just coming back from Webflow Con with some swag as you see. And we have tons of news. Web flow just killed itself because that's the right. I thought you were going to say killed. No, killed itself. So, but in a good way. In a good way. Uh, we have Open Claw 2.0. Have you heard of the the first Open Claw? That was a thing um a while ago. And we've been playing with um Claude Fable 5.1, which is which is pretty sweet. Yeah. So, that's about it uh for the news of this week. Astra I'm well I mentioned that that was the first thing that I mentioned yeah before web Astra yeah yeah that was we'll talk about all of this and this week's episode this is command AI I'm Karza and I am Sam and this is a week in design AI and dev I I completely forgot it and web flow seemingly and web flow in web flow we trust in web flow We trust. Yeah. In Web Flow we trust. In Web Flow we trust. Yes. So, uh let's let's actually start all of this. Uh do you want to start with Astrong? Well, no. Um I would say let's let's get open claw out the way because there's not a lot to talk about really to be honest. So, good. By accident, seemingly OpenClaw basically had a major release. It wasn't intentional. Um, it was a contribution from, let me find this out, Claude. Claude. 933 contributors, including 569 firsttime contributors. Open source for the win, I think. Yeah. So they released this blog post that basically I'll save you the spiel is that they planned a sort of minor you know update a visual refresh but actually over the course of two months um I don't know how two months turns into 230 days but they managed to kind of overhaul it uh beyond and sort of recognition almost. So, what they've uh what they've introduced is a visual overhaul. Actually, let me um thinking of which way around to do this. They've simplified the startup. I'm going to cancel that. Can you see that? Okay. Yeah. Um and I'm going to close this. I'm just going to run the start up once again just so you can see it. It's a bit cleaner. This was always one of those annoying things where it was just it was just awful. It was absolutely awful. The the owner, I would say, is the probably the worst person for designing UX in the history of the world, unfortunately. Smart guy, but yeah, I mean, even sort of this looks a bit, do you know what I mean? It's a bit Yeah. bit much. Um Oh, it went down a slightly different route. I um did not expect that. It looked a little bit different. Here we go. It looked like this before. So the another logo to add to the to the list of logos that they use. Um and yeah, it kind of it just installed everything which landed up on this which was let me refresh. Oh, that's um let me let me start this up. Claude open run the server. Oh, that's good, isn't it? Yeah, that's good. I've literally just installed it. Maybe I'll do this. Hang on because of course you need to know the um know the terminal to be able to run thing. Yeah. Node is required. Node blah blah blah is required. So what why why is it now suddenly not working? Literally just ran install and all of a sudden it's not working. And now I can't I don't even have um open core. It's not even turning green. So it's not even registering that I've got the app. Let's try a new window. Okay, let's just go. Let's just do what it says here. I don't know why why anyone should have to run this, but let's just X baby. I hope someone from Open Claw is watching this. The fact that all I did was run install. Oh my days. Here we go. Let's try this. Someone from Open CL is watching. It's literally just a bunch of butts. Well, put it this way. Oh, something's happening. Put it this way. I no longer use Open Claw. Uh, it was just Yeah, just absolute Oh, here we go. Right. Apparently, it was already started. Um, it was just becoming a mess. Absolute mess. Look at that alignment. Yeah, I know that that wasn't there earlier to be fair. Yeah, that's already [ __ ] misaligned. But what's quite interesting is that it automatically detects your subscriptions and things like that that you have installed. So, it's picked up the fact that I've got clawed code installed. So, it gives me all of this stuff. Obviously, I have to have my extra usage turned on because I don't like you using Hang on. Did I just see Mythos there? That's interesting. Can I use Can you use Mythos through the API? No, you're not an enterprise guy. Yeah, interesting. Anyway, uh but OpenAI codeex is a different story alto together. So, it's picked up the fact that I can use uh Open AI. Wonder if that gives me API usage of Astra. No, it doesn't. Um, and I can also have get, you know, access my claw code sessions in here as well and presumably my codec sessions. But, um, yeah, quite nice. Overall, it's still it is a lot nicer than what it was. It was a real pig's ear and the real sort of uh where it was obvious is in these settings, but it actually looks like a normal app now that you would use and it's like a lot cleaner. So yeah, big overhaul there. Um, what else did they do? I don't I'm not sure what else they actually did to be honest, but I think it's just things work a bit better. Shared cloud sessions so that you can bring people in if they are um if you want people to work together. So there's a multiplayer mode which is actually quite nice to be honest. I I'll give them that. Uh this is the automatic subscriptions and this is why they took two months. So regardless, welcome update, cleaner, slightly less sloppy, only slightly open claw 2.0. Look at that image. I honestly that image is giving Oh my god, it's it's awful. Why are they not even using like a good image generation model? This this looks like something from two years ago. Yeah. Well, I wouldn't say two years ago, maybe a year ago. This is classic GPT. Classic GPT imagery. Like I say, not the the the GPT the V2. Yeah, exactly. Right. Too bad. Yeah. Yeah. For that. Uh well, like I say, they they Yeah. They think they care about design, but then they go ahead and release things like this. But there we go. Anyway, yeah, GBT uh Open Claw 2.0. Download it now, I guess. I yeah I I don't know why are they not going the what is it called the the one the Twitter got the XAI got what what was it the Grockbot I forgot the name for a second yeah I'm not sure why didn't they go that direction because it seems like people are receiving it really Well, well, I really struggle with Grockbot and I still kind of do to be honest and I met my friend yesterday and he took me through what he was doing with Grockbot. And I will give Grockbot the benefit of the doubt in saying it's really surfaced what most people want to do with these sorts of agents, right? And it's on the cloud. It's all everything's always on the cloud. So, you could shut your laptop. You don't even have to have your laptop on and you know these agents are going to be running in the background. Um I showed him how you can do it in Claude. I showed you how you can do all that in Claude and I showed him my setup and stuff like that. Granted, I had the nerdy setup. I had the, you know, it's not as accessible the way that I do it. It works for me and I like that and I want it to work that way. n being a pivotal part and determinism is a big factor as to why I don't like Grockbot. Having said that, yeah, you hit the nail on the head in saying that this accessibility and just bringing a really easy way to just get stuff done like Dropbox does. Super simple, just does one thing really well. Yeah, they should have gone down that route for sure. Well, cool. Let's jump to the next one. So, interjection, there was another model released. This was actually really big week. Yeah, Gemini Flash 3.8 and 3.8 Flash Cyber. This got released earlier on in the week and you know, we're getting tired of benchmarks, man. But this was apparently this was scoring really well on benches. um real world usage is a different matter whatsoever but 3.8 8 flash appar people are people are enjoying I did see a few pe few tweets on people really enjoying using this thing I mean apparently here according to deep sui benchmark which is by the way maybe one of the most reliable benchmarks when it comes to software engineering it's benching higher than opus 5 or at least on like media that's insane but I don't know if I can believe that like in actual like usability That's the thing, right? Yeah. They could be benchmarks in for all we know, but you know, a fraction of the cost. We're going to go through a bunch of new model releases this week, and I think the story will always remain the same. Ultimately, just use the damn thing. Just it's a new model. Just go ahead and use it. Don't don't jump to the benchmarks and get excited about them. Just just actually use them. Or or similarly, don't jump to the benchmarks and be put off by them. Just use the damn models yourself. Yeah. Um, what have we got here? This is um Terra. Yeah. So, 61.4 versus Opus 58.6. So, and what's that one on? That's Finance Bench to be fair. So, yeah, another new model, but without that interjection, the big boy. Fable 5.1 and Mythos 5.1. So before anyone had a chance to catch up to what Anthropic are doing with their Fable model or Mythos models, whatever you want to call them, they then release a an update 5.1 as an update, extending that lead, which is absolutely incredible. Obviously, Astra's just been released. We'll get to that. But the point is they released this before anyone had a chance to pick it up. So few notable things actually just to say this is they are the same model. However, Fable has uh more guard rails around it. So it can't do certain things and only certain level of people have been trusted with the great mythos who are particularly sensitive or have some sort of relation to do with cyber security or life sciences as well. The most interesting thing is other than it being obviously amazing is that the price is estimated to be 25% less than Fable 5. And we'll get into some examples where I have seen this and it's pretty darn good. I also had this conversation yesterday as well. It's a bit boring, but the the they've introduced this EFS uh customer retention policy. Now, a lot of companies were not using Fable, were not allowed to use Fable because I don't know the specifics, but it