Maybe the real bottleneck was never designing screens, but understanding what the hell the client actually needs. Anthropic’s launch of Claude Design pushes exactly in that direction: generating prototypes, presentations, one-pagers, and visual pieces from natural language, with the ability to learn the user’s brand and produce fast results in research preview mode. (The Times of India)
The signal is clear. If an acceptable interface can appear in minutes, the value stops being in “drawing screens” and shifts toward defining problems, constraints, flows, business goals, compliance, brand tone, and real user needs. Put brutally: 90% of the process should be intelligent documentation and serious discovery; 10% should be rendering what was already well thought out. It is terrible news for people who confused Figma with strategy, and excellent news for people who know how to listen.
On that same board appears Google with Stitch, presented as an AI-native design canvas for creating web and mobile interfaces from prompts, voice, images, or mixed context. Google even named the phenomenon “vibe design,” an expression that sounds like a tired creative agency but describes something real: iterating quickly from business intent instead of starting from rigid wireframes. (blog.google)
Similarities between Claude Design and Stitch? Plenty. Both reduce friction, turn language into visuals, and make the first version stop costing weeks. Both also threaten the old model where every minor change required ceremony, tickets, and a meeting with 14 people watching a progress bar.
Differences? Also important. Stitch seems stronger in pure digital product work: app UI, web UI, prototypes, and a bridge toward code. Claude Design aims wider: presentations, marketing materials, rapid visual assets, and business artifacts, not just interfaces. Stitch thinks like a product designer; Claude Design thinks like a consultancy that learned how to run.
So, which one is better? It depends on where the pain is. If a company needs to explore app flows and move from idea to frontend quickly, Stitch seems more tuned today. If the problem is translating fuzzy needs from commercial teams into immediate visual deliverables, Claude Design looks more versatile. The adult answer, of course, is less epic: neither one replaces discovering the problem well.
Because the perfect screen made in ten seconds is still useless if nobody understood the client during the previous three months. And that, unfortunately for the market, still cannot be solved with a prompt.