On 17 April 2026, Anthropic launched Claude Design, a research-preview product that turns a conversation into polished visual work — prototypes, slides, one-pagers, UI mockups. Predictably, the headlines reached for "Figma killer." That framing is the least interesting thing about it.
What it actually is
Claude Design lives inside Claude.ai and is powered by Anthropic's vision model, Claude Opus 4.7. You describe what you want; it builds a first version; you refine it through conversation, inline comments, direct edits, or sliders Claude generates on the fly. Given access, it reads your codebase and design files during onboarding and builds a design system, so every later project comes out in your colours, type, and components automatically. You can import from prompts, documents, or a live URL via a web-capture tool, then export to Canva, PDF, PPTX, or standalone HTML — or hand the whole thing to Claude Code to build. It's available to Claude Pro, Max, Team, and Enterprise subscribers, using your existing plan limits (Enterprise admins have it off by default).
That's the spec. Here's the part worth thinking about.
The real shift: exploration stops being rationed
Anthropic put its finger on the actual problem in one line: even experienced designers ration exploration. There's rarely time to prototype a dozen directions, so you do three and pick the best of those. Everyone who's done this work knows the quiet cost — the best idea is sometimes the eleventh one you never had time to try.
When generating the eleventh direction costs roughly what the first one did, that constraint loosens. You explore wide, then bring taste to bear on a much bigger field of options. The early reports point the same way: one team described pages that took 20-plus prompts in other tools needing 2; another compressed a week of briefs, mockups, and review rounds into a single conversation. The headline number is speed. The thing underneath it is range.
Who it's really for
The "Figma killer" framing fails because it assumes one audience. There are two.
For professional designers, this isn't a replacement for the tool they run reviews and design systems in — it's a front end for exploration. Generate ten directions in the time it took to build one, then move the winner into your real toolchain to polish. The pixel-perfect, systems-at-scale work doesn't go away.
For everyone who was never going to open a design tool — founders, PMs, marketers — it's something genuinely new: a way to get from an idea to something visual and shareable without a designer in the loop for the first draft. That's the larger group, and it's where this kind of tool changes who gets to make things at all.
What it doesn't change
The honest caveats matter, especially while it's a research preview. It's not a replacement for a mature, team-based design-system workflow yet. The frontier features — 3D, voice, motion — are real but early. And the "no design skills needed" line is only half true: prompts still need specificity about audience, purpose, and visual intent, and the output still needs judgement applied to it. The skill didn't disappear. It moved.
That's the through-line with everything else happening in design right now. When production gets cheap and fast, the scarce thing isn't the ability to make a screen — it's knowing which screen is worth making. Tools like this hand you a wider field of options than you've ever had. Deciding well among them is still the job, and it's still yours.
Further reading: Anthropic's Introducing Claude Design.