I used it today to take a look at my previously built design system with Logos, branding, fonts, and everything else. After a lot of annoying tweaking back and forth, finally, I got something that was satisfactory.
Then I looked at the usage and it said I had used 95% of my Claude design usage for the week!
This isn't a real tool. This is a plaything, if that's what they're providing as examples.
I used Claude Design to see how it'd spit out a design I already had been working on for some weeks, given a dense enough prompt and a decent requirements document (I did not feed it visuals). I thought the output was pretty good! It didn't match the style we're after at all but it did do some logical content grouping and made some IA decisions I decided to pull into my own explorations. Overall I left with a good impression.
And then I was scrolling Twitter, and saw someone else post their own "success story" and the design was nearly identical to the mock up Claude Design made for me. Lol. The homogenization problem will continue to plague tools like these to some degree, much in the same way AI generated text or code or imagery has a sort of homogenous tone or feel to it.
> I used it today to take a look at my previously built design system with Logos, branding, fonts, and everything else.
The fact that you are using this language tells me you are probably more advanced than the average individual, and likely have higher expectations.
My sister-in-law has a small apparel company. She’s developed quite a bit of skill over the past six years but she really struggled at the start. She had great ideas, but translating them to something she could apply was frustrating. *Anything* that could have helped her there would have been worth a look.
Funny. My read on that language was this person has absolutely no idea what a truly robust and scalable design system and component library actually are, particularly within the scope of a successful business. Well built ones serve every facet of the organization, not just the product.
• Claude Design uses Opus 4.7, which is more expensive than earlier models.
• It's just Day 2; it's not a finished product. It's ridiculous how quickly Anthropic iterates.
• If you've been using Claude for a while, Design already knows your style and preferences. You'd have to start from scratch using a different AI design tool. I don’t doubt that'll pay dividends in the long run.
> It will never be cheaper than what it is today. Anthropic is heavily subsidizing.
We don't know that for sure—they've dropped prices before:
1. Claude 3 → Claude 3.5/3.7 generation (mid-2024 to early 2025): Haiku went from $0.25/$1.25 to $0.80/$4.00 per MTok — this was actually a price increase for Haiku, but Sonnet stayed flat at $3/$15 while delivering significantly better performance, effectively a price-per-capability reduction.
2. Claude 3/4 Opus → Claude Opus 4.5/4.6 (late 2025): This was the big one. Opus dropped from $15/$75 per MTok down to $5/$25 per MTok — a 67% reduction on input and output. This is the most significant explicit price cut Anthropic has made, delivering a far more capable model at one-third the price.
I had a similar experience with running out of usage quite quickly, after setting up one design system properly, and then getting pretty close with a second one. But it's a research preview - I'm sure it will change.
I was quite happy with what I pulled off using the first design system: I wanted a new footer section for my IPAAS startup, it generated four options, the fourth of which was quite good. We iterated on it for a bit, then I pulled it into Claude Code (that integrated feature is very cool), CC built it, I deployed it, done. (Bottom section of https://tediware.com/ if you're interested, the bit with "Origin story" on the left and the signup panel on the right).
It was not a complicated build by any means but I liked the concept it developed and it was dead-easy to make it all happen. I think the ideas in the UI are very good. Still rough, but you can see where this could go, and it's got a ton of potential.
I mean, it's fine and serves it's purpose, but I'm a bit confused what you are getting that you wouldn't get with the millions of pre-made designs and design systems? Like Tailwind UI for example.
I find that with the ubiquity of Tailwind, developers treat design as a "solved problem". What's missing is the specific evolution of one's product and the resultant information architecture. The sibling response is my experience as well, design is an incredibly interactive exercise.
Granted, not every component on every surface will need this amount of scrutiny. But I'm usually the outlier developer warning teammates that design is not a solved problem. Granted, there's a huge difference between an existing app and its evolution and throwing a nextjs landing page up in search of any life.
Even with bootstrap, design was a solved problem. What you bring with a UI designer is appeal (aka make thing pretty and enjoyable). If you want utilitarian, even the old x11 toolkit like Athena, Win 98 era widgets would do the part.
Iterative experience (experimenting with different ideas, deciding what works best) and speed of execution (once I was happy with it, making it happen required almost no work).
Yes. Even without Claude design and just Claude code, it can use existing design and build out new mockups in-app, which is much easier to demo , tweak and then implement the backend (if any) - all through Claude Code (or Codex if you prefer that). We use both and are now leaning more towards Codex over Claude
It produced great results for me, in 10 mins, and then my usage was blown and now I have to wait a week. It did let me export the ZIP, though. I tried throwing the contents of the ZIP into Stitch With Google, but it didn't work very well.
Yup it's based off their playground so plaything is the right word.
It's a wrapper around that. I definitely appreciate the better design output from Claude code but it has a ways to go before it can replace serious design contenders.
It's in research preview. I suspect limits are low on purpose. FWIW, I gave it twelve screenshots of different pages in my app and it did a really excellent job fixing them up. Consumed just 40% of weekly quota - still too high but it's probably a YMMV situation.
I don’t really buy that Claude Design will remove all the complexity around design. Vibe-coded apps using Claude look simpler because they are simpler. They’re not a gigantic product suite with extremely specific UI components tailored to each use case. The ‘simplicity’ is an illusion coming from conflating the complexity of a bicycle (a vibe coded app) with an airplane (an app like Figma).
Building the same design system component in code versus in Figma is going to be slightly more succinct in code; Figma’s primitives don’t have the sort of conditionals and control flow that code has. But code is much less malleable than drawing on a screen, and creative freedom is harder to achieve in code.
UI can fix the gap where code feels less malleable than Figma, but complexity comes largely from the worlds that humans create, and humans apparently want to create 8 modes for 4 products and 2 light/dark modes. If you want the same setup in Claude, it’ll be a little easier to maintain, but not much less complex.
I admit I'm having a visceral reaction to this analogy. A bicycle is a sophisticated product whose form is almost pure function. Despite being apparenty simple, almost no regular person can draw even a reasonable facsimile of a bicycle from memory ( https://www.youtube.com/watch?v=L0_vXZ-3LFU ). Which is to say, for actually designing a functioning bicycle, the devil is in the details, and details are exactly where vibecoded apps fall down. Our lower bound for this analogy should instead be the downhill go-kart cobbled together from scrap wood you found in the dumpster.
Figma has been in trouble for a while. All the designers at my company switched to Cursor nearly a year ago. They made live mockups that don’t even need a spec to implement, because the expected behavior is already captured in the prototype. Claude Design makes it just marginally easier.
Stitch has been around for a few months from hole and it does a better job than this. I bet designers are in the honeymoon phase of people don’t know this exists and this does my whole job phase.
Not to mention all the people hiring UX just because they don't want to deal with it themselves, not because they need something that requires a lot of skill.
So let me get this straight (Pretend I'm 50, a developer since childhood, but I can't CSS to save my life) are there shops where developers, even front end developers, have to talk to designers who are't just sketching an idea for a logo or landing page, but designers who run this Figma thing and maintain the entire products "design" in some "style database"? And the idea is that these designers - who aren't developers - should be able to tweak the look of things without changing code? Or is it usually just the front end devs that run this Figma thing, but they dislike the disconnect between it and their code?
lol yes. At least in agency world, a common approach in the last X years has been that designers create entire pixel-perfect, component-based sources-of-truth in Figma (which evolve! they aren't delivered static and complete) -- these are also what the client sees and approves, or at the very least they see branded deck slides that incorporate the Figma designs. Anyways, front end then re-implements from Figma into CSS, except it's usually best-approximation (not pixel-perfect) partially because, despite Figma allowing you to "copy CSS" for an element, it's unusable, almost inline CSS (and usually not aware of its ascendents and descendents, or any variables you're maintaining in CSS, or any class hierarchies, etc), and partially because the units of measurement aren't always identical on either side. You'll also often have multiple FE devs recreating components independently of each other (as a team effort), which can lead to drift and different implementations, which is fun. Then, depending upon the tech stack, FE might be building these components in something like Storybook [0] as a "front end source of truth", which then are either directly injected into a React or NextJS app or whatever, or sometimes they're partially or fully re-implemented again into BE components in the CMS (ex. Sitefinity). Then people ask which one is the source of truth, but really it's a chain of sources of truth that looks more like the telephone game than a canonical "brand bible". Then throw in any out-of-the-box future client efforts (say, a promotional landing page hosted outside of the main project) and you may have yet another reimplementation of part of the same design, but in a completely different system.