comes down to how they're allowed to store data. If you use Fable, obviously, they want to improve. It's because they have to No, not just that. It's because they have this strict policy around uh security. So, so anthropic has to see in a way to see like I'm doing air quotes. They have to see what prompts you are sending. They are constantly observing that. So if it's uh deemed unsecure, they route you to Opus instead of letting you use continue using Fable. That's why that's not acceptable to some of these enterprise companies because they don't want their data, their prompts to be seen by anthropic in any way, shape or form. So from my understanding they are doing this new system where they allow I think an intermediate um solution here that they they don't let the company host fable uh but there is this intermediate that the company knows the their data doesn't go to anthropic so they can exactly yeah I understand so the very safeguards that are in place to stop you to to make it not mythos basically. Yeah. That whole system is data being shared to a third not third party you know what I mean. So I understand I understand. So typic so if mythos was available then they would allow it because there doesn't they don't those safeguards don't exist and they don't need to send that data off. Yeah. Point is is that uh organizations had to use opus. they were using opus for this for what you've just described. They've introduced a new uh well as I say EFS a new data retention policy which hopefully is a lot more I think they still have those safeguards and that mechanism but the data being shared is slightly different. So really interesting. The guy I spoke to yesterday, he used to work at Meta. Used to work at Meta. He has an opportunity. He didn't get an opportunity to see what life would be like inside Meta and and big organizations that were impacted by the Fable 5 lock or sorry the the Fable 5 data retention stuff. So yeah, uh, who knows what we'll see from a software perspective now that these larger organizations can use a shitheot model or at least if they they choose they they decide that these EFS rules are better for them that they can now use Fable. We might we might see a good uh, you know, better maybe maybe Open Claw will be better. Maybe they'll actually make some serious improvements there. Yeah. Um, but yeah, as you come to expect, Terminal Bench, you know, already the low version is scoring probably on par. What's that? 26.3 25. So, already higher than uh yeah, which is the highest actually. So, it drops there, which is kind of kind of weird. And for cost as well, look at the cost is kind of shifted a little bit as well. So, all of the good stuff. I did point out here that I don't really know what they're smoking, but apparently Fable 5 is worse at aentic coding than Opus 5. Yeah, this is this is where it's weird sometimes to look at these um numbers. That's why like I I would like to just test it test these things and see how they do. I've been testing Fable 5.1. It's it's I haven't noticed like a big jump, but generally speaking, I stopped using the high version. Like I've been always like on medium almost like always. Yeah. Because it's just good enough. I noticed on the high it also takes a huge like a long time to do anything. So just medium it does the job. I'm really happy with it. Generally speaking nowadays I'm not sure what we'll get into it. We'll get into Astra but I don't know what these new things can do. Yeah. Sure. Yeah. Yeah. They are really good. Let me just finish this and then we'll get into our like observations on using Fable and all the rest of it. So you might not know this as well is that they've actually been embedding watermarks in compliance with the EU AI act. They're watermarking the output and then giving you an API that you can send output text to to detect whether it's generated by AI. So really interesting. I'm guessing it it it's in the word choices that actually have some sort of cryptographics signature or something like that. So that's quite interesting. Um and that's it. That's all I wanted to finish on to be honest. So yeah, all all the all the improvements you expect extending an already pretty high lead on the on the world's best model. Um yeah, so you were saying then that you've dropped it down to medium because you prefer the speed over the Yeah, it depends what you're doing, isn't it really? I think us in our position, I was thinking about this yesterday. us in our position doing what we do. I think we we absolutely should be using Fable and Opus at the very least. Like I think we undermine ourselves a little bit when we talk when we think about the work that we do. We're doing very complex work. You know, maybe we're not fable um x high data science deep research stuff, but the stuff we're doing are is quite complex. I know it might not feel like it is, but it probably is. So, I'm very hesitant and I probably should be using Fable a lot more for the very reason like you just said and just dropping it down to low, medium, working in Fable, but doing low, medium or high, you know. Yeah. I like using Fable. Yeah. Yeah. Yeah. I I I think I think the work that we do um I I can divide it into like two ways. One way is like we are doing this iterative type of work. So whether you are creating a web app or a website like I'm like I'm doing a lot of like marketing stuff with it. Um I I would I don't want to give it a one big task and go about my day and come back one hour later and get a huge software or like a huge website that I can't myself even navigate and then I notice like a big chunk of it is not the direction I want. Instead, I want to do this iterative like one by one feature like animation or whatever. Like I want to add these stop these things, test them and I'm not just testing like if they work or not, if they are buggy or not, but test them if they make sense in terms of like UX and getting me what I want and then just build on top of that. And then we have that other type of work where you can like give these models like on X high even the whole thing and be like improve on um security or like find patterns where we can reuse like some of the the patterns to um you know reduce complexity and things like this. This is I correct me if I'm wrong, but this is where I feel like you can use the X high version, but for smaller things, I I I think just low would be even fine. If you want to make a change to something that you already have and you know what you want, that's the point. If you know what you want and you can describe it, just use the low. M but saying all that and we we've sort of contradicted ourselves a bit because your your friends decided to use GLM 5.6 flash right and getting very good results from it. Yeah, I think 5.3. Yeah. Although what we do is complex, some of these cheaper models are catching up, right? Um absolutely. Yeah. It's kind of again I guess you got to kind of use it. I mean only people who really understand the code that's being written are going to know whether it's going to be a good model or something like that. And uh yeah, to Mark's point here, like probably probably should just be playing around with more different levels of, you know, working from the bottom upwards and just just if it gets the job done on fable low or opus low even, you know, I tried that, but it's sometimes I I see myself losing time because I'm explaining to a dumb model something that Fable would just, you know, understand. Yeah. Yeah. So, arguably, Mark, you you might waste more tokens because you're getting a dumber model to do something it's not really good for. So, it's a mixture of the two really. It's like, yeah, you know, again, just understanding the task at hand and and and choosing the right model that's capable of doing it, but not going so low just for the costsaving impact because you'll just waste tokens, you know, doing all that. Yeah, I couldn't get Haiku pushed to my GitHub easily. It was like always coming back and bothering me. And I was like, yeah, I was like, you can't it could do it, right? But it was it had this tendency to ask me things. I'm like, I don't want you to ask me things. Just just do it. I'm I'm telling you to push. It means push. Like, why are you asking me all these weird questions? Well, fable. It just does it, you know. Um, yeah. So, um, I my experience with I want to show you something that's on this page in a sec because I've just realized it's quite cool. Uh, my experience was so I've just released a video where I learn how to get AI to edit these videos. If you don't know anything about the show, we do this recording live and then next week you'll start to see some of the clips go out. Not even clips, it's wrong to call them clips, but they are they're the individual articles. Independent videos. Independ Yeah. And trimmed out, less farting around and whatever, just succinct stuff. Uh, but because of time and all the rest of it, I I've decided to focus on some of the heavy hitters, which you know, spoiler alert, I think the Web Flow one will be going out tonight when we after we talk about it. Uh, why why wouldn't I? We should be putting out more content. So, why don't I get AI just to edit them all and then we just put out the AI edited versions and then maybe I go in and edit like properly the big ones, the ones that I really want and care about. long boring story. The point is is I recorded this video and I actually tested GLM 5.6 Flash, Quen 3.8 Max, a local model, and Fable 5. And then I I edited it together and all the rest of it, and then Fable 5.1 lands, and I'm like, ah god. So then I redo the video with Fable 5.1. And a couple of things to note is my whole perception on editing with AI changed because of Fable 5.1. It was like it went from like h it's not really there yet. It's not really you know whatever it doesn't pick up on the nuance to oh [ __ ] Fable 5 has absolutely picked up on the nuance and that my whole uh what do they call it? Not thesis but my whole um your world my world changed. Yeah. Yeah. Yeah. Not quite as extreme as that, but yeah, basically, you know, it my my my outcome was different because of Fable 5.1. I've also looked at the cost. It cost $15 to edit one video with Fable 