Seven years as a UX designer reduced to a single job: translator. Wireframes to engineering, then wait. That ended in 2024. The author now runs Claude Code, Figma Make, and ChatGPT projects as primary tools, replacing static Figma deliverables with working demos, connecting design systems directly to Claude Code for interface generation, and running research synthesis inside ChatGPT. The role shifted from translator to conductor: issuing direction, evaluating output with senior judgment, staying hands-on where it counts.
The core argument is about tacit knowledge, and it is the most useful part of this piece. AI handles explicit knowledge, anything documented, named, and searchable. What it cannot access is the accumulated, inarticulate reasoning a designer builds over years. The author's 3C framework addresses this directly: Context (the full project background the AI reads every session via a dedicated file), Components (the specific tools, skills, or MCP configurations you hand the AI for specialized work), and Criteria (output standards including negative constraints, what not to generate, plus a self-audit prompt to catch generic AI defaults like blue-purple palettes and Arial fonts). The framework is practical and immediately applicable.
The piece also draws a hard line between Figma instincts, spatial and visual, developed by dragging shapes, and real product design sense, which lives in systems and time: how interfaces behave under real data, animation easing, interaction chains. These require actually running code yourself, not watching tutorials. The author includes exercises for building that muscle, a design system integration walkthrough, and a section on scaffolding your own AI workflow. Worth reading in full for the 3C breakdown and the honest account of what seven years of experience actually transfers to in this new model.
[READ ORIGINAL →]