Claude Design Changes Everything (Figma in Trouble?)

Sam Nadler: When I was aligned with how it came out, I just clicked hand off to Claude Code. It created a zip file. I then gave it to Claude Code, and within, you know, however long it took it actually took quite a while, thirty minutes. The entire web app was fully updated to the new design, and the results were amazing. Hey, everyone, and welcome to Built This Week, the podcast where we share what we're building, how we're building it, and what it means for the world of AI and startups.I'm Sam Nadler, cofounder here at Ryz Labs. And each and every week, I'm joined by my friend, cohost, business partner, Jordan Metzner. How are doing today, Jordan?Jordan Metzner: Hey, Sam. Happy to be back. Another exciting episode. Huge week in AI. We've got some awesome new excitement, some news, some cool new tools, and it seems like, the AI news is not cooling off again this week.Sam Nadler: No. Some huge releases. And before we get into the docket, don't forget to like and subscribe. We're at 25,000 subscribers on YouTube. New episodes out every Friday.So please give us a like and subscribe and and continue to support Built This Week. So we're gonna do kind of a a double coverage of Claude Design, which was released, I believe, four days ago. So very fresh. And then we're gonna jump into some of the latest news. We got some news from SpaceX, some chatty p t news, and and and then we'll wrap.How's that sound, Jordan?Jordan Metzner: Yeah. Let's go. No. I know you've been playing with Claude Design a lot. I've been playing with it a little bit.You know, again, this is an interesting time where Claude's Anthropic is able to take their model, build a new interface on top, and almost build a brand new product, and almost maybe segment killer. But, yeah, I would love to go into, you know, what you thought of the tool and how you've been using it. And and, yeah, let's go from there.Sam Nadler: Alright. Well, I've seen some actually mixed reviews of Claude Design on social media. But my experience, I've absolutely loved it. I've found the design quality incredible. I found it easy to use, and I've been pretty excited about it.I've done, I would say three projects with Claude Design so far. So all three basically start as a chat. And I would say, to me, it feels like a chat, a UX UI designer at your hip that you can chat with, engage with, have them propose ideas. And then the output I found has been really high quality, but it also gives you a ton of optionality. So when it gives you the output, you can change font, you can change color, you can obviously chat with it.Usually, the project starts obviously, you know, just prompting something. Right? I'm looking to build a deck. I'm looking to do a full redesign of this web app. I'm looking to build social ads.And usually the first thing it creates is a questionnaire. And usually there's a series of eight to 12 questions. Once those questions are answered, it creates the first set of assets, at which point it gives you kind of these tweaks that can be made in the moment and you can change. But then the rest is all managed through this left hand sidebar.Jordan Metzner: No, I think what I'm a little maybe not confused by, or I think what's different here is that this interface was shown to build ads, but you can use it to build web apps or videos or any interface. Like, it doesn't really matter what's on the right hand side of the screen. Like, is this a web app or is this a showing you the ads that they generated for you? I'm I'm almost, like, overwhelmed and confused by how good this whole thing looked.Sam Nadler: Yeah. So, I mean, Claude Design is a web app. And then what it's in this context, which is I was asking for different meta ad concepts, it really kind of breaks out how it thought about creating these different concepts for me. So this was a little bit evolved. Chatted about creating some hip train ads and then I subsequently prompted it to target women between 50 to 70 years old, which is generally speaking, of our prime demographics for this service.And it created these concepts. So it built this presentation for me. Three concepts of women who have been here for a while. I wanted it a little cheeky, self aware, and it captured that to some degree. It builds out these different concepts.Turns out I wasn't done. The things my doctor said I couldn't do anymore. The grandkids think I'm chill. The waits know better. So for this particular one, I added an image.It actually gave me an image prompt. I asked for this. I was like, create these three campaigns, give me an image prompt that's relevant to this campaign. I used nano banana for the prompt. And then I wanted in case I wanted to turn this into a video asset, give me the VEO3 prompt that would align with this campaign.And it gave me, that is what this presentation tier. And I only did that for this one. That's why there's an image here for this campaign versus this one. There's no image here. But it gave me, I believe it gave me the prompt here.So here's the image prompt and here's the VEO three prompt. But for this one, I took this image prompt, put in Nano Banana, then I gave it back to Claude Design. And as you can see, here's the ad. I mean, you could continue to iterate here. The grandkids think on chill, the the weights know on better.And it gives you so much, I think, just really, really great output. So that's that's how I was playing with like different ad creative. I wanna show a couple other let me go back to projects.Jordan Metzner: Okay. And how do you get those ads out of there? How do you get them out? How do you export?Sam Nadler: Great, great question. So there's tons, well, that's actually really interesting. So there's tons of different share capabilities. What I found was super helpful was you can actually, I didn't know what I, I knew what I liked, but I wanted to share with our marketing