Claude Design, Anthropic's AI-powered visual canvas, just got its most significant update since launching in research preview two months ago. The product is now in full beta, and the headline change is one that early users were loudly asking for: it actually stays on brand. Claude now checks its own output against your design system before you ever see it, and a rebuilt import pipeline lets you bring in components from a GitHub repo, design files, or raw uploads.
One million users, one loud piece of feedback
Over one million people used Claude Design in its first week, and the feedback was consistent. Users consistently reported three limits: fast usage burn, drift from its own design system, and a recognizable default aesthetic. This update addresses all three directly.
Anthropic rebuilt the design system import to give more flexibility and increase precision. You can now bring in one or several design systems from a GitHub repo, design files, or raw uploads. Claude builds with your components, checks its output against your design system, and makes corrections before you see it. That last part is the key shift: the model is now doing a self-review pass against your brand guidelines, not just generating and hoping for the best.
What Claude Design actually is
Claude Design lets you collaborate with Claude to create polished visual work like designs, prototypes, slides, one-pagers, and more. The interaction model is simple: you type what you want on the left, and Claude builds a working design on a canvas on the right. From there, instead of re-prompting from scratch, you comment inline on specific elements, edit text directly on the canvas, or use adjustment knobs to tweak spacing, color, and layout live.
Don't miss what's next in AI
Join 300,000+ engineers and researchers who get the signal, not the noise.
- Full access to in-depth AI research breakdowns
- Be the first to know what's trending before it hits mainstream
- Daily curated papers, repos, and industry moves