You'll get a competent UI with little effort but nothing truly unique or mind-blowing.

Impressive technology, but that old skool artisanal weirdness of yore only becomes more valuable and nostalgic.

Before these tools, when a client wanted a specific section built, we'd spend hours hunting references across the web. The output always ended up feeling like a mesh of 2-3 sites, never fully unique. Then we'd burn more time explaining the intent to the client's designers and devs, usually with multiple rounds because words don't convey layout well.

Now we throw a quick mockup together in Claude or Lovable and send it. The designer gets the idea in 30 seconds instead of a 45-minute call, then pushes it further with their own taste and the client's branding.

It's not replacing designers. Most clients don't know what they want until they see it. These tools collapse that feedback loop from weeks to minutes, so the designer actually spends their time on the parts that need human taste, not on decoding a vague brief.

Anyways, this is 100% a shot at Figma, but also catching Lovable in the crossfire. If anybody from Anthropic is reading this, if you keep developing this with features in Figma and other design tools, you'll have a major hit on your hands.

I use it all day every day with Claude Code. I sometimes wonder past code if this has had the biggest impact on my day to day productivity, either having to make do with semi-bad looking reports or have a designer design them (which is slow).

Sort of feel sorry for Figma in a way though, given all the "partnerships" (highlighting their MCPs) and case studies they've done with Anthropic and then they release this. I note there isn't a testimonial from them this time.

I'm surprised how poorly Figma have used "AI" in general - given they were the "gold standard" in taking emerging technologies (WASM etc) and making an incredible product. The Figma Make thing was incredibly underwhelming, I managed to extract the system prompt out and it's basically just Gemini 3 Pro with a design prompt. Perhaps the original team has left?

They are extremely exposed imo. While all the UI/UX designers will continue using it for the forseeable, I strongly suspect a lot of their (A/M)RR was coming from extra seats for PMs, developers, etc to view and export and do commenting on the files - not core designer usage. I think a lot of this just won't happen on Figma as much.

- The best design is original, groundbreaking and often counterintuitive. An AI model is incapable of that, it's uninspired, it will absolutely converge to the norm and homogeneity (you see it everywhere now, just scroll on ShowHN and take a look at the UIs) and produce the safest design that appeals to its understanding of the ideal user.

- Good designers will reject this, they prefer to be hands-on and draw from multiple sources of inspiration which is what Figma boards and Canva is good for, also mainly for cross-collaboration. If you've seen how quickly a great design engineer can prototype you'll know that "speed" they advertise in this video is not worth the tradeoff.

- Creatives typically have a very very very high aversion to AI.

- Non-designers will not see a purpose for this tool, basic design can already be done through Claude Code and Claude.ai, I fail to see what this could offer unless they leverage a model that is more creative and unique by default (you can not prompt/context/harness engineer creativity believe me I've tried).

- Design is a lot more than just UI. Tools like this ignore so many other important aspects like: motion, typography, images, weight, whitespace, sound, feel.

* Massive token usage, some small tasks burned through $50 of credits and did not offer $50 of value.

* Terrible at logo work. Comically bad. This is something that is "hard" so it could add great value if it could deliver.

* Repeatedly forgot prior feedback - when iterating it would re-implement prior iterations after being told why we didn't want that result which made for a very frustrating UX.

* Prone to adding visual clutter - kept adding extra elements that look "pretty" but add no value to the user.

* Seems better at "pretty" vs user focused / UX.

* Did not do a good job at using my existing design / UI library

* REALLY wanted to start from scratch. Could not be coaxed into designing part of an application, it wanted to redesign the whole thing.

I use Opus to generate Typst for that and I'm already pretty happy with that approach. It gives me a degree of control I do not have with other methods, because

1. Typst is really powerful

2. Opus is really good at surgically modifying Typst

I basically never look at the Typst code for this. Telling Opus visually what I want changed is usually good enough.

My 76yo Dad just used Claude Design to make a simple commercial tenancy/lease management CRUD app. We handed what he built to Claude Code to implement within my existing business tech stack. It perfectly followed my coding & testing playbooks (basically a multi-file CLAUDE.md with prompt & command hooks to evaluate conformance). It works… exactly as my Dad imagined.