5.1. It cost $23 to edit. Granted, it edited three videos, but I don't think that is it's it whether it's editing three videos or one video or 10 videos, I don't think it scales that much. I think if I I think if I rendered 10 videos with Fable 5.1, I think it would be between 15 and $20. Like it it's not it's not like it doesn't exponentially get not even exponentially, but it doesn't just it's not just that one video cost 15, therefore two videos will cost 30. No, I think it's just a few more tokens that it uses. So, yeah, I'm noticing the cash increase that or sorry, it's the it's cheaper when you hit a cached token. Basically, if it's already that I mean, it's going to cash your whole conversation. So, when you write write a question, answer, you follow up, it will all of that that chunk in the beginning will be cached and that will be priced differently to the new response, you know. So, you're going to save money on that. So, I've noticed all of these things. So, it's um yeah, it's it's it's a cool model. And for editing, I'm I'd be interested to actually just get Fable to do the edit completely all of the videos from this week and see how they see how they end up, you know. Um final thing on uh Fabler. Uh just noticing here it says made with Fable 5.1. And I'm like, well, I thought that was a video. I actually thought it was an image because I hadn't stayed on this area long enough. But yeah, we can change the time of day. Oh, that's cool. And I was like, well, how are they doing this? What is this? It's a canvas. It's a canvas. Yeah. Yeah. Canvas, baby. It's a canvas. So, this is 3D. Well, I say 3D. It's WebGL, right? So, it's pretty cool. Yeah. Yeah. That's really cool. The motion is really nice. Just just subtle. The blur as well. I wonder if they're actually rendering those out and then putting an effect on top of it. I don't know. Um or that whether they are loading images or videos into the canvas. I don't know. But true that but that's what I was paying attention to because the the motion of the bird is way too way too realistic. You know, I would say it locks in on that branch and the branch doesn't really doesn't do that. So that's true. Maybe. So, watch this. And that even looks different now. It's different every time. Look. Look. Watch. Oh, no. It's not. Okay. That's crazy. Yeah. So, it's maybe it flips between two. That's crazy. So, so we know it's not a video. So, we know it's not a video. It is. I want that birdie to land so you can see what I mean. I'm sure you saw it anyway, but yeah. Here we go. This this one. Did you know this is like I'm a sucker for this type of things. Um like my my new website is canvas everywhere as much as I can. Yeah, this is really cool. It's just pleasant, isn't it? Land bird. Whatever you can go check it out, I guess. But we are bird watching right now. So this is a bird watching stream now. Uh let's move on to Astra anyway. So yeah, obviously obviously when Fable 5.1 released, Astra was not maybe it was rumored, I don't know, but either way, open the party. They were hinting at it. So OpenAI have joined the party with their, you know, cybercritical god mode and they've just dropped GBT6. Astra, that's it. Astra. Uh, quite a nice website. You can, uh, play around with this. Let me show you something cool. Go on then. Go for it. Look at that. I wanted to to do the same thing like for the opening of the video. Go on then. So, yeah. So, this is GPT6 Astra with this nice, beautiful website. It's all 3D. And this is the situation. Astra waiting to get added to the model selector dropdown. Yeah, it's it's a huge one. It's apparently very very capable, but we don't have access in a kind of like not too bad way. So for every day that we don't get access to Astra, you get your GPT subscription limit reseted. You get, you know, you have that reset button that you can click. You get that every single day. So you're we are l we are using Astra but kind of I'm not too mad because I'm happy with what I have. So yeah, uh what can this model do uh beside creating this beautiful website? Well, that's kind of like a part of the whole cell. It can do really well in computer use and 3D. And the things that people have been doing with it are like insane. We'll get into these in a second, but I just want to show you this video by this guy by uh Arena AI. He's creating all of these like insane simulations all in 3D with a few prompts and there are like a bunch a bunch of these like he's creating a bunch of like historical things and I know people are in two camps when they see this like on one side it looks ass right it's like I mean these graphics they they are not amazing but considering that these were done with one or just a few prompts and thinking like this can be an educational thing like maybe you are a teacher and you want to teach something from a unique perspective. This is amazing like the level of 3D work and computer use that you can um you can do with Astra. So like this one you see the the change of the weather. So apparently it's just very very good at 3D. This was what also they mentioned in their release, right? If I'm not mistaken, like I'm changing to your screen. This is their release video. Yeah. Really, really nice. Actually, I can't Oh, yeah, we can. So, it starts off with this is apparently 7 1979. This is early days of uh AI, I guess, maybe. I'm not too sure. But then we get to 26 and everyone's using his voice. Is it Microsoft Paint? Yeah, it is. It is. So, he's drawing a rocket. So, he starts off with a yellow circle and then he makes makes it draw a rocket then in Microsoft Paint and then he asks it to make that rocket in 3D in Blender. She's um making a presentation. All of them are talking, which we, you know, we'll get into, but uh they're all doing their work through talk, speaking. They're not talking to anyone. And then he asks it to make it into a game. He's doing some law documents. Um there's a presentation coming along swimmingly. She's selling something on eBay. And then this is his game that he gets to make all Yeah. So we need to buy a projector now. So, we need to buy a projector and rent a safe space where you can talk, say whatever you want to your, you know, AI agent. Yeah. So, yeah, I see a crazy number right there. Arc AGI 99.9%. What the hell? That's crazy, isn't it? So, is this like human level intelligence now? I guess it is. Well, last I saw RKGI was on number two. So, this is news to me. Yeah. By the way, this this website is nice. It's nice. Yeah, it's so good. It's so good. One prompt. No, I'm kidding. Uh, but maybe Yeah. Um, so how does it compare to Fable? Well, Opus 5, they're comparing it to there. So obviously they're favorable in their benchmarks, but this is this is Fable. So Fable on max is not as good as Astra on low at that's crazy about a third of the cost in terminal bench and it's much cheaper in terminal bench. And if you remember when we just looked at Fable 5.1 just then they were very chuffed with their terminal bench. Um mind you this is science. So terminal bench is the company their what do you even call it their uh their benchmarking is in here is science 01 and then the other one you know I'm very confused. Um, ArcGI, they don't compare it to Fable Frontier Math, they're very chuffed about. They are comparing it to Fable five, not Fable 5 uh.1. There's Fable 5.1, which is again just scoring basically the same on uh for for a fraction of the cost. But it's Frontier Matt tier 4, and it's it says like is it like 100%. It's not quite 100%, right? I mean, it's Yeah, it's flatlining at 97.6 by the looks of things. Yeah. Interesting. It's struggling. It's struggling there. So, why do we know why we don't have access yet? I think it's probably no. I The answer to your question is no. I don't know why we don't have access. I think it's just a capacity thing. Well, they are giving away a lot of capacity though with the resets. Don't know then. Don't know. Well, is it capacity or is it compute? Is it compute that they're reserving rather than capacity? I don't know whether those two are are the same thing. Um, but anyway, I mean, this is a great model. Of course, it's a great model. It should be a great model. The only thing I'm confused about is that it's GBT 6, whereas uh last we heard GBT was on 5.6 six on all of the Terra Soul and Luna. So, are we going to get a GBT6, Terra, Soul, and Luna, or have they just skipped it all together? It's they've kind of gone the uh what other company does this sort of thing where it just Well, I guess no, it's very on brand for Open AI to just completely mess up all of their numbering systems and all the rest of it. But the answer will always be, I think, at this point is just use the damn thing yourself. Don't get caught up on the benchmarks. cuz I'm just looking to see if I had highlighted anything else. Um, but yeah, I think I am particularly interested in the computer use because and I think there was a note here on the codeex harness they've made uh let's search codeex updating the codeex harness to be signific to significantly improve the speed of computer use. So, I already think Codeex was really good at computer use because it didn't it did it differently to what Claude did. I think Claude literally took screenshots and use the mouse whereas I think this is more inclined to do it programmatically. So, you don't you can still work whilst computer use is working. So, codeex was already very good. Now, Astra is in intellectually a lot better at computer use. Plus, the codeex harness has now made it even better. So, this is really interesting. This is really interesting for me cuz that's the one thing that I really despise Claude for is its lack of ability to use my computer effectively, especially when I'm remote. Like, if I try and uh try and get something to do, like if I dispatch something, for instance, there's just always problems. There's always problems. Drives me nuts, mate. Drives me nuts. That's kind of it. I mean, yeah, this was quite a fun game. It's really good at 3D and computer usage. Uh, apparently it's not as good as um now. Uh, is it as good in in coding as good as Fable? Let's have a little look. Uh let's just search codeing some responses here. Uh fable is