Skip to content
Article Figma Oct 2026

Figma: 3 ways product design teams use the agent after GA

Published on the day the Figma agent exited beta, this article collects three concrete use cases from product design teams that participated in the preview and have since moved to daily use. The cases cover design system documentation, critique workflows, and motion design, and they share a consistent underlying pattern: teams that saw the most value defined specific tasks and constraints in advance rather than relying on open-ended prompting.

Uber’s design system team uses agent skills — custom scripts that the agent executes as named commands — to automate component documentation. Skills like /create-anatomy and /create-color generate usage guides and flag hard-coded values that should be replaced with design tokens. Staff Product Designer Ian Guisard describes a task that previously required multiple people over months as now taking a single designer an afternoon. The workflow works because the team encoded its own documentation standards into the skill rather than expecting the agent to infer them from context.

Granola, which makes AI meeting notes, uses agent connectors to pull product feedback directly onto the Figma canvas during design critiques. Without the connector, noting feedback manually costs attention mid-conversation; with it, annotations appear in the file in real time. Product Designer Paavan Buddhdev calls it a significant time saver because the act of capturing feedback no longer interrupts the flow of the critique itself.

Atlassian built animated illustration components that designers with limited motion experience can customize within defined parameters. Senior Product Designer Alexandra Pereira describes the result as maintaining consistency, accessibility, and quality while reducing the skill threshold for motion work. The agent did not create the motion system — it was used to help build the component infrastructure that made the system accessible to a wider range of designers on the team.

The three cases are useful for design teams evaluating whether the Figma agent is ready to add to a real workflow. None of the use cases required teams to change how they work at a fundamental level. What they required was time spent building the skills, connectors, and components that gave the agent concrete tasks to execute. The agent performs best as part of a prepared system, not as a general-purpose assistant called ad hoc.