Design spent a decade optimizing for the wrong thing. A Head of Design at Byte by Yum!, formerly at Walmart and Capital One, argues that design teams built their own displacement by chasing agile velocity metrics: screens shipped, handoff speed, design system compliance. When AI tools like Figma AI and Claude can now generate mockups without a Slack message, the profession is discovering that throughput was never the actual value. It was just the only part that was easy to measure.
The article's sharpest claim is structural, not sentimental. Design didn't lose strategic relevance because organizations devalued it. Design traded that relevance incrementally for salary bands and sprint approvals, selecting for designers who produced ten high-fidelity mockups by Thursday and calling it a career path. The judgment that catches a failing system before a line of code is written, the instinct that holds the whole product in tension while everyone else stares at one feature: those muscles atrophied because the job never asked for them. Peter Merholtz made this argument first. This piece extends it with inside testimony from someone who ran the performance reviews.
What makes this worth reading in full is not the diagnosis. It is the specific mechanism: AI automates what is lowest-value first, and for design, that was UI production, because UI is repeatable and describable enough to train a model on. The author's answer is not to learn better prompts. It is to rebuild the capabilities design abandoned, prevention over production, systems thinking over screen factories. That argument needs to be interrogated in full, because it raises a harder question the piece only begins to answer: what does a design organization actually look like if the interface is no longer the proof of work.
[READ ORIGINAL →]