Now, mind you, my Dad had some outdated UI expectations (he asked Claude Design to style use a Windows 98 aesthetic!), but my point is: we just built a fully-functional, deeply tested, domain-expert-designed app, in less than a day.

Seems like Claude is actually building almost like a layered Figma wireframe that you can do fine grained adjustments afterwards (e.g. adjust font size).

Interesting that Canva provided a quote of support. I'm not familiar with the differentiation, but it seems like this will directly siphon customers from Canva, right?

I wonder what other features they're cooking right now.

Here's the prompt as of around release-time yesterday. You can verify in your own network requests.

https://gist.github.com/mcrowe/9da081f52e0740886d39e730852db...

Anthropic has distribution on their side, their engineers are excellent (I have ran with them across the ggb in the past and they work 12 hours plus a day regularly.)

I think what actually might be slowing them down is the public releases and pr lol, not ideas or execution

It's also just a beautiful product. The interaction model, styling and integrations via exporting is all super thoughtful

Is that globe made from whole cloth or is there a bespoke "telecom globe" widget that it dropped in? Could I ask for mock up of molecules with the same fidelity of knobs, down to nucleus size and such?

Deriving a bland average of creativity is the saddest thing you could do. I don’t even enjoy design and I find this offensive.

> You've hit your usage limit. Claude uses a different usage limit. Resets Mon 12am

It's Tuesday. What are we doing here, Anthropic?

This was also the last day until my subscription ends. Good riddance.

Exact Transcript as follows:

-- Claude: Let me fix these precisely: <empty>

-- Me: ok

-- Claude: <empty>

-- Me: hello?

-- Claude: <empty>

-- Me: there's still some issues with 01 live and 02 - your business.

-- Claude: <empty>

-- Me: hello?

-- Claude: Sorry about that. Let me look at those two and figure out what's still wrong. <empty>

--- Me: hello?

-- Claude: Screenshot _illo_check.html Viewing image illustration-pillar-1.svg Viewing image illustration-pillar-2.svg <empty>

Lately it is more and more ShadCN as well.

TailwindCSS is a masterpiece but ironically doesn’t really get its fare share while “Build on top of TW” frameworks make money.

TailwindCSS is the final evolution after all other frameworks always had its benefits but also massive limitations.

BEM anyone?

TW is really elegant a new paradigm in its purest sense and brilliantly executed. No wholes could be poked in it for years and the extensibility shows, how brilliant it is.

Bootstrap will always be held dearly but it was about browser quirks etc first. Important milestone but stands no chance against TW.

This app is pretty slick, this will funnel a huge number of customers away from Figma + Canva imo.

Very interesting though

I'm wondering how i can CONTINUE that in this design thing, can i import something? Because they show it the other way... you can start and edit, and then export to claude code.

Until then, I guess it's back to just using CC

As someone who's thinking about side project-ing a game, this caught my eye.

I am curious to explore what Claude can yolo in terms of a retro style indie game... One who's audience might only be me.

I have been doing fine just instructing Claude code to use Tailwind and reference design documents

What I found valuable is the design.md that was produced. It's a guide for building each component. So using these tools becomes akin to PSD to html we used do. At least that's when I find them most effective.

half the design problems I've seen weren't bad taste. they were bad communication between the person who knew what they wanted and the person who could build it.

Page not found Claude can help with many things, but finding this page isn’t one of them.

when logged in.

Edit: I see they have a new token tier 'Claude Design' so that's good.

Anything but that

Ultimately, this really just shifts the focus towards product design and ideation rather than UI design.

Can't even test two designs. No idea if PDFs used up so much budget.

Fed it my figma file to it and creating a design system. And yeah, it looks like it, but it's still wayyy off at the same time

Maybe AI is not good at everything, yet.

That link is redirecting me to https://claude.ai/404, anyone else?

If you’re confused about why everyone is shipping their own canvas tool - this is what I wrote when I reviewed Stitch from Google in my newsletter:

“every SOTA (State-of-the-Art) model can already do this. Give it a prompt, it'll spit out an HTML design. Ask for 10 responsive variations, you'll get them. Stitch is a Ul and context harness on top of Gemini, in the same way Claude Code is a harness on top of Claude's models. This means every Al lab will likely ship their own version, and they'll all interoperate because at the end of the day it's all HTML and markdown.”