square. So no, I mean on max on max. So you're there's high. So it's actually on low it's as good as fable with coding. Interesting. So it's Yeah. Very close. Interesting. Yeah. Yeah. So yeah, just and I see 5.6 down there. And I I've been just having so much fun with 5.6. There we go. Um on I don't use it on max, but I do use it on high or X high. Yeah, X high is what I use it on mostly. M again like I think someone who understands the code would be able to like it will get the job done but has it done it in the most efficient way and realistically are you doing challenge challenging tasks where it touches loads of files and there's API calls and by the sounds of things you're you're focused very much on the front end which soul is I've been doing a lot of web yeah webgl and uh 3JS type of things yeah yeah Yeah. So, yeah, pick your poison, boys and girls. You've got a plethora of models to use. Don't be afraid to try some of the Chinese ones. You might surprise yourself. But if you're doing something a bit more, you know, uh uh let's say intricate or client sensitive or whatever, you might choose to jump into the Astra camp or the Fable 5.1 camp. I'm fully baked in onto the Claudia ecosystem. They know exactly what they're doing there. uh it would be very difficult for me to separate myself and everything I've got going on in Clawude to to suddenly start using a different harness or whatever or different model. But that all being said, uh I don't know what I'm where I was going with that. That all being said, I don't know. Yeah. Yeah. Cool. Let's see. We will have access uh I guess soon and we'll put it into test. Yeah. So, now is the time. Is it the time already? It's that time. You got your hat on. You've got your lanyard. You've got your little 3D printed mascots. Yeah. I will get my stuff uh ready for the screen share. How can I engage our dear viewers? What are you using as are you GPT a camp GPT or camp claude? Where are you? Are you experimenting with local models or who out there is using let's call them Chinese models open source models or whatever and actually doing real work you know client work with them be interesting to know and also any what work do you do with your models because I'm just a coder I do a lot of business stuff I do a lot of business chat with my models my claude Um but yeah so we have uh we have comments here saying only for I don't mean for what number four yeah only for user in live ah well uh people are uh join us on on YouTube no it's it was just like people on on Instagram so join us on YouTube and Twitter yeah And also we have another comment saying team Claude. Yeah, I'm mostly also team Claude, but I mean I'm burning tokens wherever is giving me the most amount of them while we're in the golden age with cheap AI while while they last. We got to burn them. Cool. Um All right. Uh should we should we start? Yes. All right. So, Webflow Conf just happened and by the looks of it, you see I was in a watch party and we got a lot of announcement and we are going to go through everything and tell you things that you probably missed out from the live or like from other videos. So, uh two big things happened. Web flow is reinventing itself by killing itself in a way. So the old web flow with the designer that we know that got some updates but the biggest most exciting update is what they are doing next which is a new software a desktop app called source which is completely different from the web flow you know it it's a different category of tools altogether. So let's actually get to it. So this is source by web flow and let's let's see what it has to offer. Okay. So this is how it generally like looks like at the start. You can open a web flow site. So a website that you already have on the I'm calling it old web flow or web flow designer. You can open that. You can start a new project from scratch using AI obviously or you can connect your own uh codebase. In the example that they gave, they actually showed us um a a completely like customcoded uh Nex.js app that they brought in. And actually maybe the best would be to show you that video. So I have the most interesting part of the whole um conference here. Let's let's just go through a few basically it's just a minute or two of this video. So this is source they are calling it they it has a new name source by web flow and here it comes in a second. They start by entering by uh loading a nexjs project into it all locally and then they ask the AI to basically run they will run the server. They they open up the project inside of source and here you already see a bunch of things. I I'll go through them one by one but you have your brand, you have your components. So basically these are the pieces of your project that you see but the most interesting part something that gave us a wow and I don't know how much of that is real we'll get our hands uh on it and test it but they ask it to create a variation a more dynamic variation of the hero section and it does in a canvas and then you choose the picker and then you have these UIs that are being generated on the fly by the AI to tweak the AI generated uh UI and effects whatever you you have. But this is not like web flow. Where is the you know old web flow designer? Here it is. So you can actually interact with the elements on the canvas. You can directly change the text or change the styling or dive deep into the code and literally code because this is a project that you can load from any GitHub that you have or push back you know your updates to the main or deploy to web flow cloud. So this is kind of like in the first glance the whole gist of what it can do. I have a few other things that I want to show you and basically would you go through them. So here we see kind of like how it looks the UI it looks like cursor maybe maybe I called that before it happened. So it looks like uh cursor in a way you have agents and you have these things that they call views and we will get to them. So here we have also you see your sessions. Um you have other tools in it as well. This one it's called assets and they show assets I guess in both web flows. Now we have two we have source and we have the old web flow. They show it there as well. So here this is like the the Figma tool vaver what is it called? Vaver uh or something like that. you have this nodebased system where you can generate assets like this is a part of web flow offering I guess now if you remember we covered this a while ago web flow actually uh acquired a startup through which they now do this uh or like with the help of that startup they have this new I'm not sure is it completely new or like this interesting take on agents where as you see here you have your projects you have your views you have like campaigns and projects but you also have agents so what's an agent so here they have like these different personas in a way so you have your context agent you have your channel assets you have uh QA agent I guess you have life cycle so you have like these different types of agents with different levels of access and they might connect to Slack, they might connect to Google Docs and these connect to your website and basically can manage your website. So, uh that was kind of like very interesting, but I um I want to show you something even cooler. I guess where was it? I guess I just clicked it. Um here. So, what about the CMS? Maybe this is the most interesting part of all is the nextG CMS. And when I say nextg, it does look like a proper modern like completely revamped, completely headless nextgen CMS where you have here your content, you have your pages and here for example you have the blog but they also quickly showed the homepage. You have these concepts of blocks and basically you're feeding your pages with the data from the CMS and I'm not just talking about like collection pages that's like collection pages and collection items the way they existed in the old web flow. No, like in a modern way you want to have like these different blocks the content of which is coming from the CMS and you can basically create any structure that you want with any sort of like templatization and that's seems like to be exactly that they they also show that you can preview this and that was the basically the highlight of this for me. You obviously have things like activity and you see like a human doing something or an agent doing something. This is really helpful especially for big clients. They want to have that level of uh control. Then they have this inbox thing that I want to show you now. Uh it's I'm trying to wrap my head around it. The inbox thing is kind of like the sessions that you have you know you do things in claude and cursor and chat GPT and you have your sessions and this to me it looks like that but it's kind of like different goals and tasks that humans or agents take on and you basically see okay uh this person is working on the site and this is an agent and this is I don't know like they have different icons I guess like the site has different things to work on for example analytics. So this is working on organic traffic or whatever. So you have these inbox and you can go through a team of people and agents in a way. Uh they they showed a lot of things regarding QAing and AB testing and things like that. So I'm I'm not going to go through all of that. But also you have now this granle control also on the canvas directly where you can click on any element and change the style or like these basic things that tools like lovable and riplet also have. So they are adding that as well. So you don't necessarily need the CSS style panel. You can do things also directly on the canvas. You can access the layers the same way you are doing it in web flow but I guess in a more professional but still yet friendly way. You see the tags directly in web flow you can't do this exactly and obviously yeah the the style panel a bit modernized I would say then they showed us views and they talked about views a lot. So views are essentially pieces of um software in a way or like a combination of different tools like you can combine the data coming from Google Analytics with some what whatever else that you have and you create a view that's visualizing the