team some different options. And you can actually have it create a link where basically it creates a presentation that they can comment on or even edit directly themselves within Claude Design.There's you can export it. So in this, if you can see my screen in the share options, there's you can download the project as a zip, you can export as PDF, you could send to a client per se, you could send to Canva, you could export as standalone HTML. And this one's the really interesting one. I thought hand off to Claude Code. So when you're doing a website redesign or web app redesign, you can directly link your local code base to Claude Design.You can go through the process, chat it up. I've done this once. It gave me an absolutely beautiful design. When I was aligned with how it came out, I just clicked hand off to Claude Code. It created a zip file.I then gave it to Claude Code. And within, you know, however long it took, it actually took quite a while, thirty minutes, the entire web app was fully updated to the new design and the results were amazing. It was, you have now, you know, shown a couple of members of our team who are maintaining their own web apps for their own purposes how to do this. And about an hour ago, they were showing me the new designs, they were brilliant. Brilliant.So it's been there's lots of different ways to get this out of the the web app and share with teammates, share with clients, give to Claude code, etcetera.Jordan Metzner: Awesome. Super, super cool. Okay. Yeah. You were gonna jump into the another example here.So you've got multiple tabs in the chat and multiple tabs in the app. Is that how this works?Sam Nadler: Yeah. So I, you know, I don't know if I'm doing this correctly at the end of the day, but I have multiple chats. Usually when I'm starting something new, I just start a new chat. So it tells you right here, start a new chat to save, you know, save on contacts.Jordan Metzner: Understood. Cool.Sam Nadler: So it's been prompting me that quite a bit. I think I've been overusing the tool. But yeah, you start a new chat when you're on a new kind of avenue of building the project. But here is a internal web app that I basically redesigned. This is from internal app that we use, and this is the exact functionality of our app.This is what it looks like in production, but it gave me a very, in my opinion, polished redesign of the app and it gave it to me page by page. So this would be the homepage, this would be another version of the page. And it's a little squished here, but I took it to this is the one where I exported to hand off to Claude Code, gave it to Claude Code, and within thirty minutes, I had it live in production. Amazing. Fully redesigned all the way through.So I've, you know, I've been having a lot of fun with it, and I think it's gonna be it's gonna be really empowering to our team. Building with AI shouldn't take months. We do it in weeks. Ryze Labs helps companies design, build, and scale AI products with elite engineers from Latin America. Real teams, real velocity, real outcomes.This is Rise Labs.Jordan Metzner: Now let me ask you a question. If this was just, like, rolled into Claude code for you, then would you just abandon this? Right? You just you just want these, like, design based features in Cloud Code anyway, right, or codecs or something similar. Right?Sam Nadler: I think maybe maybe it could be rolled into co work somehow and but, yes, absolutely. I'm a little surprised they they didn't they made a separate product.Jordan Metzner: Well, think I mean, as you know, for software development purposes, it's just much easier to build a web app than it is to integrate something into a native desktop application. So it just seems to me that, like, they were able to build this as an easy way to test to see if people would like it or not. And if they like it, then just roll it over into the desktop application, which I think we've seen before with Cowork and some of their other tools. And, you know, it just makes sense like that because one, you've got a web dev team that can move really quickly, and then also just like you're able to test these hypotheses on whether or not customers actually like these products. Because, you know, I think we're talking about Cowork and now Claude Design, and these are products that customers have been loving and using all the time.But I'm sure there's been some other Claude tools that they've launched that have not got as much traction and therefore maybe don't don't warrant the necessity or need to implement them back downstream into the native application. So this does seem like a gateway ish drug, but, yeah, it just seems like a really awesome tool to improve design. And, mean, I it it definitely feels like a little bit of a Figma killer in my opinion, but I don't know. How do you think?Sam Nadler: Yeah. Totally. I think, like, you know, I'm I I would say my Figma skills are pretty awful, and this is incredibly easy. I mean, we it's it's as easy as chatting with any AI and it does a really good job and it's intuitive. I've already redesigned multiple things and I think top tier.So, we'll see. I mean, obviously I think Figma will have its place, but this is I think I think this is a step it feels like a step for me in my what I'm doing, a step function in my capabilities.Jordan Metzner: Totally. Well, let's jump in to the news as we have some really good news articles this week.Sam Nadler: I don't know the exact deal terms, but I think, SpaceX and Cursor SpaceX has the option to buy Cursor, I believe, by the end of the year for 60,000,000,000 or breakup fee of 10,000,000,000. You know, obviously, SpaceX has the compute and Cursor is is the coding agent, but a pretty interesting auction that Elon drove here.Jordan Metzner: Yeah. So let's let's jump into a little bit. So, you know, I think, like, you know, before the news was announced, I think everybody pretty