More generally, this is a competition of where the product development work starts and lives. The business value will accrue to those who become a destination, not a plugin.

AI can now power a new sort of tool that supports the entire process (not just coding or just design). So there’s no reason for Anthropic or any other lab to give that up to another tool.

The design problem to solve post-ai isn’t this it’s how the space for thinking fits into all this, getting to the end result slower so human ideation can play out. This is just optimized for the first generic output + tweaks.

Unfortunately it's going to be a tough few years until that happens, where it really does feel like the idiots are winning.

THE IDIOTS ARE WINNING

Reusable components. Shared styles. Auto layout. Variants. Design tokens. So many useful plugins! Interactive prototypes. Dev mode handoff. Versioned libraries. A single source of truth keeping designers, engineers, and PMs speaking the same language.

It's fucking bliss when you use it right. And best of all, anyone on the team can build with the legos the designers gave us!

Teams finally started treating the design system like infrastructure instead of decoration.

Then Figma Make showed up.

It generates pages that ignore your components, skip your tokens, and treat every screen like a fresh snowflake. Looks fine in the mockup. Lands like a brick on the dev team. Now every page is bespoke and every handoff is a rebuild. Fucking sucks.

We had it all figured out, then we fucked it up.

AI was supposed to accelerate momentum. But it just so completely lacks maturity. AI Tools are dragging teams back to 2010, when every page was a Photoshop file (with way too many layers and shitty names for them) and none of the designers can agree on what radius to assign to their buttons.

Good design systems are a discipline. If your AI tool doesn't respect the library, it's not a designer. It's an intern with a Molotov cocktail waiting to destroy your productivity.

I haven't used Claude Design yet, I'll try it, but I LOVE Figma, and I hate Figma Make. I don't have high hopes for Claude here.

Really hope Anthropic didn't notice my ingenious work and quickly copy it.

> do you see our current html

Hi-fi design Interactive prototype html

> do you see our current html

Hi-fi design Interactive prototype html

● Claude Yes — I can see your html/ folder. It has 7 files:

files-list.html math-result.html rows-view.html table-view.html text-editor-doc-view.html trending-topics.html unique-result.html Want me to open them up and take a look at what's in there? And what would you like to do — polish them, turn them into an interactive prototype, build something new on top, or something else?

> yeah in table view i'm thinking about how to deal with cells that may be big. i guess we want to do something different than excel or sheets in that auto expand rows with cells that contain big values. we can show like the first -- what does twitter show, 280? so we can show 280 chars and then [...]

● Let me take a look at the current table view to understand what we're working with.

Got it — I see the current table view. Right now every cell is clamped to max-width:220px with white-space:nowrap and text-overflow:ellipsis, so long values just get cut off at whatever ~30 chars fits the column. You're thinking about the opposite end: what if a cell genuinely holds 2,000 chars of text? Neither pure clip nor Excel-style row expansion is quite right.

Before I start sketching, a few quick questions so I don't guess wrong:

> Questions answered: - truncation_threshold: ~280 chars (new tweet) - expansion_ux: nothing right now -- it's not interactive - row_height_policy: Grow freely like Notion/Airtable - content_types: Not sure yet - indicator_style: [...] at the end of the text - newlines: Render them (row grows) - copy_affordance: no interaction right now - column_width: Fixed width, all columns equal - variation_count: 2 variations - aesthetic_direction: Same bones, but let you push it a bit

● Got it — so: non-interactive, rows can grow freely, 280-char cap with [...], newlines render, same bones but push it a bit, 2 variations. Let me sketch this out.

Set up variations file with two table treatments V1: Notion/Airtable-style free-grow with [...] cap V2: pushed aesthetic variant Wire both to the same sample data w/ long cells Let me peek at a couple of the sibling files so the variations stay in-family with the rest of the widget set.

