The RAD team structure was intentionally lean: a product manager, backend engineer, frontend engineer, quality engineer, a field lead to connect us with store associates, and me as the product designer. We were pulled from our regular teams and given permission to ignore other company responsibilities. This was rare—total focus, total dedication. We also received AI resources, though they ended up being more hype than help.

Six weeks. That was our constraint, our challenge, and honestly, our excuse to move fast and break the usual rules.

The research phase had to be aggressive. I conducted generative interviews across different states because DMV requirements vary wildly—what works in California doesn't work in Virginia. One in-person trip, followed by eight video calls in week two, gave me the foundation I needed.

What I heard was pain. Real, visceral frustration. Associates described the old system with the kind of exhaustion that comes from fighting broken tools every single day. But I also heard something surprising: when I showed operational partners ways we could remove unnecessary steps, they were eager to simplify. I'd expected pushback, bureaucratic resistance. Instead, I found allies who were just as tired of the complexity as the frontline associates.

The validation came through emotion. When associates saw early prototypes during our four usability testing calls, their relief was palpable. One associate literally said she wanted to cry—in a good way. That kind of reaction tells you everything about how badly the old system was failing.

Here's where things got chaotic in the best and worst ways.

I worked in Figma during the interviews. Not after—during. I'd be on a call, listening to an associate walk through their workflow, and I'd be adjusting the prototype in real-time. The moment the call ended, I'd huddle with the team to iterate before the next round of calls started.

This approach was exhilarating and exhausting. We called it "echo exhaustion"—trying to rapidly incorporate new learnings into prototypes led to overwhelming fatigue. I shifted to extreme working schedules and burned out early. The continuous pressure from company demands, combined with being isolated from other teams, meant key decisions were being made without the usual collaborative dialogue.

What I'd do differently: At the start, we should have created a shared playbook outlining work preferences and necessary focus times. We recorded meetings but full transcriptions proved minimally useful. We made the right call avoiding AI for prototyping—our existing design frameworks (CarMax's Material UI and Associate Tool Framework) were sufficient, and AI would have just added documentation overhead for developers.

I leaned heavily on principles that kept me grounded: "Design like you're right, test like you're wrong." And I thought beyond the six-week timeline, focusing on how this project would connect to the larger ecosystem and persist into the future. Less is more became my mantra when facing feature creep.