much believed that the the coding race was being led and won by Anthropic. I think most people believe that OpenAI was kind of in second place to Anthropic, especially in enterprise, especially considering how much Anthropic has grown over the last few months. And, know, we've seen some of the drama between Sam and shareholders and board members.You know, we saw the Ronan Farrow article came out on OpenAI. And, you know, we've got, you know, in in kind of third and fourth place, we've got Google with Gemini and their Gemini CLI and some of the other Gemini tools they've built. And probably in, you know, third or fourth place, depending on how you wanna rank them, you had Grok with, you know, their tools. They've got image generation. They've got a real time API.Obviously, Elon's models are a little more risque and, you know, telling dirty jokes and things like that as he's kind of been pushing for. But, you know, I think he'd been mentioning over the past few months that he's fully aware that Grok is behind in the in the enterprise, like, coding space and that, you know, he was working really hard to improve that. I think at the same time, what's interesting is that, you know, Elon had just set up one of the largest GPU data centers with NVIDIA just, you know, a few months back. And, you know, Jensen was talking about how it was one of the largest data center builds from scratch of all time. And if you read this article and you drive through it, what you see is that XAI or SpaceX, now they've kind of been merged, is offering excess compute downstream to to Cursor, but also with this, like, option to buy.And I've seen a few people talk about this, that this is actually gives SpaceX some interesting optionality because they're able to kind of understand and define whether the next level up for Cursor is really about inference or about, like, kind of, you know, generating a new model that can then compete with the Claude codes and the Codexes of the world. But, obviously, a $60,000,000,000 incredible price point. I think the public data says Cursor was running at about $2,000,000,000 run rate, so this is 30 x on revenue. You know, I know that, like, the most advanced teams, like, if you think about our, like, really advanced AI teams have kinda, like, moved on from Cursor, as we mentioned, gotten out of Cloud Code and Codex. But, you know, we're still seeing lots of enterprises just starting to adopt Cursor.I know Cursor is really well embedded inside of universities and other enterprises where it takes a lot longer to adopt new new coding tools. You know, they launched their own model that the Composer two model, I think, that was, a wrapper on top of the open source models a few months ago. If you recall a few weeks ago, maybe that was. But, yeah, I think, you know, maybe many people thought Cursor was left to dry or, you know, was kind of gonna slowly dwindle, but I don't know. I think what this deal means that, you know, it kinda puts puts SpaceX and Grok and Cursor back into maybe that number three slot against, against Google.And I think it just makes the market more interesting, probably makes prices more competitive as far as kind of token prices. And, as you know, Elon plays the long game, so, you know, you can't bet against him. I don't know. What are your thoughts?Sam Nadler: Yeah. For sure. I mean, it just feels like we're still in the first inning of a very long game. No. I'm excited.I'm excited to see what will happen. Obviously, you can't count Elon out, but it's a it's a bold move. That's for sure. The other article we're going to cover today is yesterday, ChatGPT launched Images two point zero. I know we've both had very limited time to play with Images two point zero.I think your reaction has been underwhelming, while my reaction has been, it's actually pretty good. I think what I've heard, mainly on social media, but according to this article as well, is that the text fidelity is really good. In previous models, you would write something like, you know, Jordan.Jordan Metzner: You'd get, like, gibberish. You'd get gibberish. Yeah.Sam Nadler: Yeah. You'd j r d n or something like that. Mhmm. Mhmm. But apparently, for this, it's, like, really high text fidelity, which is, you know, exciting.Jordan Metzner: Yeah. I've been playing with it a little bit. I still I thought I was a little underimpressed, to be honest. I think some of the examples they showed were more impressive than the results I was getting. This background that I generated today was generated using Image two.And, you know, on almost every episode, Sam, I generate a artificial background, a new one. I don't know if there's a significant difference between, like, my previous backgrounds that were being generated by Nana Banana or not. But, yeah, overall, you know, it's interesting to see ChaChipiti into the space and get deeper and compete against Nana Banana, where it seemed like Google had almost, like, run away with it a little bit for a while. But, yeah, you know, it's interesting, especially because they shut down kind of the Sora business of, like, generating content. So it it is a fun, cool space.I think it's, like, the more models and the cheaper it gets to generate images and video, the better. And I think it just makes, like, working on the Internet a little bit more fun. But I think, like, you know, one thing that stood out to me, Sam, you said it's, like, just the early innings. You know? SpaceX has kinda filed an IPO to go public.OpenAI is a private company. Anthropic is a private company. So, you know, we've got, like, three out of the four leading leading laboratories right now. We're all private. Obviously, Google is a public company with, you know, a huge cash pile, but it just shows how early it is.I mean, I'm I'm in the thinking about the Internet kind of before Facebook and Google were public is very early, early, early days. So, you know, thinking about