work that you need to do or the information that you need to digest and that's about your site and running it and the campaigns that you have and so much more and it looks really cool. It it looks like also they they mentioned you can get you can make these yourself install what web flow already gives you or there will be like a community where people can create these views and you can get views these views from different people. Here we have a bunch of like other product shots. So obviously as as you've seen so far I don't have access to it yet but um you bet once we have access I will uh I I will show you more and everything. So let's just have a look at here the um the page. So now it's called webflow.com/source. We look at the landing page and you see it's for marketers, agencies, designers, developers, agents, and it has basically the things that we talked about. I'm just going through everything to make sure that we yeah, we we talked about pretty much everything here. So, let's have a look at um kind of like the summary here. So, there are like the thing with assets and the agents. Oh, they they also mentioned that the agents you will see them working as they will work. Um I'll just go through this actually to see yeah they have a bunch of other things that I want to talk about separately. All right. So that was basically source. This is the future of web flow. This is where they are going. And to me it looks like this is how they want to transition. They never said that they are killing the old web flow. But I have a really hard time imagining the product actually being able to to survive this AI revolution or whatever you want to call it. Like I don't want to build in this way anymore. I don't want to deal with the CMS limitations in this way. It was good for 2019. It's not good for today. You know what's good for today? Astro. Speaking of which, I want to talk about like the other part of the stuff that they announced. Um, you we we were talking you you've heard of us uh talking about uh Astro in the past weeks and I started using Astro and Timothy Ricks as well. So I guess the biggest news was Timothji Ricks actually joining Webflow after it was also interesting he joined after he started to talk about Astro and building a tool for Astro and like a visual builder copying web flow essentially and now he's a part of the web flow team I guess working on some improvements from my understanding to the old web flow to the current web flow designer which got a bunch of updates and now let's talk about them. So visual editing for AI code components. So in web flow you could always do this like you could build code components with react. Now there will be a visual editor for like you can click on them on the canvas so you don't have to use the property panel panel necessarily. Um I want to go through actually them. Uh yeah, here you see you have this nice UI to let's see if I can zoom in. Yeah. Uh to edit those. Then here we have the components inside of the CMS and not just the CMS inside of the rich text. This is a big one. This is a super big one. So, previously in the rich text, you could have just a set of elements that they would give you access to. And if you wanted to add like a more complex component, let's say a slider, you could not do that unless like you would do this with custom code. We've done this plenty of times, but it's not the most uh SEO friendly way of doing it. But now you can literally in the rich text area type slashcomponent and you can choose from the components you have. There is a catch though. You cannot use components that have CMS collections or components with um slots. So you can use just in a way simple components but it's not very simple because with with components we have conditional visibilities and we have var variance component variants. So basically your rich text can become pretty much whatever you want. It's still not that nextgen CMS that you can get with something like payload, but it's it's much better than, you know, the the the old CMS with all the limitations that it had. So, I'm actually maybe about from all the new updates, maybe to the web flow to the web flow designer, this is probably the most interesting one or the most useful one to me because this was very limiting. uh not having it. Now we have this view with all of the break points next to one another and it just it it makes it so much more visible how useless the mobile landscape is because it's just you know small tablet sizes web flow just let's get well I I guess the platform will be depreciate in I don't know in a few years I guess. So yeah anyway it it's a nice update. Uh, speaking of not so nice updates, this one I'm I'm trying to wrap my head around it. It's called popovers. And you know, we had these ba basically elements before like with a drop down where you had kind of like a popup way. This is to me that, but it's called a popover. And I guess this is maybe the marketing language they are using. Anyway, it it's called popover and it's a new element, but I don't get fully like I why why is this a thing? Because we have the MCP and we have the custom element. You can create anything actually with the custom element. But yeah, um we have another nice update with to the components with slots. You can now add a default component inside of a slot and you can actually prevent some components from being used in in the slot. So this is like adding some restrictions which is really nice guard rails uh for your clients. Uh they added agentic translation. I honestly didn't know that this didn't exist. I'm not using it all that much and I think yeah for people who use uh localization I guess it's nice. Um they talked about style props for components when Timothy Ricks was on the stage and I guess that's what he will be work I'm just guessing this is what he will be working on. We'll see about that. They also said that uh the the whole performance has increased by 80%. for example, the publishing is 80% uh faster. Site rendering 43% uh faster. So yeah, but I haven't seen any of it. I honestly I've been using it, but I haven't noticed. So I don't know, maybe I don't have a big enough uh website there. So yeah. Um that's kind of like all the updates, but that raises a question. If this is the updates, what is that they didn't talk about? Because they talked about source, they talked about the new web flow updates. What they didn't talk about is like they didn't talk about things like made in Webflow, which you see in the navigation. It's right here on my screen. This was this used to be a big thing because it is the the Web Flow community. people sharing resources and bills with one another. And it's to me it's quite sad. You know, last week Timothy Ricks, he closed down his um community. He has since joined Webflow. Um by the way, Christian, he he actually from the community he he predicted this he predicted this uh clearly. But anyway, so made in web flow components and sharing and things like with the community. That's that that's something of the past. They didn't talk about it. I don't think they they care about this anymore. I guess with AI that doesn't anymore doesn't matter anymore. There are a bunch of component libraries, but I myself I don't use um many of them actually almost any of them because you know you just do them do these things with AI. the community didn't got mentioned actually at all. Uh speaking of like things like uh the what do we call it now the experts the expert partners they didn't really talk about expert partners at all they didn't talk about web flow experts at all they didn't mention templates at all and I understand I'm not like saying oh they should have mentioned templates no but templates used to be such a big part of Web Flow Con in the past few years, they would mention like people making a lot of money with these. They would kind of like incentivize you to also make templates. And I see in framer uh in the framer world, this is still a thing. Templates and components and that part of the community. It's just apparently not an important thing in Webflow anymore. And I understand like these templates no matter how beautiful they are. I I think it is a shrinking uh market, a shrinking uh world. Like you get a beautiful template, you still have to make it your own. But you know, why not start from scratch with AI and make something of of your own? So yeah, that's that's kind of been it. You've been quiet. I know. I went on a rant essentially like I I had these things prepared. I You saw like that. Um but yeah, that's kind of been it. I actually I missed something. I missed something. They also announced a few things that didn't make sense and I think you didn't see these. So these are probably new to you as well, Sam. This is what I talked about. They have assets studio. Look at this. What do you see? These are a bunch of images. All of these, they have dimensions. These are images. They have templates, fonts, boards, and I don't know what these are. You have active jobs. I don't know what these are either. I guess these are running campaigns. And we we'll just go through them. So, this is truly like not like the web flow we know. it is just so out of place for so many of us community members. But now you can generate images with web flow uh either like from scratch I guess or like a mix of AI and the assets that you have and similar to how you have this with some other AI tools now starting to offer this. You can actually click into the text and change the text. So it's not the whole thing only like AI generated you AI generates it but you can edit it uh manually and it's not just images it's videos as well they showed uploading a video to web flow to web flow assets to be more specific and it the AI edits the video you talked about uh editing videos with fable 5.1 well web flow does that Now uh it can remove the a and um and like loom and actually cleanshot x which is uh something that I personally use. I used it actually for making even these screenshots and they came up with something like that as well. So this is just everybody is doing everything I guess. Um, so it can cut the video and then it can um they they showed like it can create like different versions for