Yah you get this inner platform effect where designers start unwittingly creating their own version of css using Figma and it gets really bespoke really fast.
@kevinsync's answer is 100% correct and it's been this way for the last ~~~20? years? at least - only it was "Photoshop files hold the (design) truth" before - now it's figma.
But yes, the "design to code" gap has always been where designers' intentions were butchered and/or where frontend developers would discover/have to deal with designs that didn't take into account that some strings need more space, or what to do when there are more or less elements in a component, how things should scroll in real life, how things should react to a variety of screen sizes, etc.
this short meme video is funny/not funny because it hits too close to home - https://www.youtube.com/shorts/r6JXc4zfWw4 - but yes, "designers don't code and developers don't design", roughly speaking
and then of course you meet some people who do both really well... but they are pretty rare. :-)
It's a real shame that people bought into this false dichotomy, because the base reality is that people who work in web dev that stubbornly pick either code or layout are more of a liability than an asset.
I don't believe that people who can design and code are as rare as folks seem to believe, either. What seems more likely is that there are a LOT of coders who are extremely fluent in CSS but aren't particularly gifted when it comes to making things look good.
It wasn't that long ago that designers understood that they couldn't just hand off a 2D comp of what they want to see. The job isn't done until the output can be integrated into the app. Nobody gets to launch cows over the wall and go for lunch.
Nah yeah Photoshop .PSD's were totally normal for website designs. I got extremely proficient at building functioning websites based on PSD files, going back as far as the days of using nested <table> structures with 1x1 transparent spacer.gif images :) I built hundreds of websites from .PSD files, and Fireworks was pretty much non-existent in my experience.
Sure, and also Illustrator sometimes, and Photoshop at other times. Some of the designers I know (very famous for their ui/web work) never touched vector components and just had a ton of layers in Photoshop and air/paintbrushed everything. Hence the meme...
I wouldn't even classify it as a mistake, just a difference in experience regardless of what Adobe's intentions were.
As someone who has done front-end development for both web and mobile devices for a very long time in the pre-Figma days I was handed a lot more designs that were mocked up in Photoshop than Fireworks.
Yes, Figma is the de facto standard for how designers hand off a UI design to developers in large (and some small) companies.
It's kind of horrible, but I guess it's better than previous alternatives. But, it's not better than a tool that works with code directly and mostly automates away the tedium or translating a visual design into code. I haven't tried Claude Design, but I know I don't find Figma enjoyable (but I'm not much of a designer...I'm more comfortable with code than with pages and pages of options in a GUI).
I'm not sure about "without changing code" but I have definitely seen the believe that Figma represents something authoritative about the product instead of, say, the product being authoritative for itself.
Perhaps because I have a similar bio to yours, I am allergic to this view.
UX designers I encountered have mostly been tasked on ensuring consistency across the various product (A lot of devs are very cavalier about spacing and font sizes). Sometimes they proposed new flows and layouts, especially when the product needs a coat of paint.
So tools like Figma is nice in that regards as it's simpler to iterate on (From simple to hardest: Sketch on whiteboard|paper, Wireframe tools like Balsimiq, Figma|Sketch, css code) because it's pure fiddling with various properties. Figma has direct feedback while the code may require a compilation phase.
As someone who spent quite some times these days to reverse the figma protocol[0] I can't agree more with
> Figma accidentally excluded themselves from the training data that would have made them relevant in the agentic era.
Their binary format is so much of a "let's reinvent everything" which I think come from the fact it's a tool you can use for web design, android app design, ios design and anything-you-want design that it became a jack of all trade and so mapping it to web is not a perfect 1:1 translation.
And for being useful to agent, any people who got to implement the figma from a UX guy know that even human can't know truly the intent of most figma design, so how a LLM could ? Common source of question that even the UX guy has no answer for:
1. Ok this button looks great, but in German how will it look ?
2. Oh and actually this button does not look great when i put in CSS, it wraps on two line, you cheated again with the letter spacing, did you ?
3. How does it look on a phone that is not an iphone ?
4. You know that doing a border with a gradien is not possible in CSS, so what should i put ?
5. How does it look on a 4k screen ?
6. etc.
I know that most of these question can be answered by props and autolayout, and I've been asking the 5 question above these days on a figma that had these but it's just that the UX guy is not that mythical beast that "know-how-to-use-figma-right"
So I can't wait for these tools that are html behind to catch up, even more if we can have the prompt with it. (As a developer I never got to see the prompt the product manager made to the UX guy)
Great article, the last couple of paragraphs made me laugh! I love the part about things not masquerading as something else and being honest about what they are.
I was wondering if PenPot (https://penpot.app) might be sitting pretty in this new agentic era, considering that they took the direction of designs being actual markup, unlike the canvas approach in fig - if that’s even something that interests them.
I got deeper into Figma than I ever had to before yesterday.
It turns out there’s no way to use Math in variables built in to the product. The most common plugin is 99 dollars. To add what is obviously mustache JS. To browser based software. It’s not good.
This design tool space died a long time ago for me when InVision shut down and pivoted to a digital whiteboard. It’s a really difficult space.
But the fundamental problem is that it’s hard to get a design system right long-term, especially because it’s so intertwined with your code and whatever component library you use, which is a layer your designer will never touch. I don’t really see Claude Design fixing the fundamental Storybook hell of designing reusable and pretty components and layouts, but I don’t see Figma or any other tool solving it either.
What’s the solution? It feels like something that needs to be fixed more deeply at the component level.
The solution is an open, flexible, scriptable and drawable canvas where design and code co-exist in exact harmony. Design changes directly modify front-end code, and front-end code directly modifies design equivalently. I see the endgame as a model where designers and FEE's are co-owners and co-authors of the front-end with zero handoff.
What if the approach isn't reusable, but instead is rebuildable? We are stuck in the mindset of creating components that we can grab and plug in to new designs. When we have a component that we like, why not ask the tool to create a markdown definition of it. Later on, when we're doing a new design where we would like to reuse that component, we tell the tool to read the markdown and use that whenever they need to use that component. I think the future will be much more flexible and interesting.
I don’t know what level of complexity you’ve seen in your software buildouts, but at anything “enterprise” scale, building your own components from scratch is a recipe for absolute disaster. There are so many nuances, especially with accessibility, and edge case bugs, that I would really strongly recommend against it. And by extension, I’d be against this approach.
Maybe you could make it work if everyone agrees on a base set of headless components to use, but we seem to be moving in the opposite direction with things like ShadCN.
> I don’t really see Claude Design fixing the fundamental Storybook hell of designing reusable and pretty components and layouts
FWIW Claude Code is decent at scaffolding those out if you have a good set of examples for it to work from.
But the argument is that is unneeded as we move forward as making changes and extracting things and such becomes basically "free". I'm not so convinced, but I do see the argument.
I tried Claude Design yesterday and fed it the UI of an app I am working on and what I had created already. I asked it to mock up 6 new "layout skins" using different colors, different icon and button placements, different vibes and it churned out 6 good designs with hi-res images plus Claude Code instructions and the ZIP file to boot. Didn't hit the usage, took about an hour. Did a really good overall and I liked the process and thinking structure and question asking.
Stitch has way more usage allowance, and it has been absolutely fantastic for some things, but it can get trapped in weird places where whatever you tell it to do it thinks it did it, but it didn't. It's like that Family Guy blinds meme:
I've worked on design tools for the last few years.
This article is a fundamental misunderstanding of both the design space, and figma as a company.
Just a few of my thoughts:
- Figma was always about building a successful company over a successful product. Figma started with a much more ambitious aspiration, and had the ability to deliver through talent like Evan Wallace. A lot of it started with showing how capable webgl was in the browser. And yet, a lot of things like 3d features don't exist because they had the awareness to really hone in and focus on building a specific thing that made them money because everybody in the company ended up with an expensive seat price.
- Seriously, Figma is a company that's about design tool second, and about getting a product that businesses use first. To that end, it's already succeeded through the IPO, subsequently, who knows what the market is going to look like. Figma having a war chest is in many ways much better than having a technically impressive demo that might evaporate.
- People at Figma, 100%, know everything in this article. And not just figma people, like anybody and everybody that's tried to build a design tool has had these thoughts. It's very obvious that ui/ux is the interesection of design/dev/pm. It's also very obvious that it should stick close to the source of truth, to something like code.
- The problem is, that it's almost underselling it to say that it's MASSIVE challenge to execute on these ideas because of how easily it bleeds into building not just a design tool, but a coding, data management, architecture, etc. tool
- I could talk at length about all the challenges and potential solutions, but that's neither here nor there.
