Affirm

The Capital team members I was designing for understood their domain deeply: the mechanics of the deals, the language, the edge cases. I didn't, not at first. What I learned was that earning their trust meant demonstrating that I understood their workflow, even if not the underlying finance. The redesign reflects that shift. The original tool was organized around data structures that made sense to engineers; the new one is organized around how the Capital team actually thinks about their work. Getting there required paying close attention in stakeholder sessions and asking questions that probably sounded basic. The answers shaped every navigation decision I made.

Working with Cursor changed how I think about my role as a designer. When an AI is generating code from your prompts, the design lives in how precisely you can describe your intent and in your ability to evaluate whether the output matches it. I found myself being more explicit about decisions I'd previously made on instinct: why this component behaves this way, what the conditional logic should be at each step, how state should change on user action. Explaining my process to an AI collaborator made me more deliberate about it.

It also required enough programming literacy to evaluate output quality. I didn't write the code, but I needed to recognize when the structure was clean and when it would create problems downstream. That technical legibility, understanding what good code looks like without being able to write it, turned out to be as important as the design work itself.