The handoff tax is real and it compounds. Every sprint, designers ship intent and receive approximation back. Easing curves get rounded. Hover states lose frames. Spacing snaps to the nearest token. No one files a bug report. Users just develop a vague sense the product is slightly less trustworthy. At Alan, this came into focus around a single button color change that took weeks to ship, not because it was complex, but because the person who spotted the problem had no path to fix it. Their 'Everyone Can Build' initiative gave designers, PMs, and ops teams that path. Over two quarters, those teams shipped 283 pull requests directly to the codebase, each paired with an engineering reference for final merge.
The decade-long argument about whether designers should code was always the wrong frame. The barrier was never technical. It was organizational. Tools like Claude Code did not teach designer Ella Moran at Tracksuit to write production code. They removed the justification for keeping her out of the codebase in the first place. The argument worth reading in full is not about efficiency gains or reduced annotation overhead. It is about authorship. When the designer who decided what a transition should feel like is also the person who sets the easing curve, the consideration and the medium are no longer separated. That gap is where product quality silently degrades.
The piece does not skip the uncomfortable part. Frontend developers whose primary function was translating design intent into code are in a materially different position than they were two years ago. The translation layer is compressing and AI is doing enough of it consistently enough that the economics of the role are shifting. What the author argues survives that compression is engineering judgment: assessing whether a contribution breaks the architecture, whether a codebase remains maintainable in two years, the calls that require understanding the full stack. Whether that argument holds as the tools improve further is the question this piece raises without fully resolving, which is exactly why it is worth reading.
[READ ORIGINAL →]