- On AI, I guess other people's guess is as good as mine, but my gut feeling is that while data is important, SOTA AI is generalist enough that the base models, the thinking they're able to do, is better than having a lot of custom data. Especially because ui design is front-facing - you can just scour the web in contrast to private financial documents, or legal documents for example.
I tried yesterday for about an hour to have Claude design make me a simple logo (just the symbol) and didn’t get anywhere good. I’m sure for certain things like UI it’s great, and so is just Claude code, but this Claude Design thing very much to me feels like a demo and not a product. Maybe one day!
Claude Design is just a big opinionated prompt: https://www.lobsterpack.com/blog/claude-design-trenchcoat/. Among other things, it knows it isn't that great at drawing SVGs, so it won't try unless you force it to. For a logo, try drawing it with vanilla Claude Code as if it was a separate project: ask a "design agency" to ask you questions, answer them, then make a detailed brief of what you want to draw, then make it output an exact plan of what and where will it draw, lastly ask the chatbot to do the actual drawing using sub-agents for drawing individual components. Also add a "render it to a raster and make sure it looks right" step as well.
Some good points, but as a whole - I'm not sure if I agree. Sketch lost to Figma because of it's design tooling & multiplayer. Physical products still get designed before being constructed - I don't see that going away. If anything, I think Figma should stop trying to play both sides of the field and decide what it wants to be.
> "To Figma: I can see a world where this post does numbers in the Figma internal Slack. If that’s the case and you’re reading this from Figma: this wouldn’t have happened if you hired me last year when I was interviewing. Your loss, big dawg."
What a counterproductive way to end an otherwise good set of points. Gives the appearance of bitterness and a desire for vengeance.
The article makes a good point about how Figma's non-open data model is limiting their utility as the source of truth.
But I think it's part of a larger mistake Figma is making: they seem to have shifted to an extraction mindset too early, assuming they'd captured the market, right when the ground beneath them is starting to shift.
It's most visible in their pricing model evolution, which is now explicitly anti-collaboration. Figma used to be the obvious default because you could quickly share files with non-designers, so they could view and make small edits without fuss. Now that requires a paid "seat", along with a confusing mess of permission flows.
It's platform wide too. I taught a college design class recently, and had students sign up for Figma because it seemed archaic not to teach them to use it. Instead of just giving any ".edu" address a free account (like they used to) students are forced through a 3rd-party process of uploading transcripts to prove education status. A few of my students got rejected or ran into confusing errors, and never got access… Now I have to re-evaluate whether its worth using when teaching the class again. (And this is for a population with near-zero short-term purchasing power, but huge potential long-term value… why add barriers?)
This is such a weird self-inflicted wound for a collaboration platform to make. The big tools that won on collaboration (eg. Google Docs, GitHub) have understood that low-friction sharing is critical to becoming the default choice. And that being the default is a flywheel that drives adoption, both in users and in tooling.
It makes more sense if you see it through the lens of Figma trying to juice short-term numbers for their IPO. But it's sad to see because it had so much long-term potential.
I miss the days of having a native desktop design app with a perpetual license.
What Figma achieved technically in the 2010s was amazing. Coded the app in C++ and then used WASM to deliver it as a multiplayer web app.
But now it's trying to be too many things. Why did they ever feel the need to add slides and this other stuff.
Their MCP is poor (sure, they'll improve it).
The app struggles with larger files and performance is sloppy.
And don't get me started trying to design data grid heavy apps.
And they could easily follow Adobe's lead. Enshittify and lock you out of your account whenever they feel it's necessary (remember what happened with Venezuelan Adobe users a few years ago?)
Either Penpot gets their act together and will become the opensource design canvas for open-weight AI models or we will see another open source solution that will fill this space.
> I miss the days of having a native desktop design app with a perpetual license.
You can go that route with Affinity Designer [1], owned by Canva, who partnered with Anthropic on Claude Design [2]:
We’ve loved collaborating with Anthropic over the past couple of years and share a deep focus on making complex things simple. At Canva, our mission has always been to empower the world to design, and that means bringing Canva to wherever ideas begin. We’re excited to build on our collaboration with Claude, making it seamless for people to bring ideas and drafts from Claude Design into Canva, where they instantly become fully editable and collaborative designs ready to refine, share, and publish.
I have it, jumped on the affinity band wagon years ago after Adobe started their enshittification process.
After Canva bought Affinity, you now have to authenticate with your email from time to time when you launch the desktop app. Annoying and why do they do that?
React, Nextjs, Vue, Nuxt, and Angular are pretty much what AI is the very best at coding in my experience. Probably because they are all meant to build essentially the same thing with different curtains.
Fashion designers can, in general, work with fabric, yes. And an interior designer should probably have some idea of how to paint at the very least. To me with web design so much of what matters is encoded in the CSS and HTML that it is the final design product. Anything produced before is a sketch, a concept, but it's not a design.
For example designers and developers both use the computer as their primary medium of working. Their outputs resemble each other very closely, despite having a different underlying form.
Contrast that to the interior designer building a house, well those are different mediums. There is no efficiency gain from the interior designer designing the plan and also implementing it. Where as with a designer working in code there is one.
Fashion designers do indeed make clothing by hand, it's a very important part of their craft. This example disproves your stance.
I've been a developer for over 2 decades and I've been using AI in our react codebase for the past 3 months. Outside of some optimizations there's not much a designer couldn't debug through Claude Code. 90% of the industry is toast.
I want to be wrong because I'm watching the death of my entire career, but everything I've seen is pointing to this as an inevitability. We are shipping better and more secure code, and doing it easily twice as fast. Many development teams can be cut in half today with no reduction in output. I don't want to say it out loud at work yet, but we're actually producing too much.
I've been writing code for 50 years and it looks now that we have seen sunrise and are about to see sunset on humans writing code by hand.
Is that bad? Not to anyone who has managed dev teams and is familiar with the incredibly tortuous and painful business of trying to corral a bunch of humans with varying skill and enthusiasm levels to create software. We have tied ourselves in knots with things like Agile just trying to work around the fact that software development is so slow and arduous.
Many times back in the waterfall days I have written up design documents to kick off dev teams on multi-week or month projects. Now I could feed those into Claude Code and get results in days. This stuff is exciting beyond belief in just getting shit done.
This is a golden era for any established company with an existing customer base. My question to them would be "with Claude Code, why aren't you carving through that massive backlog of feature requests that has been building up over the years?".
A lot of people seem to look at this as job threatening, and it surely is for junior devs. But for companies that already have a strong senior talent bench, it's time to raise the ambition levels and ask not how many jobs can be shed, but instead just how fast and hard can we go now we have these new superpowers.
This is so context-dependent. Coding some generic crud app is indeed becoming more automated, but most of the stuff I end up building is just way outside the capabilities of current LLMs to accomplish without significant steering and gasps hand written code. Most of the stuff LLMs are good at ones hotting are the same things a non-coder could build with no-code platforms anyways. Which is great imo, it means we can utilise our skills and expertise on stuff that is more "cutting edge".
AI is impressive but this same sea-change happened at least twice before - the era when computers went from being rooms full of women(354) to machines programmed in machine language(892) to those with screens, keyboards and even assemblers (assembly language, especially macro assemblers, were considered seriously high level at a time), to mid-level languages like C (considered needlessly complex and slow at one time, now considered barely above a macro assembler), to high-level languages like Java and even higher ones (arguably) like Rust.
Every one of those transitions has resulted in more programmers - though not necessarily the same programmers.
But things really are different this time. Computers and software were nascent industries with lots of room to grow, lots of software to build in previous transitions. Today software and technology companies are the biggest in the world. Every industry uses software. Getting your web app, mobile app or game discovered is actually a huge problem today because we have so much software. There is not infinite demand for software, or for anything else, even if it seems that way in the early days.
The Olde Days were custom code for a business doing business things (only they could afford a computer, and some universities), then there was a Cambrian explosion of software sold by individuals or small developers to many (80s and 90s), then we've moved to large companies and SaaS.
I think we're about to cycle back to "custom code" except now it's for everyone, by AI - you don't need to find the to-do app of your dreams; you can code one for yourself in a fever-dream.
The era of "write Wolfenstein 3D in a few months and make millions" are gone, but they've been gone a long time already.
Or - there was a HN discussion on this half year or so ago - there's consolidation again, and there will be AI, but no code. Domain expert talks to the AI, perhaps with an expertised intermediary. AI spins up a whole new 'software platform' for the customer.. internally. Offers all the UI that is needed to work with it.. still 'in the cloud' i.e. in AI data centers. Customer happy, devs less happy.