Good — now I have the visual family. The shared vocabulary across widgets:

#f5f5f3 page, #fff card, #d0d0cc borders, #f0f0ec headers System font, 12–14px Subtle alternating row tints, no drop shadows, flat corners Color used sparingly; #666 for secondary text My plan for the two variations (side-by-side in a canvas so you can compare on the same rows):

V1 — "Notion-ish, stay in family"

Keep current palette, borders, zebra striping Rows grow vertically as needed; white-space: pre-wrap so newlines render Cap each cell at 280 chars; if longer, slice to 280 + […] inline Same fixed column widths, everything else identical V2 — "Same bones, pushed"

Tightens the grid (hairline borders instead of panels, more whitespace, denser typography) Column rail with sticky row numbers Soft fade on the bottom of ultra-tall cells instead of mid-text […] when content is really long (hybrid: char cap + visual) Monospaced tabular digits for row numbers; type column ⌾ / ▤ markers subtly hinting content type; column width slightly wider for breathing room Let me build it.

Now let me build the two variations. I'll set up realistic sample data with a mix of short values, tweet-length values, long-long values, and ones with newlines, so the behaviors are visible.

Now the two table components.

Clean. Two variations on the canvas, both rendering the same 7 rows so you can compare directly:

V1 — in-family. Your current widget, with white-space: pre-wrap so rows grow, newlines render, and a soft grey […] appended when a cell goes past 280 chars. Row 3's summary (double-long prose) and row 4's URL show the cap. V2 — same bones, pushed. Hairline borders instead of filled header panels, monospaced tabular row gutter, column letters as uppercased small-caps above the header names row, tighter typographic rhythm, cells > 280 → […] legend in the footer. Row 7 shows how newlines render naturally. Both slice at the nearest word boundary when one is within ~40 chars of the 280 cap, otherwise hard-cut — so […] never lands mid-word.

A couple of things worth calling out, not changed:

280 is pretty generous when a row has multiple long cells side-by-side — row 3 is ~2 cells × 280 = a very tall row. If that ends up being painful in real usage, dropping to ~180 or adding a per-row max-height with fade might be worth a follow-up. Neither variation is interactive per your answer; there's no click-to-expand, hover, or copy. Easy to bolt on later. Fork verifier agent Both variations are live on the canvas. Let me know which direction feels right and I'll fold it back into html/table-view.html." -----

It's basically similar to their Artifacts system in a harness I guess

You talk to it, it consumes "usage" and no design system created.

Had a similar experience with their online code sessions.

All this slop generated code is so freaking bad. We'll get drowned in so much slop code that at some point only these garbage machines will be able to "work" with it.

Part of this weirdness is the continual relativism of design. A taste-meme is good or cringe only relative to the prevailing social environment, never itself. An AI can never do "design", properly understood, because design is the work done by a reluctance motor, spinning endlessly in a Sysphean quest to align itself with a moving magnetic field and producing torque by side effect.

All efforts like this can do is capture the field alignment at an instant in time. It cannot do work. It cannot produce motion, not as long as its weights are as fixed as the field lines of a neodymium magnet. The instant AI design is good, it becomes bad through the act of becoming good.

Producing work through motion of taste may be one of the last human endeavors to be absorbed.

It's not X, it's Y! all over again?

Semi-relevant XKCD: https://xkcd.com/1425/

>NEVER use generic AI-generated aesthetics like overused font families (Inter, Roboto, Arial, system fonts), cliched color schemes (particularly purple gradients on white backgrounds), predictable layouts and component patterns, and cookie-cutter design that lacks context-specific character.

> Interpret creatively and make unexpected choices that feel genuinely designed for the context. No design should be the same. Vary between light and dark themes, different fonts, different aesthetics. NEVER converge on common choices (Space Grotesk, for example) across generations.

[1] https://github.com/anthropics/claude-code/blob/main/plugins/...

With that said, this seems like it's just claude code + simplified devtools in a container? Will need to spend more time with it to see how EG commenting influences things, but 5 minutes in, this seems like it's needless formalization on top of how I already used claude for deisgn

This will fail because it's already a forgotten side project within anthropic, and anthropic also has pretty bad product DNA as a company. Their headcount is already too large and the culture is already set. They grew revenue so fast they speedran the stage at which you could build software product chops into the company culture (think Google circa 2008 vs. Google circa 2018).

They should focus on what they're good at: the actual AI models and B2B sales. Let OpenAI play early Google and churn through 100 different consumer product experiences to see what sticks, they're better positioned for it anyways.

Imagine if a designer were hired to custom design the lightswitches in every building. We need to get back to off-the-shelf interfaces and stop wasting smart people's time reinventing UI widgets.

Go ahead and roast me.