Skip to content
Article Figma Blog Oct 2026

Figma Sightlines no.3: Raise the bar for how teams use AI

Figma’s Sightlines is a recurring series for design and product leaders. The third issue, published in October 2026, addresses a problem that has become common in organizations where AI adoption is underway: individual contributors use AI tools productively, but the team as a whole moves only a little faster, and the quality of decisions hasn’t improved.

The issue collects perspectives from participants in Figma’s Leadership Collective events and frames them around three connected claims.

Speed is not the goal. AI can generate more output faster, but more output moving in the wrong direction is not progress. The suggestion here is to redirect AI use toward exploration rather than production — running more design variations to find the best direction before committing to it, testing edge cases earlier, and asking AI to challenge assumptions rather than execute them. Several contributors noted that they used AI to produce a first draft and then stopped, when the greater value was in using it to generate five versions and understand why one worked better than the others.

Adoption is a team practice, not an individual skill. A designer who has built a sophisticated prompting workflow does not automatically make the team better. Learning that stays with one person creates gaps. Contributions that transfer across the team — shared prompt libraries, internal demos of workflows that worked and ones that didn’t, AI sessions integrated into existing critiques or planning meetings — move the whole team forward. Leaders who use the tools themselves, and say so, signal that the learning is expected and valued.

Protected time to experiment matters. Two participants in Leadership Collective events described creating intentional space for low-stakes AI experimentation: weekly sessions where the goal is to learn something about the tool rather than ship something. These are not hackathons. They are small, regular, and attached to real project questions rather than invented ones. The signal they send to the team is that figuring out what doesn’t work is a legitimate use of time.

The piece is short and practical in format, written for managers and team leads rather than individual contributors. It is most directly useful for people navigating the early-to-mid stages of AI adoption — past the initial experimentation phase, but not yet at a point where AI practices are routine across the team.