AI isn't a technology that replaces programmers, it's a technology that replaces generic human beings. The manager of your agents will be an agent, too.
Realistically this doesn't mean all pure designers go away. Large orgs can have a small team that set the overall style guide and designs important pages, and the rest of the org just follows using AI to iterate.
> Outside of some optimizations there's not much a designer couldn't debug through Claude Code. 90% of the industry is toast.
I've seen the "debugging" and "coding" that non-coding designers are attempting to vibe-code. 90% industry is definitely toast, but not the 90% you're thinking of. Most industry is going the way of Microsoft that cannot even display a start menu in under a second
>A leading question is a query that suggests the desired answer or puts words into a witness's mouth, often guiding them toward a "yes" or "no" response.
All right, here's a statement: your designers won't even know when the code is wrong. Just because it compiles it doesn't mean it's fine. They lack code judgment in the same way your coders lack design judgment.
In response I suggest that the engineers using AI also lack code judgement (because they are not reading it either). I don't think questioning the AI use is the actual topic here, it is the shifting roles. Who says it's the designers that are taking the new meta-role? It's probably the FE's honestly.
The role shifting doesn't mean that it's the best path forward. I'm simply stating that it is happening.
As we know from THERAC-25, etc., comprehensively verifying that code works the way it's expected to is not actually very easy - it's perhaps one of the hardest parts of building any system more complex than a toaster.
Your designers are going to be looking at the layout; they're not going to notice if it's slow, uses too much memory, is not maintainable, doesn't follow repo patterns, etc.
Do you think it is reasonable to expect a person with an arts degree to know this?
> Do you think it is reasonable to expect a person with an arts degree
Of course it is.
The only people who think your fucking college degree determines your knowledge level and ability are teenagers and people who are so deeply untalented that it’s the only way they feel qualified.
When you can control the model layer like Anthropic, you get more leverage over the traits of the persona, enough so that the system feels closer to havin consistent expert design judgment built-in that complements the 'truth-to-materials'.
Excellent post. I share the author's sentiment which is essentially "to hell with Figma, at least fix Sketch". Been feeling very lonely in may hatred towards Figma, which is for a whole bunch of reasons (among others, it's an incredibly shitty, memory and CPU hungry Electron app that looks and feels worse than any more or less well designed web site), but now after reading this I realize the number of reasons has doubled.
It may look like a crappy Electron app, but Figma has a quite interesting architecture. The browser editor is developed in C++ and cross-compiled to JavaScript with emscripten. The rendering engine looks like its handling HTML, but it's actually rendering their own document format for cross-browser consistency. They have their own CRDT implementation to handle multi-user edits.
I think my biggest question is who cares? What does having an interesting internal architecture have to do with the “its electron though” ideological attack.
It is made to perform much better than your typical electron app would. Saying electron-based == shitty is complete misunderstanding of the technology. Although i dislike Figma as much as the next guy, their app was in many ways very impressive. See Figma's cofounder old articles at https://madebyevan.com/figma/
(author of the post here) I cut a paragraph how Figma costs cuckoo bananas money for your entire team for the privilege of enduring this byzantine nightmare. And they paywall certain features, which you likely can't get authorization for, so you have to do more hacks on top of hacks on top of the “gold standard” practices I shared in the blog post. The price ramp is not gradual.
man, I dont even use Figma for personal & side projects because its so expensive. I still occasionally fire up sketch or freehand it.
Figma is a work tool only and I'm disappointed by its MCP tooling which feels late and behind where it should be, I just feel forced to use Figma Make which stays in their walled garden without practical utility and connections to my actual codebases
I was making a new brand file in Figma and was about to define all the variables and aliases, paths, all that. Took one look at it and just made the 8 colors I needed and went on with the design.
Honestly, I never understood the move to create an artificial dichotomy between design and code with a heavy layer of tooling.
I suppose that a layout engine made sense in the context of Flash, and you saw the future of the web as a set of keyframe animations. But the notion that there's a lot of value in creating a very heavy, high-friction abstraction between the UI/UX and the platform it ultimately runs on was always going to be a loser.
In the end, it turns out we're all just web developers, regardless of your weapon of choice.
I promise I am fine lol, I actually had a really productive chat with their panel last year and found it wasn’t the right role for me anyway for some logistical reasons.
I feel like we're leaning way too much into the "vibe" aspect of using LLMs at the moment. There is definitely a good use case for LLM's here, but is just prompting your way into a design really the best method here? I feel like something in between Figma and Claude Design would give designers the control they want, but still removing the friction of going between design files and the code impl.
Oh no. Figma has variables and instances and it's hard to debug bad colors.
And here I am with Claude Code... That so far generated a 2000-line CSS file for a 7000-line app consisting of literally three web pages [1]. Where almost every single color, component, class and style is duplicated at least two times. Where custom classes are fighting with Tailwind classes (yes, there's also Tailwind ON TOP of custom CSS) that are fighting with inline hardcoded style= declarations.
Figma is definitely going to suffer the vibe-coded design slop-app from Anthropic.
[1] 7k lines are almost justified for the functionality in them, and I tried to keep an eye on the code. It's harder to keep an eye on CSS
Have had similar dramas with CSS. Opus/Sonnet seem to have far less taste and discipline in writing CSS - constantly they are trying to slam in changes with !important or inline styles rather than think through how to do it properly.
We can blame AI for slopcoding but it's just what we've been doing for decades now - who cares if it takes 9000 lines to do what in theory Linus and Knuth and Carmack could do in 3000? Our CPUs eat a million lines of code for breakfast and demand more.
I used it today to take a look at my previously built design system with Logos, branding, fonts, and everything else. After a lot of annoying tweaking back and forth, finally, I got something that was satisfactory.
Then I looked at the usage and it said I had used 95% of my Claude design usage for the week!
This isn't a real tool. This is a plaything, if that's what they're providing as examples.
I used Claude Design to see how it'd spit out a design I already had been working on for some weeks, given a dense enough prompt and a decent requirements document (I did not feed it visuals). I thought the output was pretty good! It didn't match the style we're after at all but it did do some logical content grouping and made some IA decisions I decided to pull into my own explorations. Overall I left with a good impression.
And then I was scrolling Twitter, and saw someone else post their own "success story" and the design was nearly identical to the mock up Claude Design made for me. Lol. The homogenization problem will continue to plague tools like these to some degree, much in the same way AI generated text or code or imagery has a sort of homogenous tone or feel to it.
> I used it today to take a look at my previously built design system with Logos, branding, fonts, and everything else.
The fact that you are using this language tells me you are probably more advanced than the average individual, and likely have higher expectations.
My sister-in-law has a small apparel company. She’s developed quite a bit of skill over the past six years but she really struggled at the start. She had great ideas, but translating them to something she could apply was frustrating. *Anything* that could have helped her there would have been worth a look.
Funny. My read on that language was this person has absolutely no idea what a truly robust and scalable design system and component library actually are, particularly within the scope of a successful business. Well built ones serve every facet of the organization, not just the product.
Things to keep in mind:
• Claude Design uses Opus 4.7, which is more expensive than earlier models.
• It's just Day 2; it's not a finished product. It's ridiculous how quickly Anthropic iterates.
• If you've been using Claude for a while, Design already knows your style and preferences. You'd have to start from scratch using a different AI design tool. I don’t doubt that'll pay dividends in the long run.
They can iterate fast, because their devs and only their devs have access to the best Claude Code on the planet.
It will never be cheaper than what it is today. Anthropic is heavily subsidizing.
> It will never be cheaper than what it is today. Anthropic is heavily subsidizing.
We don't know that for sure—they've dropped prices before:
1. Claude 3 → Claude 3.5/3.7 generation (mid-2024 to early 2025): Haiku went from $0.25/$1.25 to $0.80/$4.00 per MTok — this was actually a price increase for Haiku, but Sonnet stayed flat at $3/$15 while delivering significantly better performance, effectively a price-per-capability reduction.
2. Claude 3/4 Opus → Claude Opus 4.5/4.6 (late 2025): This was the big one. Opus dropped from $15/$75 per MTok down to $5/$25 per MTok — a 67% reduction on input and output. This is the most significant explicit price cut Anthropic has made, delivering a far more capable model at one-third the price.
Be glad it's not Day 200: Opus models are only getting more expensive to use.
I had a similar experience with running out of usage quite quickly, after setting up one design system properly, and then getting pretty close with a second one. But it's a research preview - I'm sure it will change.