different uh social channels and then you can push it through web flow I guess to these different channels. I don't have access um I I'd be interested in testing it but yeah I don't know. uh they announced that the the whole asset things um with campaigns. So campaigns is also again like leaning really hard to marketing. They they literally call themselves marketers now. So web flow is for marketers now. They they the the two words that they used most during the entire conference were two things marketers and agents. That's that's that's what they basically talked about marketers and agents. So here you have also campaigns and you see like brand you have like the assets and you can create these um campaigns and I guess these are pages which is a bit strange because now we are back in the the web flow like build mode but for marketers and yeah you you can create a c campaign and then uh I guess this is the asset Yeah, this is the asset. This has nothing to do necessarily with that, but kind of like both part of the same thing. You you push these assets to these social channels, but you also create the whole um landing pages for them. And they they showed also I want to show here uh you want to also measure the data with AI and you want to make sense of the data. So they still have the insights and the all the data and so on. They mentioned a few other things uh that I see here I added to my notes. Um this this was already released. I don't know if you have used it or not but you can add skills and basically these are just MD file instructions added to your web flow project and you can generate these now. So you can when you hit generate it looks at your whole web flow build and it creates documentations for how a component is used. And I guess I guess that's useful. But I I'm a bit hesitant because this is only useful if I would as a developer go through this and confirm this because I'm seeing way too much of meat proxy being like becoming a a thing where people generate an audit with AI and then give it to another AI to you know fix the things. And I noticed like I it happened multiple times in the past two weeks where I actually start noting down things that AI mentions that don't make any sense. So I would be very hesitant to you know just generate this type of documentations and give it to another agent to fix something that you don't understand. I would I would be just hesitant. uh it's not as simple as just you know um passing passing the information between two agents. They also released not really announced releases which are just kind of like the way git works I guess but simplified for your website. So you have like these different versions of your site and like different people and agents work on it and then you have you have different versions and then you can merge them together and then publish the merged version. So but we don't know if this is an enterprise thing or or not. So yeah that's that's kind of been it. Uh it was a lot to take in. Uh it was a lot of marketing jargon. It was a lot of like it was for marketers in a way the whole thing. Um but yeah we have now source still waiting on trying that. Uh, it looks promising, but I think it can only be relevant if it's free, at least to try. And it can be really exciting to me to use it to start building something with it if I can use my own harness and subscriptions and and whatnot. So, we'll see about that. And once uh that's out, uh, we'll cover it. Any questions, thoughts? falling asleep there. Yeah. Yeah. It's a lot of marketing. I feel free to cut down the video, by the way. Like I don't I don't know. I We'll be cutting that down. Yeah. Yeah. I went through everything, right? That's not necessary. Yeah. Um Well, yeah, you kind of mentioned at the end because one of the biggest things they also didn't mention was price. So, we don't know whether this is going to be a free tool or a paid for tool. I've said it before, my my not even opinion, my suggestion is that they make the tool free, whatever, because that's a given, but also they make it um publishable on Web Flow for for free. Now a lots of these other tools they have I think those will be I think they should be paid for. I suggest they should be paid for but the problem is Facebook the problem is web flow are competing they're competing with free right devs or sorry not even devs sorry web flow designers now know how to build a website and host it on Versel. So that's their biggest competition right now for builders. But it doesn't seem like their presentation at all focused not at all next to nothing on the builders and those are the people that are going to later on sell to enterprises to you know it's got they've got to they've got to balance out that focus a little bit. So not mentioning the price will really make or break this product to be honest. Um otherwise there there is going to be a massive if you have to pay for this tool let alone actually using web flow once you've got the tool then there is going to be so much uh what do you call it when there's mass like exodus like a mass exodus of web flow designers cuz they just like I would it's so easy for me to just not use web flow why would I go to this the other thing I will say oh go M no no no you're you're completely about the price I I think you're completely right like people but you I want to just mention that you're talking about us like small agencies and freelancers you're not talking about enterprise companies or like big big companies but no I am talking about those because the people that work for the enterprise companies need to yes the ones they've already got sure they'll move on to this and everything will be fine. But they are going to put a complete stopper to new companies because no one's going to be building with web flow if it's a paid for tool and a paid for e you know whatever. um even just a little bit of money I think people will just leave like people want to use the builders want to use web flow but if they have to pay something then they're going to be like h it's not worth it and so they are shooting themselves in the foot because new enterprise won't be able to find web flow developers who use this web flow there's this there's nothing attractive about this like that you just can't do elsewhere it holds the current audience it holds a current enterprise client and gives them something nice and new, but it does not it's not attracting new uh new prospects as it looks to me. Interesting. So, I I went through like this whole video. I just explained what Web Flow explained, right? It was a recap. Like, if I would give my two cents on this, I feel like they they overstaffed this tool. It looks like that to me. It has way too many surfaces. Like it has views, CMS, assets. It's like so many things that I don't necessarily care about. I don't know if I care about all of these things or it looks to me like instead of them coming and saying like here we have a tool that is basically lovable, you know, uh because lovable is very simple like at least for the start of it. uh they they came with something like to show hey actually we we thought about everything and it has so many things things that I don't know if I care about or like I don't want those things to get uh to I don't want these features to block me from using the tool itself by all being all making the the product being like a paid or expensive product I like the idea of having CMS and assets as like that tool to generate assets and brand and campaigns, but I want to have the option to not use any of these things and just use this with my own agent subs with my own open AAI and anthropic subscriptions that I got to use maybe some of their agentics. So like I I guess a lot of these are just MD files behind the scene. Correct me if I'm wrong, but like a lot of these I I would like to use the simple stuff first and then and then maybe I find brand useful or like campaigns useful and then I I buy it. Uh you know, maybe I can have a free trier, you know, I can try it and then buy it. I think that's where they can compete. If not, I find it to be honest really hard to I like I maybe if my client want to pay for this because it's nicer to them, that's fine. I wouldn't advise them against it, but I myself I'm already paying for claude and I'm pretty happy with the claw desktop app. And if I want a CMS, I add a CMS. If I want an asset generation tool, I mean, I've showed it in the show before. I can create one that does you know what I need the way I need it. So it's a difficult time for software altogether because I can create the software that I need in the way that I need it. And as I've shown before, like with when it comes to creating the website itself, it's just becoming incredibly easy to do with things like Astro and just AI, getting something really good up and running. Um, even the CMS, the reason I'm saying like the Web Flow CMS, the current Web Flow CMS is not compelling anymore is because you have so many better free options that you can use. And I just don't want to pay for yet another AI subscription. I want this to be free to let many people build to not do what Web Flow used to do. you still have to pay for freaking code being able to add even one line of CSS to a free web flow project. Like you still can't do that. That that's insane. Like the we're in the age of you know AI creating anything with code and you still can't do that. So I want web flow to not repeat the same thing with source. Make it free. let people build and the way you said it, these people who will end up building with this, they will also end up building for the big companies. Um, like let the community use this, build with this, um, share things with each other and maybe this is a way to, yeah, let the community come back to life in a way again. But honestly, we we watched the the show with a bunch of like uh web, you know, Web Flow community people, a bunch of like likeminded people. We saw all of this. We were really happy with the Timothy uh announcement. Uh we were excited about source, but we kind of like had our thoughts that it would be cursor repletable like. And after we saw it, we were basically like, cool, now let's go back to building with codeex and claude