Blue. These yeah. Yeah. This is, like, you know, in the early two thousands.You know, I know, you know, Microsoft and Amazon, these companies went public with much smaller market caps with much less venture capital raised. But the fact that, you know, OpenAI and Anthropic have so much revenue pre IPO, SpaceX too, I mean, even though the multiple's a little crazy, but, yeah, it just shows how early it is. I mean, you can imagine these companies be public for ten to fifteen, twenty years and have a totally different valuation than, obviously, where they're gonna go to market at.Sam Nadler: Yeah. I feel like about six months ago, it really felt like Google was making huge strides. They were really catching up. But it's been, you know, it's I could be rock here, but it's felt like they've been a little quiet over the last quarter or so, where Anthropic is obviously, like seems like it's really, like, taking the lead. And then OpenAI is, you know, still still contending.Do you expect anything do you know of anything on the horizon for from Google?Jordan Metzner: No. I mean, I've tried their new real time chat. It's pretty good. Some of the new Gemini tools, it definitely seems like they're a little bit as far as, like, competing in the, you know, in the code code race against, you know, Anthropic and OpenAI. And, you know, some news on Twitter has come out about how, you know, some teams at Google have adopted AI quite well.Some teams have not. It seems like the teams that have have been using Claude code. So, you know, maybe even at Google, they're not using, you know, Gemini as its tool set. But, you know, I I think what you have to take a step back on. You know, it's kinda like building you know, every time these models launch, it's almost like building a hotel.You have a city. You don't have enough, you know, hotels for the tourists to come to your city. So, like, great. Let's get a hotel. But from the idea that, like, hey.We don't have enough hotel rooms to, like, the hotel opens, takes, you know, two or three years. And by that time, you hope the demand is then increased. And then by that time, you need more hotels. And so you're kind of, like, always on this thing where you're behind on the amount of, like, you know, availability that you have based on the demand when you're in this, like, growing market. Now if you apply that type of model here to, like, how long it takes to do the training set of some of these high end models.You know, I don't know how long 4.7 took to train or, you know, or 5.4 from from from OpenAI. But, you know, if it takes three months, six months, nine months to train some of these models, then it seems obvious that you're gonna go through these quiet periods where it looks like some of these major laboratories are doing absolutely nothing. But in the background, they're at maximum training. Now, the the problem you have is that the longer this iteration is, the the more risky it is to throw, you know, approximate iterations in so that they then get built later. So, you know, you start the building of the first hotel today.Do you start the next hotel next month, or do you wait until your hotel is built? Or do you wait halfway through knowing that you don't know whether your hotel is gonna get built on time or not? Right? So you start to get into this space where it's like, Okay, well, how much money do we have if we already have a model that's going to be done in, let's just say, ninety days from now, do we start another model today with some of the things that we didn't get to put in the last model, or do we wait in this one's fill? So I think you have this cycle of understanding, kinda when do we start our next model because we know it's gonna take, like, nth length of time, and then what significant improvements can we make because, like, it does seem that, like, once that model train starts, it's not easy to stop it, start over again, or, like, implement new datasets.So we'll probably see some innovation in this space, like the ability to, like, you know, they call it, like, steering essentially, like, while a model is doing its work, kind of implement or aggregate new information. But, yeah, I think the models are getting heavier. The times to develop them are getting longer, and so you're starting to see, you know, maybe some lulls in the in that time period before, like, you know, one of these models comes out and announces something large. But I think, you know, just to wrap up on Google really quickly today, and I don't know if that was in our news stack, but I think it's Google's developer conference this week, but they just announced two new TPUs. Sam, I don't know if you saw that.If you can pull up the article, it'll be great. It's on top of TechCrunch and everywhere else, more tech meme. But, yeah, they had launched two new TPUs. Elon tweeted that TPUs are underrated. As you recall, Jonathan Ross, who invented the TPU at Google, just sold Grok to NVIDIA.And so, you know, Google becomes a major inference player. I think they just signed another inference deal this morning. I had read some news about it, and I know they just signed, like, another inference deal with somebody else too. Yeah. There's just been it just you know, you can't count Google out.You can't bet against them. They got a ton of money. Obviously, it's led by really smart people, and they're gonna continue to develop and and build, you know, things to keep them in the race. You know, this is just one more week in, like, the tech space. You know?And this isn't any anything special. It's just literally, like, one more week. Great episode. Really fun to chat, and super super excited to to next week's episode. I think we got some guests coming up over the next few weeks.Sure if next week or the following, but know I we got some great founders coming on board from some great AI companies. Yeah, see you all next week. This was really awesome. Bye bye.