I was quite happy with what I pulled off using the first design system: I wanted a new footer section for my IPAAS startup, it generated four options, the fourth of which was quite good. We iterated on it for a bit, then I pulled it into Claude Code (that integrated feature is very cool), CC built it, I deployed it, done. (Bottom section of https://tediware.com/ if you're interested, the bit with "Origin story" on the left and the signup panel on the right).
It was not a complicated build by any means but I liked the concept it developed and it was dead-easy to make it all happen. I think the ideas in the UI are very good. Still rough, but you can see where this could go, and it's got a ton of potential.
I mean, it's fine and serves it's purpose, but I'm a bit confused what you are getting that you wouldn't get with the millions of pre-made designs and design systems? Like Tailwind UI for example.
https://tailwindcss.com/plus/ui-blocks
I find that with the ubiquity of Tailwind, developers treat design as a "solved problem". What's missing is the specific evolution of one's product and the resultant information architecture. The sibling response is my experience as well, design is an incredibly interactive exercise.
Granted, not every component on every surface will need this amount of scrutiny. But I'm usually the outlier developer warning teammates that design is not a solved problem. Granted, there's a huge difference between an existing app and its evolution and throwing a nextjs landing page up in search of any life.
Even with bootstrap, design was a solved problem. What you bring with a UI designer is appeal (aka make thing pretty and enjoyable). If you want utilitarian, even the old x11 toolkit like Athena, Win 98 era widgets would do the part.
This is just completely false. But I have a feeling there's no way you're going to change your mind.
"make things pretty" would be a graphic designer or artist. Are you saying the entire arm of Product design is a made up value?
Iterative experience (experimenting with different ideas, deciding what works best) and speed of execution (once I was happy with it, making it happen required almost no work).
Thats fair. Could you have the same iterative experience with an LLM, but starting with a prebuilt base and iterating from there?
Yes. Even without Claude design and just Claude code, it can use existing design and build out new mockups in-app, which is much easier to demo , tweak and then implement the backend (if any) - all through Claude Code (or Codex if you prefer that). We use both and are now leaning more towards Codex over Claude
It produced great results for me, in 10 mins, and then my usage was blown and now I have to wait a week. It did let me export the ZIP, though. I tried throwing the contents of the ZIP into Stitch With Google, but it didn't work very well.
Yup it's based off their playground so plaything is the right word.
It's a wrapper around that. I definitely appreciate the better design output from Claude code but it has a ways to go before it can replace serious design contenders.
It's in research preview. I suspect limits are low on purpose. FWIW, I gave it twelve screenshots of different pages in my app and it did a really excellent job fixing them up. Consumed just 40% of weekly quota - still too high but it's probably a YMMV situation.
I don’t really buy that Claude Design will remove all the complexity around design. Vibe-coded apps using Claude look simpler because they are simpler. They’re not a gigantic product suite with extremely specific UI components tailored to each use case. The ‘simplicity’ is an illusion coming from conflating the complexity of a bicycle (a vibe coded app) with an airplane (an app like Figma).
Building the same design system component in code versus in Figma is going to be slightly more succinct in code; Figma’s primitives don’t have the sort of conditionals and control flow that code has. But code is much less malleable than drawing on a screen, and creative freedom is harder to achieve in code.
UI can fix the gap where code feels less malleable than Figma, but complexity comes largely from the worlds that humans create, and humans apparently want to create 8 modes for 4 products and 2 light/dark modes. If you want the same setup in Claude, it’ll be a little easier to maintain, but not much less complex.
Most of the times people just want a bike or a car. Not everybody needs an airplane. This is going to hit Figma very hard.
I admit I'm having a visceral reaction to this analogy. A bicycle is a sophisticated product whose form is almost pure function. Despite being apparenty simple, almost no regular person can draw even a reasonable facsimile of a bicycle from memory ( https://www.youtube.com/watch?v=L0_vXZ-3LFU ). Which is to say, for actually designing a functioning bicycle, the devil is in the details, and details are exactly where vibecoded apps fall down. Our lower bound for this analogy should instead be the downhill go-kart cobbled together from scrap wood you found in the dumpster.
> Most of the times people just want a bike
with a pelican on it
that runs on a local model and looks better than Opus
Figma has been in trouble for a while. All the designers at my company switched to Cursor nearly a year ago. They made live mockups that don’t even need a spec to implement, because the expected behavior is already captured in the prototype. Claude Design makes it just marginally easier.
The people that want just a bicycle wasn't going to buy figma
Stitch has been around for a few months from hole and it does a better job than this. I bet designers are in the honeymoon phase of people don’t know this exists and this does my whole job phase.
Not to mention all the people hiring UX just because they don't want to deal with it themselves, not because they need something that requires a lot of skill.
So let me get this straight (Pretend I'm 50, a developer since childhood, but I can't CSS to save my life) are there shops where developers, even front end developers, have to talk to designers who are't just sketching an idea for a logo or landing page, but designers who run this Figma thing and maintain the entire products "design" in some "style database"? And the idea is that these designers - who aren't developers - should be able to tweak the look of things without changing code? Or is it usually just the front end devs that run this Figma thing, but they dislike the disconnect between it and their code?
lol yes. At least in agency world, a common approach in the last X years has been that designers create entire pixel-perfect, component-based sources-of-truth in Figma (which evolve! they aren't delivered static and complete) -- these are also what the client sees and approves, or at the very least they see branded deck slides that incorporate the Figma designs. Anyways, front end then re-implements from Figma into CSS, except it's usually best-approximation (not pixel-perfect) partially because, despite Figma allowing you to "copy CSS" for an element, it's unusable, almost inline CSS (and usually not aware of its ascendents and descendents, or any variables you're maintaining in CSS, or any class hierarchies, etc), and partially because the units of measurement aren't always identical on either side. You'll also often have multiple FE devs recreating components independently of each other (as a team effort), which can lead to drift and different implementations, which is fun. Then, depending upon the tech stack, FE might be building these components in something like Storybook [0] as a "front end source of truth", which then are either directly injected into a React or NextJS app or whatever, or sometimes they're partially or fully re-implemented again into BE components in the CMS (ex. Sitefinity). Then people ask which one is the source of truth, but really it's a chain of sources of truth that looks more like the telephone game than a canonical "brand bible". Then throw in any out-of-the-box future client efforts (say, a promotional landing page hosted outside of the main project) and you may have yet another reimplementation of part of the same design, but in a completely different system.
[0] https://storybook.js.org
I've directly experienced this and it is roughly as sane and effective as it sounds.
We only have the Figma and Storybook layers (product not agency) but these two comments paint an accurate picture of the absurdity. Thank you!
This is exactly my experience of working in an agency. Made worse by Figma defaulting to 1440px so every design only really works at that width.
Why don’t we just teach the designers code / the coders design? This feels very Programmer-Analyst split
Yah you get this inner platform effect where designers start unwittingly creating their own version of css using Figma and it gets really bespoke really fast.
@kevinsync's answer is 100% correct and it's been this way for the last ~~~20? years? at least - only it was "Photoshop files hold the (design) truth" before - now it's figma.
But yes, the "design to code" gap has always been where designers' intentions were butchered and/or where frontend developers would discover/have to deal with designs that didn't take into account that some strings need more space, or what to do when there are more or less elements in a component, how things should scroll in real life, how things should react to a variety of screen sizes, etc.
this short meme video is funny/not funny because it hits too close to home - https://www.youtube.com/shorts/r6JXc4zfWw4 - but yes, "designers don't code and developers don't design", roughly speaking
and then of course you meet some people who do both really well... but they are pretty rare. :-)
It's a real shame that people bought into this false dichotomy, because the base reality is that people who work in web dev that stubbornly pick either code or layout are more of a liability than an asset.
I don't believe that people who can design and code are as rare as folks seem to believe, either. What seems more likely is that there are a LOT of coders who are extremely fluent in CSS but aren't particularly gifted when it comes to making things look good.
It wasn't that long ago that designers understood that they couldn't just hand off a 2D comp of what they want to see. The job isn't done until the output can be integrated into the app. Nobody gets to launch cows over the wall and go for lunch.
It’s just one more example of people realizing that the code is the spec.
> only it was "Photoshop files hold the (design) truth" before
You mean Fireworks. Photoshop was for graphic design. Web designers used Adobe Fireworks. Figma is a successor to Fireworks, not Photoshop.
Nah yeah Photoshop .PSD's were totally normal for website designs. I got extremely proficient at building functioning websites based on PSD files, going back as far as the days of using nested <table> structures with 1x1 transparent spacer.gif images :) I built hundreds of websites from .PSD files, and Fireworks was pretty much non-existent in my experience.