because that's the reality of our world. Yeah. My impression, my feedback, my initial feedback from watching this and understanding the tool is that Web Flow are still hellbent on creating an ecosystem that you cannot leave. Mh. And it sounds like on the surface that they've made everything open and whatever. Again, when they announce the price, that'll be where we finally learn how how much they've let go of that ecosystem uh mindset. I think it's the wrong mindset is coming. I think it's coming across as greedy. Um, but that's not the way that they are going to compete because people like you have have discovered open source and what it means to be an open-source platform and to have intero interoperability. Let's see how much it costs. Um, there was one other thing I was thinking about. Um, that's what I was going to ask you like who is going to use this tool? There was just so much that's gone. Marketers There's so much that's gone into it. Like no one is going to be using the build and then jumping into the inbox and then you going into the analytics platform and then generating images like very few people are in that whole world and arguably the marketers didn't need a tool. They they would they would much prefer prefer to probably work in the browser or through their agent. They didn't need to add that to source. I don't think that's where the primary use um you know if you were to look at the user data of who's using source I don't think they're going to be in those windows and those areas I really don't this needed to be a tool web flow this needed to be a tool for your builders no one else everyone else is is happy with their MCPs and logging into the website no one asked for any anything like this people wanted a new way to build. I agree. And and I think if they would do this for builders, we the builders, we could provide the marketers, the like other like SEO special specialists, anyone else, we could provide them with the tools they need the way they needed with this tool. Like with source, we could build the Nex.js website, you know, we could build it with astronjs, whatever. and build the tools they needed. I mean, we literally did this for a client. I mean, you did it. They you did their brand pipeline image gem image tool like the shader tool that you made. This is something that we're doing like like everyone is doing this, right? So, I don't see the value in a lot of this for me. But again, like if my client buys this like and finds out about this and they buy it, sure. And they find it useful. But I I'm not sure if I would even like tell them, hey, buy this. I my it's just not my clients. Like my clients are not marketers. My clients are business owners. You know, they have businesses to run. They care about their site being, you know, beautiful, being performant. they care about like an a CMS they can rely on edit you know if they need to but a lot of them I also see they are starting to do things with MCPS and this tool let's see let's let's get access to it and I hope I can you know um I can do something with it and it's not like you know too expensive or yet another subscription with or another blocker like a way to it gets in the way of building websites. Uh I'm going to give Ship Studio a try. I don't know why I haven't by the way. I really don't know like told you about ages. I know. Like you told me about it. I I known about it but I just I don't have the Okay, let's talk about it for one just a second. is I don't have the appeal for the style panel anymore. I did I thought I needed I installed uh Stacky. I used Saki like every single day for a week, more than a week probably. I tried the style panel. I didn't need I It's just Yes, it's faster to do the styling directly, but I'm building so much so fast with agents that I don't know my own uh way of like styling things any and I don't care I don't I care about the quality. I care about the quality how how how like the spacing is across the entire project uh uniformly um applied but I don't care about how it's like what what the class is called. I don't care about that. So by that I can't go in and change it. I could but it's not you know it's not the most efficient use of my time. So, I noticed even though I try to change stuff visually, I noticed I'm better off to tell the agent to do it for me and then I check if it's correct. And in the meanwhile, as the agent is doing this slowly because it is a slower process, I do other things. I there are just so much there's so much work to do, so much to create. So yeah, I'm like I'm not like fiddling with uh with those small details anymore. So maybe that's why I haven't tried this yet because it felt to me like a visual editor that I don't need. Um well, but we'll see. Well, going back to Web Flow, I think the price will determine everything. Yeah. And is it that important? like it's important to me but I don't know if it's important to the clients that web flow was talking to the web flow was talking to marketers marketers and it's like I hear marketers I hear enterprise because like companies that have a marketing department these are not tiny companies right this it's not like a uh the founder doing marketing they have like at least a marketing department people who call themselves marketers That's that's enterprise to me and we know that they don't care about price as much. I'm more yes I'm I'm for myself I'm very concerned about price but I'm also very concerned about not having access to you know my subscriptions. I have Claude and I'm happy with the GPT 5.6 six uh that I use and the claude 5.5 that I use and I I want to plot I I want to similar to ship studio which I haven't used but I get the concept like use my own agents and then use the web flow source as a visual thing and on top of the visual thing we have the composable CMS maybe I can use it for free and publish to web flow that IO for free. Uh but once I want to publish to my main domain, I pay for it. Yeah. Yeah. Um there was something I was going to say. You have a lot to say about this and then I think about something that I want to say and then you move on. I'm sorry. Yeah. I'm sorry. Um it was something about Oh, yes. So, uh, interestingly, we had a comment on one of our previous videos. Someone was very upset thinking we're thinks we're trying to we're trying to clickbait by [ __ ] on Web Flow. And I think I could I could not agree, disagree, sorry, more that we want Web Flow to survive. We want Web Flow to do well. You more than me. Obviously, I have no use for Web Flow anymore. But you you you want we're not but it's it's we we have to call it out. We have to call it out as we see it. And I just don't think we're looking at a product that that will that will work. I mean I might we might eat our words but I I I really think I can't stop focusing on this price aspect. price will determine whether they're just being greedy and trying to lock continue down this ecosystem route or they want their builders to to have a nice shiny new tool with a bunch of like nice little extras that probably won't get used. But hopefully you'll be able to, you know, customize the interface and from a user user permissions perspective like someone in your team has only access to certain area, you know, certain elements and whatever of the app. So it cleans it up a bit. You don't need your marketer to be looking at the you don't need sorry your developer to be looking at the stat analytics of the website. You just don't need to show them that or whatever. Um yeah I want I want web flow. You want Web Flow to do well. You want we want better tools particularly as you're still building Web Flow websites and still need to manage Web Flow websites. You want a tool that's going to be good because right now web flow as it stands is is slowing you down. It is that that that's the point. Yeah. The things that I'm building with web without web flow I can't do them with web flow anymore and that's yeah it's slowing me down. I if I convert my Astro um Astro project and like try to bring it to web flow I I'll just quit. I I'd rather quit. I can't do it. What you'll have to do is just use Web Flow Cloud and just upload that Astro project to the root of the project. But why why do right now just so you can build in web Astro with your agents and get it running within web flow? But I don't need uh I don't need web your clients is is what I'm thinking like clients. Well, like with my own agency website that I'm rebuilding and I've been showing it. Um, right now it's on GitHub. Uh, sorry, on um Yeah, but I'm talking about your clients. I'm talking about your clients. But like, let's say you get a client, they want to build in Web Flow. They want you they want to build a website in Web Flow. Obviously, if they want to be able to control it in the editor, you cannot build it in Astro. But if they simply want to because you can link the CMS with a dev with a with a web flow cloud project I think I think not 100% sure on that but if they just want to be able to use the CMS and that's but I can use payload but if they but stop think web flow stop think web flow they want web flow and they want the CMS they want the CMS web flow and Maybe they like the analytical stuff. They like the analytic stuff and they've got all manner of other websites in Web Flow. Yes, I get your point. But to be very clear on this, and I think this might sound harsh, but most of most of our clients, and I'm not saying just mine, they would not know what Web Flow is without us. They would not care about Webflow. That's a different subject completely. That's a different subject completely. And that's why I think they need to focus on the builders. But yeah, so if they come to me and they say we we want web flow and I show them something better, I think I can, you know, show them something better today. Maybe we are we're talking about today. Maybe source will change that. But if my client comes and says, I want a an a marketing site and I want this and that. If it's more complex than what Web Flow can do today, Mhm. I can show them other things. Sure. Take our previous previous project we've just done. There's I think it's yet to launch. I They were adamant they wanted that on Web Flow. I was