Claude Design — Figma Is Dead

It's not another prompt-to-code demo

The thing that clicked for me around the 14-minute mark 13:52 is that Claude Design isn't trying to be ChatGPT-with-a-canvas. The onboarding walks you through mood, color, and layout before it generates a single pixel. That's a designer's intake form wearing an LLM coat.

For non-designers — the exact audience I've spent years making WordPress tutorials for — this is huge. You're not staring at a blinking prompt trying to guess the magic words. You're answering questions you'd answer on a discovery call.

Sketch, don't type

Around nine minutes in 9:11 we got to the part I can't stop thinking about. You can literally sketch on top of the generated page to communicate edits. Circle a button and write "bigger." Draw an arrow from one block to another. Later in the stream 44:00 we pushed this harder and it held up.

Text prompts are a useful fallback, but if you've ever tried to describe a layout tweak in words — "make this like 10% more padded but only on mobile and not on the sidebar" — you already know why drawing it is ten times faster.

This is how humans talk to designers. It's wild that it took the AI crowd three years to figure that out.

The desktop app is doing a lot of work

Most of my day-to-day is juggling client sites, so the thing I really wanted to stress-test was workflow. Around the half-hour mark 33:43 I walked through how I'm using the Claude desktop app now — multiple project "terminals" in one workspace, one per site. Switching between a grooming landing page and an unrelated client build doesn't require swapping IDEs, tabs, or context.

Pair that with native macOS tools — Preview for image compression, the built-in background removal 26:15 — and suddenly a chunk of my Photoshop / Figma / third-party plugin stack is just… gone. I haven't opened Photoshop in weeks and I haven't missed it once.

Is Figma dead? Probably not. Is the bottom half of the market gone? Yeah.

Around 31 minutes in 31:30 Enmanuel and I got into the honest question: does this replace Figma, Elementor, WordPress? After pushing the tool hard for an hour, here's my read:

• High-end agency work: No. Complex overlapping compositions, pixel-perfect art direction, brand systems — Claude Design struggled to execute these cleanly 47:43. The human touch still wins at the top of the market.
• Low-to-mid market: Over. If you're a small business that would have hired a freelancer off Fiverr or signed up for Squarespace, you are going to ship faster and cheaper with a tool like this. Full stop.

That's the same delta I've been watching play out in code for two years 57:00. It's here for design now.

Where it's still rough

I don't want to oversell this. We hit token limits twice on stream 46:50 53:57. The usage caps are real, and if you're mid-flow on a client demo, watching the tool ask you to wait is not fun. Subscriptions solve money; they don't solve throughput when the servers are busy.

And as I said above, anything asking for specific artistic positioning — overlapping hero images, precise offsets, unusual grids — still required hands on the CSS. Claude Design gets you to 80%. The last 20% is still you.

What I'd tell someone starting today

• If you're non-technical and have been putting off learning design tools, don't. Start with Claude Design and learn the vocabulary by doing.
• If you're a freelancer charging $500–$2,000 for small-business sites: your customer is going to try this in the next six months. Your offer needs to be the thing the tool can't do yet — strategy, copy, photography, ongoing maintenance.
• If you're an agency doing complex brand work: you're fine. For now.

Figma isn't literally dead. But the job most people were using Figma to do? That job is getting done in Claude now. And it's only going to get better from here.