I was in web agencies since like 2002-2015, always got PSDs from either clients or internal designers
100%. I was always told to slice the PSD, fireworks never entered the conversation in the agency world I was a part of.
Sure, and also Illustrator sometimes, and Photoshop at other times. Some of the designers I know (very famous for their ui/web work) never touched vector components and just had a ton of layers in Photoshop and air/paintbrushed everything. Hence the meme...
I've even received designs in PowerPoint
Everyone used whatever they were familiar with regardless of the purpose of the application.
Many designers stuck with Photoshop sadly. Back when I did agency work it was absolutely normal to get PSDs of mockups.
I think his point was made regardless of his mistake
I wouldn't even classify it as a mistake, just a difference in experience regardless of what Adobe's intentions were.
As someone who has done front-end development for both web and mobile devices for a very long time in the pre-Figma days I was handed a lot more designs that were mocked up in Photoshop than Fireworks.
Yes, Figma is the de facto standard for how designers hand off a UI design to developers in large (and some small) companies.
It's kind of horrible, but I guess it's better than previous alternatives. But, it's not better than a tool that works with code directly and mostly automates away the tedium or translating a visual design into code. I haven't tried Claude Design, but I know I don't find Figma enjoyable (but I'm not much of a designer...I'm more comfortable with code than with pages and pages of options in a GUI).
I'm not sure about "without changing code" but I have definitely seen the believe that Figma represents something authoritative about the product instead of, say, the product being authoritative for itself.
Perhaps because I have a similar bio to yours, I am allergic to this view.
In a bigger company, yes, as it's required to make sure every engineer implements everything in the same way without differences in style over time.
UX designers I encountered have mostly been tasked on ensuring consistency across the various product (A lot of devs are very cavalier about spacing and font sizes). Sometimes they proposed new flows and layouts, especially when the product needs a coat of paint.
So tools like Figma is nice in that regards as it's simpler to iterate on (From simple to hardest: Sketch on whiteboard|paper, Wireframe tools like Balsimiq, Figma|Sketch, css code) because it's pure fiddling with various properties. Figma has direct feedback while the code may require a compilation phase.
As someone who spent quite some times these days to reverse the figma protocol[0] I can't agree more with
> Figma accidentally excluded themselves from the training data that would have made them relevant in the agentic era.
Their binary format is so much of a "let's reinvent everything" which I think come from the fact it's a tool you can use for web design, android app design, ios design and anything-you-want design that it became a jack of all trade and so mapping it to web is not a perfect 1:1 translation.
And for being useful to agent, any people who got to implement the figma from a UX guy know that even human can't know truly the intent of most figma design, so how a LLM could ? Common source of question that even the UX guy has no answer for:
I know that most of these question can be answered by props and autolayout, and I've been asking the 5 question above these days on a figma that had these but it's just that the UX guy is not that mythical beast that "know-how-to-use-figma-right"So I can't wait for these tools that are html behind to catch up, even more if we can have the prompt with it. (As a developer I never got to see the prompt the product manager made to the UX guy)
[0]https://github.com/allan-simon/figma-kiwi-protocol
Great article, the last couple of paragraphs made me laugh! I love the part about things not masquerading as something else and being honest about what they are.
I was wondering if PenPot (https://penpot.app) might be sitting pretty in this new agentic era, considering that they took the direction of designs being actual markup, unlike the canvas approach in fig - if that’s even something that interests them.
Wasn't the whole Penpot UI itself made in SVG? Or did they change that along the way?
I got deeper into Figma than I ever had to before yesterday.
It turns out there’s no way to use Math in variables built in to the product. The most common plugin is 99 dollars. To add what is obviously mustache JS. To browser based software. It’s not good.
This design tool space died a long time ago for me when InVision shut down and pivoted to a digital whiteboard. It’s a really difficult space.
But the fundamental problem is that it’s hard to get a design system right long-term, especially because it’s so intertwined with your code and whatever component library you use, which is a layer your designer will never touch. I don’t really see Claude Design fixing the fundamental Storybook hell of designing reusable and pretty components and layouts, but I don’t see Figma or any other tool solving it either.
What’s the solution? It feels like something that needs to be fixed more deeply at the component level.
The solution is an open, flexible, scriptable and drawable canvas where design and code co-exist in exact harmony. Design changes directly modify front-end code, and front-end code directly modifies design equivalently. I see the endgame as a model where designers and FEE's are co-owners and co-authors of the front-end with zero handoff.
Dreamweaver?
Flash! Those were the days lol.
What if the approach isn't reusable, but instead is rebuildable? We are stuck in the mindset of creating components that we can grab and plug in to new designs. When we have a component that we like, why not ask the tool to create a markdown definition of it. Later on, when we're doing a new design where we would like to reuse that component, we tell the tool to read the markdown and use that whenever they need to use that component. I think the future will be much more flexible and interesting.
I don’t know what level of complexity you’ve seen in your software buildouts, but at anything “enterprise” scale, building your own components from scratch is a recipe for absolute disaster. There are so many nuances, especially with accessibility, and edge case bugs, that I would really strongly recommend against it. And by extension, I’d be against this approach.
Maybe you could make it work if everyone agrees on a base set of headless components to use, but we seem to be moving in the opposite direction with things like ShadCN.
> I don’t really see Claude Design fixing the fundamental Storybook hell of designing reusable and pretty components and layouts
FWIW Claude Code is decent at scaffolding those out if you have a good set of examples for it to work from.
But the argument is that is unneeded as we move forward as making changes and extracting things and such becomes basically "free". I'm not so convinced, but I do see the argument.
I tried Claude Design yesterday and fed it the UI of an app I am working on and what I had created already. I asked it to mock up 6 new "layout skins" using different colors, different icon and button placements, different vibes and it churned out 6 good designs with hi-res images plus Claude Code instructions and the ZIP file to boot. Didn't hit the usage, took about an hour. Did a really good overall and I liked the process and thinking structure and question asking.
How does this compare to Google's Stitch? As opposed to Anthropic's, it's free largely due to the largesse of Google.
https://stitch.withgoogle.com/
Stitch has way more usage allowance, and it has been absolutely fantastic for some things, but it can get trapped in weird places where whatever you tell it to do it thinks it did it, but it didn't. It's like that Family Guy blinds meme:
https://tenor.com/search/family-guy-blinds-gifs
Wow, that produced pretty bad results (app). Claude design is much more useful than this but I've only spent around 20 minutes with each.
I've worked on design tools for the last few years.
This article is a fundamental misunderstanding of both the design space, and figma as a company.
Just a few of my thoughts:
- Figma was always about building a successful company over a successful product. Figma started with a much more ambitious aspiration, and had the ability to deliver through talent like Evan Wallace. A lot of it started with showing how capable webgl was in the browser. And yet, a lot of things like 3d features don't exist because they had the awareness to really hone in and focus on building a specific thing that made them money because everybody in the company ended up with an expensive seat price.
- Seriously, Figma is a company that's about design tool second, and about getting a product that businesses use first. To that end, it's already succeeded through the IPO, subsequently, who knows what the market is going to look like. Figma having a war chest is in many ways much better than having a technically impressive demo that might evaporate.
- People at Figma, 100%, know everything in this article. And not just figma people, like anybody and everybody that's tried to build a design tool has had these thoughts. It's very obvious that ui/ux is the interesection of design/dev/pm. It's also very obvious that it should stick close to the source of truth, to something like code. - The problem is, that it's almost underselling it to say that it's MASSIVE challenge to execute on these ideas because of how easily it bleeds into building not just a design tool, but a coding, data management, architecture, etc. tool
- I could talk at length about all the challenges and potential solutions, but that's neither here nor there.
- On AI, I guess other people's guess is as good as mine, but my gut feeling is that while data is important, SOTA AI is generalist enough that the base models, the thinking they're able to do, is better than having a lot of custom data. Especially because ui design is front-facing - you can just scour the web in contrast to private financial documents, or legal documents for example.
I tried yesterday for about an hour to have Claude design make me a simple logo (just the symbol) and didn’t get anywhere good. I’m sure for certain things like UI it’s great, and so is just Claude code, but this Claude Design thing very much to me feels like a demo and not a product. Maybe one day!
Claude Design is just a big opinionated prompt: https://www.lobsterpack.com/blog/claude-design-trenchcoat/. Among other things, it knows it isn't that great at drawing SVGs, so it won't try unless you force it to. For a logo, try drawing it with vanilla Claude Code as if it was a separate project: ask a "design agency" to ask you questions, answer them, then make a detailed brief of what you want to draw, then make it output an exact plan of what and where will it draw, lastly ask the chatbot to do the actual drawing using sub-agents for drawing individual components. Also add a "render it to a raster and make sure it looks right" step as well.