like, "Look, you can build with agents. We don't really need a CMS because you don't touch the designer all too much. I can give you payload as a CMS." They were like, "No, we we just want it on we just want it on Web Flow." So, you might I know your assumption is um that you won't get clients like that, but you may very well get clients who are just adamant on being on Web Flow for the wrong reasons. Maybe you need to be open to the fact that the end of the day, they're the one paying the bills. So, you need to just agree, you know. Um just to be clear, I'm not trying to move them away from Web Flow. I want web flow to be a modern tool that I can reliably use. Yeah. So the last related semi-related question in all this and it's to do with young Timothy. What happens to Staki now? What happens to Lumos? Now listen, there were there were some conspiracies going on both in person and online. people saying, by the way, I genuinely have no idea if any of this is true, but there were talks about Webflow saw Timothy, the biggest creator. He's he was the not just the biggest creator, he was one of the last standing creators for Web Flow, right? He was teaching Web Flow people to not use Web Flow. He was showing us how great uh you know, Astro is. He was literally like learning Astro. He was uh streaming his He was streaming himself, learning Astro, building with Astro. He built Stacky, a visual builder that was to me always like how is he not scared of Web Flow, you know, hitting him with a uh cease and desist or something because he was like literally copying the Webflow UI. Uh anyway, so there were these conspiracies that because he was bringing people away from Web Flow, maybe Web Flow hired him entirely or partially to to to be like, "Put a stop on this." Yeah. So if you want to be hired, they bought his followers basically. They bought his reach. Yeah. They bought his Yeah. But Yeah. But he's not going to do content uh on the web. No, I know that. I know that. But he he has fans that trust and follow him and his opinions. So they will in turn I mean there might be something in his I don't know what his contract looks like but there might be something in his contract to say you can't produce content on I mean he what you say he had to close his his um community but that was the Patreon contract. Yeah, it might contract. I don't think that was a part of his contract because he replied uh to a comment saying that he didn't want the people to get charged for something that he's not sure uh continuing to develop and I do trust him by the way. It's a strange thing, but I do trust. Hang on. Listen to what you've just said there. Like that there's a word for that. Like but why is he not going to continue to contribute to it? Of course, he doesn't want to charge people because he's not contributed, but why isn't he contributing to it? Yeah. Yeah, I get. So, I said I trust him, but I don't say like necessarily I trust, you know, a big corporate like, but he also didn't tell the truth. Maybe. No, no, no, no. That's not what I'm saying. I'm saying like I don't know if I can trust Web Flaw on that. So, that's why I'm like maybe they they they actually maybe got him partially at least so he doesn't continue on that path of Yes. That's what I'm saying. So, it's it's a it's a semi non-disclosure agreement. It's a look, we we if we if we hire you, you you got to stop. You have to talk about this because it's part of your contract. Exactly. And it's a standard by the way. Yeah. Um because interesting, you know, that you know that LinkedIn post or tweet or whatever that went viral a few weeks ago. Yes. Do you think he would have written that if he had an upand cominging uh employment contract signed by Web Flow? It might give you an idea of how quickly this thing came about at what stage in his journey of transitioning away this this might have happened. And what was that two weeks ago? This may have literally come together in the last two weeks. By the way, you remember like Web Flow bought GSAP. They It was a lastm minute thing. They even said that the ink on the paper hasn't like dried yet. It Yeah, it it I don't know if it was I don't Either way, it was like a really like like thing that they did right before the the entire um conf. The thing with Timothy probably was the same. um Christian from the community. He said he he's quitting his community because he's joining web flow. And I he he predicted this correctly. And I said no for a reason. My reason was in the post that Timothy wrote, he said the community has been shrinking. And I said, if Web Flow would hire him and, you know, look at the whatever he writes, they probably wouldn't want him to write that that part of the the post. The rest was fine, but that part like the community has been shrinking. It's a bit too negative for a company like Web Flow, even though it's completely true. The community has been shrinking. Anyway, that was my reasoning to think that he's not joining web flow with an additional reason that I think is still relevant. I was like, what is he going to do? Like he's not an engineer like an engineer like you know like a software engineer and he's not going to join just to make content because he has the love for creating. So he's a builder. So what can he do? So I guess he will be doing like vibe coding contribution maybe uh and maybe some other engineers would look at the code before merging it or you know whatever like overlook it but he will be shaping the product. I think he knows a lot more than Well, he does, right? I think he knows engineer. Well, he knows a lot about front end. But I think like building the whole product, I think that's a different beast because you need to think about the compatibility with like the whole system that they have, right? which makes you think why he's working on the web app because it's a front end potentially a front end system but agents like you don't need to worry about it as much so it's hard to say we don't know I mean then they would spend next thing we could work for I know yeah I don't know just make a we we well stacky is still up and you could like literally take it and uh you know contribute to it but I think I'm not going to do this. I'm gonna uh test ship studio and I think what what is his name? Julian is Julian, right? Julian. Yeah, Julian. I think Julian would be like looking at uh source and be like build us build us like before we get source we get everything from source in uh source in ship suit which is a strange world. uh like a one man can effectively copy all of that but we'll see to what extent. Yeah. Yeah. So there we go. Either way, I'm excited to use s source. I'm excited to use the new CMS. Uh I'm excited to actually I want to try a bunch of like new other CMS uh solutions. I want to use payload again. I have used it in the past but I want to use it again. Uh I might not even have a CMS for my next website. Like why not just use MD files? I don't know. Uh we'll see about that. I I mean I mean we are really entering some strange time with software. So but what you can bet on is we will share all of our experience with you for sure. Do you know what I want to use? What? Astra. Astra check. Anyone got Astra? Anyone using Astra? Not Astro. Astra. Oh, Astro. Astro. Oh, no. Astra. Astra. GBT. GBT6. Astra. Yes. Yes. I want to use that as well. Astro. Check. Negative. Yeah. Negative. Cool. Well, is that all we got time for? It was a long one. Any last words? Um, don't be a meat proxy. Don't be a meat proxy. There you go. Stop being a meat proxy. I Some people are already a meat. Stop being a meatrox. I am not a meat proxy. I'm a human being. You're not. I don't know why snigger all of a sudden. Yeah. Yeah. That was quite a good impression, I think. Yeah. Um, it was cool. All right. Well, that is all we've got time for. I'll be probably trying to edit this Web Flowly video more succinctly for the for the channel. Go check us out on YouTube and give us a subscribe. Um, I think do you reckon we could hit 5,000 subscribers by the end of the year? Because I think I think we could if you if you if you can help us, all 51 of you I can see on Streamyard sign side. Maybe I don't know how many is on Instagram side. Uh if you all head on over to Kamando Show or search commando show uh on YouTube, give us a subscribe if you haven't already. It would really mean the world. We are We are We've got some good We got some good um ambitions for the show. We just need more exposure. Yeah, I think we had the most amount of like viewers uh simultaneously on our YouTube. I think we had 21. What's that? Sorry. Right now. Right now we have 19. So 19 of you are there. So comment something. What? Like why is there so many of you there and not commenting? Like I I don't understand. What drew you to the video? I reckon it's Web Flow. I reckon it's Web Flow title. Yeah, of course. Yeah. What interests you? Yeah, that's why you are here and that's why we are we are building our own source. Uh just just comment and we'll do it. No comment. So, it's stop being a a meat proxy. Okay. No, 22 people. 22 people watching on on YouTube. Maybe it's all Julian. Um, well, they all might be watching at different points. See, someone might have joined but then rewound the thing. They're not watching live. I don't know how it works. I don't know how it works. But, uh, yeah. And we were also like 8 seconds behind. Yeah. You know, so who knows who's actually genuinely live with us here. Anyway, give us a subscribe. We love you guys. Um, that's all we've got time for. We will be uh great great podcast. One of my favorite. Thank you so much. That means the world. We uh we're we're we're doing our best. We are, do you know? And actually, sometimes we are some of the earliest people. Some of the other content creators, they're like a week behind on some of the stuff we talk about. So bearing in mind that we we do tend to be try to be as quickly as possible on the stream and talk about things before they gain any sort of traction. But uh yeah, appreciate appreciate the love. Um join us next week, same time, same place. This has been Command AI. I've been Sam and I've been Kabarza. Wait, that was Don't be a meat proxy. Cheers. I'm just so annoyed with