FTC Disclosure: TechSifted uses affiliate links. We may earn a commission if you click and buy — at no extra cost to you. Claude Design has no affiliate program — links to Anthropic in this article are direct, non-monetized. Editorial opinions are our own.
Three days. That's the gap between Mike Krieger stepping off Figma's board and Anthropic announcing Claude Design.
Krieger is Anthropic's Chief Product Officer — and Instagram's co-founder, for context. He resigned from Figma's board effective April 14, 2026, citing no disagreement with company policies. The standard boilerplate. Then on April 17th, his employer unveiled a tool that goes directly after Figma's core value proposition: turning an idea into something visual.
Not a coincidence.
What Claude Design actually is
Claude Design is Anthropic's new AI workspace for generating visual assets from text prompts. Prototypes, pitch decks, UI mockups, one-pagers, landing page drafts — you describe what you want, and Claude writes the HTML, CSS, and JavaScript to render it live in a preview panel.
The product launched as a research preview on April 17th, powered by Claude Opus 4.7 — the same model release that brought multi-agent coordination and a significant jump on coding benchmarks.
What separates this from just prompting Claude in chat mode is the dedicated workspace: an integrated live-preview pane, brand-awareness onboarding, and a set of interaction modes built specifically for design iteration.
During setup, Claude scans your codebase and design files to extract your brand tokens — colors, typography, component styles, voice. Once it's learned your system, every subsequent generation applies those rules automatically. Your startup's primary color and font stack show up in every prototype without you specifying them again. For teams with actual design systems, that onboarding step could save hours of copy-paste work across early-stage assets.
The refinement loop
This is where it gets interesting.
You don't just get a static output you then wrangle in another tool. Refinement happens through four channels: chat conversation, inline comments on specific elements, direct text editing, and — this part's genuinely clever — custom adjustment sliders that Claude generates for you based on the specific output. Spacing, contrast, density, layout hierarchy. The sliders are contextual to what you built, not generic settings you'd find in any export panel.
That's not something Figma's AI features offer. It's also not how Canva works. The closest comparison I've seen is what some AI presentation tools have been building toward — but with a design-system layer underneath and code output underneath that.
Export options: PDF, shareable URL, PPTX, or send directly to Canva for full collaborative editing.
Who this is actually for
OK, here's where I want to pump the brakes on some of the launch coverage, because I think the target user is more specific than "everyone."
Anthropic says Claude Design is built for "people who aren't starting from a design tool." Which sounds like a massive audience — basically everyone who's ever spent two hours on a blank Figma canvas trying to put together something for a stakeholder meeting.
But more specifically: early-stage founders who need pitch deck visuals before they have a designer on staff. Product managers who need a wireframe to communicate an idea to engineering without waiting on design bandwidth. Marketing leads who need a quick landing page concept for internal review. Content teams who need slide templates and don't want to pay a Figma Professional seat for occasional use.
Not: professional designers who live in Figma eight hours a day. Not: agencies delivering polished, production-ready assets. Not: product orgs with established design systems and handoff workflows already humming.
This is what Anthropic calls the "zero-to-one" phase of product creation. Getting from nothing to something visual, fast. I've consulted with enough product teams to know this phase is consistently where cycles get lost — waiting for design resources, explaining concepts in prose that should just be a mockup, building slides from scratch that get thrown out after one review anyway. If Claude Design compresses that loop even by half, that's real.
What it doesn't do yet
Real talk on the gaps, because launch hype tends to bury these.
No real-time collaboration. Figma's actual value in design teams isn't just design — it's co-editing. Multiple people in the same file simultaneously, watching cursors, resolving comment threads. Claude Design doesn't have that. It's a single-user workspace, at least right now.
No persistent component library. Figma's design systems — components, tokens, named snapshots with version history — don't have an equivalent here. Claude Design learns your brand on intake, but it doesn't maintain a component library you can update and propagate.
Developer handoff is limited. Figma Dev Mode gives engineers precise measurements, CSS values, exported assets, and code annotations. Exporting HTML/CSS/JS from Claude Design is genuinely useful for early-stage mockups, but it's not a production handoff spec.
No plugin ecosystem. Figma has thousands of plugins — accessibility checkers, content population, icon libraries, auto-layout utilities. Claude Design is starting from zero on third-party integrations.
None of these are dealbreakers for the use cases I described above. For established design teams running a professional workflow? These gaps matter. A lot.
Pricing
Included with your existing Claude subscription. Claude Pro is $20/month. Access also comes with Max, Team, and Enterprise plans.
No separate tier, no add-on cost. If you're already paying for Claude Pro — and the cost picture for the Claude product family is actually reasonable for what you get — Claude Design is just there.
For comparison: Figma Professional runs $15/month per editor, Organization is $45/month per editor. At any reasonable team size, Figma gets expensive fast. Claude Design, within an existing Claude subscription, doesn't add a line item.
For a solo founder or a two-person product team already paying for Claude Pro, the value proposition is unusually clear.
The Figma question
Is this a Figma killer? No. Not in April 2026.
But that's probably not the right frame. The more interesting question is: who does Claude Design prevent from becoming a Figma customer?
Early-stage founders, solo PMs, marketers, technical writers — people who need something visual but couldn't justify a Figma seat, and who've been cobbling together slides in Google Slides or commissioning one-off design work. If Claude Design is good enough for their use case — and it might genuinely be — they don't convert to Figma. That's not an existing Figma user leaving; that's a future Figma user who never arrives.
That's the real competitive threat. Not churn, but forgone acquisition.
Krieger understood this. He's been in design tools his entire career — from Instagram's visual product to his time at Anthropic. Stepping off Figma's board three days before Claude Design's announcement wasn't accidental timing. That was removing a conflict of interest before it became a governance problem.
The smarter read is that Claude Design in April 2026 is a flag-planting exercise. A research preview with real capability and real gaps. What it looks like after six months of iteration and feedback from Claude's subscriber base is a different conversation — and a more concerning one for Figma's growth projections.
For now: designers, keep using Figma. For everyone who's been losing hours in the zero-to-one design phase without a design background, Claude Design is worth a serious look.
Top comments (0)