"The traditional design process is dead," IDE becomes Claude's favorite: Everyone writes code, and is least afraid of bugs

The new lead is an AI

It has been reported that Claude, Anthropic’s AI assistant, is being used internally as a "design lead" and that integrated development environments (IDEs) — not sketchbooks or Figma files — have become its preferred workspace. Huxiu (虎嗅) published the account, saying Claude's workflow centers on code-first prototypes and rapid iteration. Is the traditional handoff from designer to engineer over? Reportedly, many teams are already treating prototypes as executable artifacts rather than static comps.

What this means for teams and tooling

The shift is blunt: everyone writes code and teams are "least afraid of bugs." In practice that means designers learn to ship runnable interfaces, product managers expect live demos instead of slide decks, and engineers spend less time translating visuals into working features. For Western readers: Anthropic is a California-based AI firm best known for the Claude family of language models, while similar pressures are emerging across major Chinese cloud and AI players such as Baidu (百度), Alibaba (阿里巴巴) and Tencent (腾讯), each racing to build tooling that blurs the line between design and development.

Broader implications and the geopolitical backdrop

There are clear upsides — speed, clarity, fewer misinterpretations — but also new risks: intellectual-property control, supply-chain security for code-heavy prototypes, and governance over what an AI "decides" about user experience. Geopolitics matters too. Export controls and chip restrictions shape who can run heavy models locally, influencing whether code-first design remains the domain of big, well-funded firms or becomes broadly available to startups and smaller teams.

Outlook

Whether the phrase "the traditional design process is dead" proves hyperbolic remains to be seen. But the direction is unmistakable: design is becoming executable, and AI is accelerating that trend. Organizations that adapt their workflows, toolchains and governance will likely pull ahead — the rest risk being left with polished mockups and slow handoffs.