Designers Deliver Executable Front Ends
UX lead at real estate firm on running a website redesign with Claude Cowork
The important shift is that AI turns design from a static spec job into a code producing job, which pushes developers up the stack into systems work. In this workflow, the designer is not handing over mockups and animation notes, but a live Vercel site and GitHub repo with responsive layouts, CSS, HTML, and interaction behavior already working. That removes weeks of front end translation work and makes engineering focus on wiring real company data, APIs, and production infrastructure.
-
The old workflow for a similar investor relations redesign used two designers for weeks, then 10 to 15 developers across front end, back end, integrations, and QA over roughly three months. The new workflow compressed design and front end build into one designer using Claude over about five days, with a smaller dev team stepping in mainly for integrations.
-
The handoff changed because the designer delivered executable code, not Figma files. Claude helped set up Vercel, generate HTML and CSS, and organize a GitHub repository, so executives reviewed a working staging link. Developers then inherited something they could run immediately, instead of rebuilding visual intent from screenshots and annotations.
-
This only works because the designer carries a reusable context packet across chats, including Markdown session exports, brand rules, screenshots, URLs, and prior page code. That keeps shared components like navigation and page styling mostly consistent across parallel builds, although visual drift still has to be caught in review.
The next step is a split where designers increasingly own the presentation layer and engineers specialize in data, integrations, and platform reliability. As these tools improve at preserving context and reducing layout drift, more web teams will treat live code as the design artifact, and Figma style handoffs will matter less on design heavy marketing and IR sites.