Skip to content
Article Figma Oct 2026

Figma: Four team rituals that move AI from individual experiments to shared practice

Published on Figma’s blog on October 10, 2026, this article by Emma Webster addresses a gap in how AI adoption is usually discussed. Most guidance focuses on individual techniques — which prompts to use, which tools to try. This piece focuses on the team layer: how to move from a handful of enthusiasts experimenting independently to a team where AI use is a shared, expected part of the workflow.

What the article covers

Webster draws on design and product teams at four companies — Figma itself, Sephora, Mercari, and Atlassian — and identifies four rituals each used to make AI practice spread:

Peer-led study groups. Jake Geller, a design operations manager at Figma, organized a program called AI Collective. He identified skills designers wanted to build, split them into five groups by topic, and had each group meet for one hour per week over six weeks. Sessions followed a template: about 20 minutes of teaching and 40 minutes of hands-on practice. A Slack channel supported ongoing questions between sessions. The key factor was a concrete goal — building a specific plugin rather than exploring AI generally — which made learning tangible and made peers comfortable asking questions they wouldn’t ask in a larger setting.

Process-focused sharing forums. Sneha Narahalli at Sephora runs a biweekly meeting where designers share AI projects. The format changed after early sessions showed only polished final results, which left colleagues feeling like their own messier experiments were failures. After the change, presenters walk through their process, including what failed before they found an approach that worked. Figma runs a similar segment called “How I AI” at company all-hands meetings.

Cross-discipline training. Aki Alkhatib’s team at Mercari runs hands-on sessions where designers learn basic engineering tasks — using an AI coding tool, setting up a GitHub repository — and developers learn design fundamentals like visual hierarchy. The goal is not full proficiency but the ability to produce a useful starting point outside one’s own discipline. A designer who can generate a working prototype, even a rough one, and an engineer who understands typographic hierarchy, can both give more substantive feedback when reviewing each other’s work.

Prototype-first critiques. Charlie Sutton at Atlassian changed the rule for design critiques: designers must show a working prototype first, let the group interact with it, and explain their reasoning afterward. Because every critique now requires a prototype, designers practice AI prototyping tools with the frequency needed for the skill to actually develop.

Who it is useful for

Design leads and operations managers working out how to move from patchy individual AI use to something that functions as team practice. The article is practically specific — each ritual includes the structure used and the reasoning behind it — which makes it easier to adapt than general guidance about building an “AI culture.”