Claude Design is very explicitly oriented towards product UI design. It's not trying to be a product that can make you a logo.
You should try the Arrow SVG model by Quiver, should be much better at that sort of thing since it's made for that.
Anthropic has no image generation models, right?
We've got an LLM using CSS and emojis and maybe pelicans riding bikes (SVGs).
Yes, they only do SVGs.
I'm actually glad they're focusing on code, and code adjacent tooling only.
Some good points, but as a whole - I'm not sure if I agree. Sketch lost to Figma because of it's design tooling & multiplayer. Physical products still get designed before being constructed - I don't see that going away. If anything, I think Figma should stop trying to play both sides of the field and decide what it wants to be.
Recent and related:
Claude Design - https://news.ycombinator.com/item?id=47806725 - April 2026 (732 comments)
> "To Figma: I can see a world where this post does numbers in the Figma internal Slack. If that’s the case and you’re reading this from Figma: this wouldn’t have happened if you hired me last year when I was interviewing. Your loss, big dawg."
What a counterproductive way to end an otherwise good set of points. Gives the appearance of bitterness and a desire for vengeance.
Do you genuinely think that was earnest?
The article makes a good point about how Figma's non-open data model is limiting their utility as the source of truth.
But I think it's part of a larger mistake Figma is making: they seem to have shifted to an extraction mindset too early, assuming they'd captured the market, right when the ground beneath them is starting to shift.
It's most visible in their pricing model evolution, which is now explicitly anti-collaboration. Figma used to be the obvious default because you could quickly share files with non-designers, so they could view and make small edits without fuss. Now that requires a paid "seat", along with a confusing mess of permission flows.
It's platform wide too. I taught a college design class recently, and had students sign up for Figma because it seemed archaic not to teach them to use it. Instead of just giving any ".edu" address a free account (like they used to) students are forced through a 3rd-party process of uploading transcripts to prove education status. A few of my students got rejected or ran into confusing errors, and never got access… Now I have to re-evaluate whether its worth using when teaching the class again. (And this is for a population with near-zero short-term purchasing power, but huge potential long-term value… why add barriers?)
This is such a weird self-inflicted wound for a collaboration platform to make. The big tools that won on collaboration (eg. Google Docs, GitHub) have understood that low-friction sharing is critical to becoming the default choice. And that being the default is a flywheel that drives adoption, both in users and in tooling.
It makes more sense if you see it through the lens of Figma trying to juice short-term numbers for their IPO. But it's sad to see because it had so much long-term potential.
I miss the days of having a native desktop design app with a perpetual license.
What Figma achieved technically in the 2010s was amazing. Coded the app in C++ and then used WASM to deliver it as a multiplayer web app.
But now it's trying to be too many things. Why did they ever feel the need to add slides and this other stuff.
Their MCP is poor (sure, they'll improve it).
The app struggles with larger files and performance is sloppy.
And don't get me started trying to design data grid heavy apps.
And they could easily follow Adobe's lead. Enshittify and lock you out of your account whenever they feel it's necessary (remember what happened with Venezuelan Adobe users a few years ago?)
Either Penpot gets their act together and will become the opensource design canvas for open-weight AI models or we will see another open source solution that will fill this space.
> I miss the days of having a native desktop design app with a perpetual license.
You can go that route with Affinity Designer [1], owned by Canva, who partnered with Anthropic on Claude Design [2]:
We’ve loved collaborating with Anthropic over the past couple of years and share a deep focus on making complex things simple. At Canva, our mission has always been to empower the world to design, and that means bringing Canva to wherever ideas begin. We’re excited to build on our collaboration with Claude, making it seamless for people to bring ideas and drafts from Claude Design into Canva, where they instantly become fully editable and collaborative designs ready to refine, share, and publish.
[1]: https://www.affinity.studio
[2]: https://www.anthropic.com/news/claude-design-anthropic-labs
I have it, jumped on the affinity band wagon years ago after Adobe started their enshittification process.
After Canva bought Affinity, you now have to authenticate with your email from time to time when you launch the desktop app. Annoying and why do they do that?
Might go back to Affinity 2.
> Might go back to Affinity 2.
Sure, but they stopped updating Affinity 2; at some point, it's going to stop working unless you never upgrade your operating system.
Front-end, UX, design, and product have become one role. The market is just realizing it slowly.
So your designers debug your React code now when the AI messes up?
In my opinion this should have always been the case. All designers should be able to code and do html/css. It's the medium of design.
HTML and CSS, sure, but modern frontend design is way more than that; it's a jungle out there.
React, Nextjs, Vue, Nuxt, and Angular are pretty much what AI is the very best at coding in my experience. Probably because they are all meant to build essentially the same thing with different curtains.
> All designers should be able to code and do html/css. It's the medium of design.
So a fashion designer can mass produce clothing? So an interior designer can build a house?
This designer should has never held.
Fashion designers can, in general, work with fabric, yes. And an interior designer should probably have some idea of how to paint at the very least. To me with web design so much of what matters is encoded in the CSS and HTML that it is the final design product. Anything produced before is a sketch, a concept, but it's not a design.
The analogies you have offered aren't great.
For example designers and developers both use the computer as their primary medium of working. Their outputs resemble each other very closely, despite having a different underlying form.
Contrast that to the interior designer building a house, well those are different mediums. There is no efficiency gain from the interior designer designing the plan and also implementing it. Where as with a designer working in code there is one.
Fashion designers do indeed make clothing by hand, it's a very important part of their craft. This example disproves your stance.
That’s what CEOs think lol. Let’s see if it pans out!
"I have an error when i click on the orange button, see screenshot. fix it."
I've been a developer for over 2 decades and I've been using AI in our react codebase for the past 3 months. Outside of some optimizations there's not much a designer couldn't debug through Claude Code. 90% of the industry is toast.
I want to be wrong because I'm watching the death of my entire career, but everything I've seen is pointing to this as an inevitability. We are shipping better and more secure code, and doing it easily twice as fast. Many development teams can be cut in half today with no reduction in output. I don't want to say it out loud at work yet, but we're actually producing too much.
I've been writing code for 50 years and it looks now that we have seen sunrise and are about to see sunset on humans writing code by hand.
Is that bad? Not to anyone who has managed dev teams and is familiar with the incredibly tortuous and painful business of trying to corral a bunch of humans with varying skill and enthusiasm levels to create software. We have tied ourselves in knots with things like Agile just trying to work around the fact that software development is so slow and arduous.
Many times back in the waterfall days I have written up design documents to kick off dev teams on multi-week or month projects. Now I could feed those into Claude Code and get results in days. This stuff is exciting beyond belief in just getting shit done.
This is a golden era for any established company with an existing customer base. My question to them would be "with Claude Code, why aren't you carving through that massive backlog of feature requests that has been building up over the years?".
A lot of people seem to look at this as job threatening, and it surely is for junior devs. But for companies that already have a strong senior talent bench, it's time to raise the ambition levels and ask not how many jobs can be shed, but instead just how fast and hard can we go now we have these new superpowers.
This is so context-dependent. Coding some generic crud app is indeed becoming more automated, but most of the stuff I end up building is just way outside the capabilities of current LLMs to accomplish without significant steering and gasps hand written code. Most of the stuff LLMs are good at ones hotting are the same things a non-coder could build with no-code platforms anyways. Which is great imo, it means we can utilise our skills and expertise on stuff that is more "cutting edge".
AI is impressive but this same sea-change happened at least twice before - the era when computers went from being rooms full of women(354) to machines programmed in machine language(892) to those with screens, keyboards and even assemblers (assembly language, especially macro assemblers, were considered seriously high level at a time), to mid-level languages like C (considered needlessly complex and slow at one time, now considered barely above a macro assembler), to high-level languages like Java and even higher ones (arguably) like Rust.
Every one of those transitions has resulted in more programmers - though not necessarily the same programmers.
354: https://en.wikipedia.org/wiki/Computer_(occupation)
892: https://en.wikipedia.org/wiki/The_Story_of_Mel
But things really are different this time. Computers and software were nascent industries with lots of room to grow, lots of software to build in previous transitions. Today software and technology companies are the biggest in the world. Every industry uses software. Getting your web app, mobile app or game discovered is actually a huge problem today because we have so much software. There is not infinite demand for software, or for anything else, even if it seems that way in the early days.
