Challenge

A team all over the map — and a landscape that wouldn't sit still

Half the team was new to AI, a few resisted it, and the design tool landscape was shifting under us mid-evaluation.

I coached the team on the basics and built GPTs to guide our process and raise quality. Then Figma released Agents, Claude released Claude Design, Google launched Stitch, and our engineers adopted Cursor. In hindsight, I'd move faster next time.

Solution

Test it myself, then let the team own the recommendation

I tested on personal projects first, then let two senior designers own the recommendation.

At home, I tested how little Figma work I could do before jumping into code.

Each time, some design exploration in Figma was critical. Sometimes that meant a UI mood board, a few key screens, core components, or some mix of the three.

Then I gave the team room to find their own process through coaching, experiments, and lunch-and-learns where designers taught each other. To decide, I picked our two most experienced, most curious designers and gave them a month to evaluate tools in their real project work. They ran ideas by the team and delivered the final recommendation.

That let us cut through the noise and land on an approach built for how we actually work, not what was trending that week.
Results

Daily AI use — and buy-in that stuck

Every designer uses AI daily, and we've standardized on Figma, Claude, and Cursor.

AI now shows up in research, copywriting, competitive analysis, idea generation, prototyping, and design. More detailed prototypes are producing sharper insights from user research, and the process keeps evolving.

The bigger shift was buy-in. Nobody felt this was handed down to them. They felt like they built it.