Is agentic design living up to the expectations?
Postdigitalist Research · 9 November 2026
Read the originalPostdigitalist reviewed the evidence on AI agents in design and found that measured time savings, about 20% in a Figma trial, come from tasks with a defined target.
Postdigitalist, a site that publishes research on AI and design in English and Spanish, examined how far AI agents that change designs, prototypes and code save design teams time. It read a Figma trial with 100 participants, a survey of 906 designers and accounts from DoorDash, Watershed, Stripe and Samsara. Its report separates what these sources measured from what is still unmeasured.
- In a preprint by four Figma researchers, designers and product managers using Figma Make, Figma's prototyping tool, took roughly 20% less time on tasks they completed. Each task came with a reference image.
- Access to Make raised completion of the hardest task by 19.9 percentage points, so the average time compares partly different groups of people who finished.
- At DoorDash, a designer's code change goes through engineering approval and then a design review. At Watershed, every designer submits changes to production and engineers review the code.
- In a study of 280 designers presented at CHI 2025, language models did not improve how well designers reframed a problem. The gap between experienced and less experienced designers widened.
The authors advise a team piloting a design agent to compare one recurring task with and without it, counting review and fixes until the change can be approved. Several company accounts come from a report by two investment firms whose partners include Stripe.
Compare one recurring design task done with and without an AI agent, timing the first version, the review and the fixes until the team can approve the change.
Name who may approve and who may revert a change made with an AI agent, and set technical permissions to match.
Derived by Working Surface from the article; more in the Playbook. Source line: The key measure is how long it takes the team to get a change it can approve.
14 November 2026: as AI agents produce more code and tests than people can inspect, teams are writing down which work gets checked closely, by whom or by what, at what cost, and how they will know it worked.