The Olde Days were custom code for a business doing business things (only they could afford a computer, and some universities), then there was a Cambrian explosion of software sold by individuals or small developers to many (80s and 90s), then we've moved to large companies and SaaS.
I think we're about to cycle back to "custom code" except now it's for everyone, by AI - you don't need to find the to-do app of your dreams; you can code one for yourself in a fever-dream.
The era of "write Wolfenstein 3D in a few months and make millions" are gone, but they've been gone a long time already.
Or - there was a HN discussion on this half year or so ago - there's consolidation again, and there will be AI, but no code. Domain expert talks to the AI, perhaps with an expertised intermediary. AI spins up a whole new 'software platform' for the customer.. internally. Offers all the UI that is needed to work with it.. still 'in the cloud' i.e. in AI data centers. Customer happy, devs less happy.
AI isn't a technology that replaces programmers, it's a technology that replaces generic human beings. The manager of your agents will be an agent, too.
I think the real question is which of the four roles is going to be the one that takes over. Probably people who were already UX-Engineers.
I would ask this: which is the worse failure mode —- design not quite right, or users can’t access the app?
On the other hand, teaching taste is quite hard, and is what people respond to and what designers learn.
Teaching programming is a bit of mostly solved problem, today anyway.
Realistically this doesn't mean all pure designers go away. Large orgs can have a small team that set the overall style guide and designs important pages, and the rest of the org just follows using AI to iterate.
> Outside of some optimizations there's not much a designer couldn't debug through Claude Code. 90% of the industry is toast.
I've seen the "debugging" and "coding" that non-coding designers are attempting to vibe-code. 90% industry is definitely toast, but not the 90% you're thinking of. Most industry is going the way of Microsoft that cannot even display a start menu in under a second
Leading question, feel free to ask a more honest one.
How is it a leading question?
They entailed scenario that isn't entailed by the person's claim.
i.e. The OP doesn't need to answer yes to their question for OP's claim to be true, yet their question pretends otherwise. (non sequitur)
>A leading question is a query that suggests the desired answer or puts words into a witness's mouth, often guiding them toward a "yes" or "no" response.
It was just restating what you already said; no need for this specious response.
Please quote me where I said 'my designers debug react code that AI messes up.'
I did not say anything of the sort.
See also: https://news.ycombinator.com/item?id=47819428
All right, here's a statement: your designers won't even know when the code is wrong. Just because it compiles it doesn't mean it's fine. They lack code judgment in the same way your coders lack design judgment.
Thank you.
In response I suggest that the engineers using AI also lack code judgement (because they are not reading it either). I don't think questioning the AI use is the actual topic here, it is the shifting roles. Who says it's the designers that are taking the new meta-role? It's probably the FE's honestly.
The role shifting doesn't mean that it's the best path forward. I'm simply stating that it is happening.
It's very easy to know when code is wrong: it doesn't work the way it's expected to. So you explain to the AI what's wrong and the AI fixes it.
This isn’t meant to be sarcastic: have you ever worked for a real company?
As we know from THERAC-25, etc., comprehensively verifying that code works the way it's expected to is not actually very easy - it's perhaps one of the hardest parts of building any system more complex than a toaster.
Thankfully the CRUD app that is being developed by some random startup is not likely to cause as much harm as the THERAC.
Your designers are going to be looking at the layout; they're not going to notice if it's slow, uses too much memory, is not maintainable, doesn't follow repo patterns, etc.
Do you think it is reasonable to expect a person with an arts degree to know this?
> Do you think it is reasonable to expect a person with an arts degree
Of course it is.
The only people who think your fucking college degree determines your knowledge level and ability are teenagers and people who are so deeply untalented that it’s the only way they feel qualified.
When you can control the model layer like Anthropic, you get more leverage over the traits of the persona, enough so that the system feels closer to havin consistent expert design judgment built-in that complements the 'truth-to-materials'.
Basic web development is completely over, and will be automated end to end, product, ux, design, and the code.
I have a complicated nextjs webapp, and I havent had to write front end code in six-nine months now.
If it’s used by < 1000 people, it’s not a complicated app
FedRAMP will be the death of me.
so like 90% of software?
Excellent post. I share the author's sentiment which is essentially "to hell with Figma, at least fix Sketch". Been feeling very lonely in may hatred towards Figma, which is for a whole bunch of reasons (among others, it's an incredibly shitty, memory and CPU hungry Electron app that looks and feels worse than any more or less well designed web site), but now after reading this I realize the number of reasons has doubled.
It may look like a crappy Electron app, but Figma has a quite interesting architecture. The browser editor is developed in C++ and cross-compiled to JavaScript with emscripten. The rendering engine looks like its handling HTML, but it's actually rendering their own document format for cross-browser consistency. They have their own CRDT implementation to handle multi-user edits.
[0] https://www.madebyevan.com/figma/building-a-professional-des...
[1] https://www.madebyevan.com/figma/
[2] https://www.madebyevan.com/figma/how-figmas-multiplayer-tech...
I think my biggest question is who cares? What does having an interesting internal architecture have to do with the “its electron though” ideological attack.
It is made to perform much better than your typical electron app would. Saying electron-based == shitty is complete misunderstanding of the technology. Although i dislike Figma as much as the next guy, their app was in many ways very impressive. See Figma's cofounder old articles at https://madebyevan.com/figma/
(author of the post here) I cut a paragraph how Figma costs cuckoo bananas money for your entire team for the privilege of enduring this byzantine nightmare. And they paywall certain features, which you likely can't get authorization for, so you have to do more hacks on top of hacks on top of the “gold standard” practices I shared in the blog post. The price ramp is not gradual.
I think this is what kills them, not AI. I know 2 companies (enterprise level) who are migrating away from Atlassian atm just gets too costly.
man, I dont even use Figma for personal & side projects because its so expensive. I still occasionally fire up sketch or freehand it.
Figma is a work tool only and I'm disappointed by its MCP tooling which feels late and behind where it should be, I just feel forced to use Figma Make which stays in their walled garden without practical utility and connections to my actual codebases
Figma is going to regret not having sold to Adobe.
I was making a new brand file in Figma and was about to define all the variables and aliases, paths, all that. Took one look at it and just made the 8 colors I needed and went on with the design.
Can someone please help me understand why people are obssessing over this as a Figma killer and not, more clearly, a shot at Replit, Lovable, etc.?
The AI app builders are part of the AI economy already. They are essentially re-selling tokens/compute. It's an existing horse race.
Bringing a Figma killer to the market is converting a conventional software sector into an AI one. So it's more disruptive.
Honestly, I never understood the move to create an artificial dichotomy between design and code with a heavy layer of tooling.
I suppose that a layout engine made sense in the context of Flash, and you saw the future of the web as a set of keyframe animations. But the notion that there's a lot of value in creating a very heavy, high-friction abstraction between the UI/UX and the platform it ultimately runs on was always going to be a loser.
In the end, it turns out we're all just web developers, regardless of your weapon of choice.
I hope the authors mom is not on the internet. Cursing in capslock. Good grief.
If you get to the post script it sounds like they are just pissed they didn't get hired by figma.
I promise I am fine lol, I actually had a really productive chat with their panel last year and found it wasn’t the right role for me anyway for some logistical reasons.
I feel like we're leaning way too much into the "vibe" aspect of using LLMs at the moment. There is definitely a good use case for LLM's here, but is just prompting your way into a design really the best method here? I feel like something in between Figma and Claude Design would give designers the control they want, but still removing the friction of going between design files and the code impl.
Oh no. Figma has variables and instances and it's hard to debug bad colors.
And here I am with Claude Code... That so far generated a 2000-line CSS file for a 7000-line app consisting of literally three web pages [1]. Where almost every single color, component, class and style is duplicated at least two times. Where custom classes are fighting with Tailwind classes (yes, there's also Tailwind ON TOP of custom CSS) that are fighting with inline hardcoded style= declarations.
Figma is definitely going to suffer the vibe-coded design slop-app from Anthropic.
[1] 7k lines are almost justified for the functionality in them, and I tried to keep an eye on the code. It's harder to keep an eye on CSS
Have had similar dramas with CSS. Opus/Sonnet seem to have far less taste and discipline in writing CSS - constantly they are trying to slam in changes with !important or inline styles rather than think through how to do it properly.
We can blame AI for slopcoding but it's just what we've been doing for decades now - who cares if it takes 9000 lines to do what in theory Linus and Knuth and Carmack could do in 3000? Our CPUs eat a million lines of code for breakfast and demand more.
Strange then that so many apps are barely usable on these powerful CPUs. Perhaps, CPUs can no longer demand more