I develop ideas through working prototypes and carry design decisions into production. Working in code lets me test assumptions early and tune interaction details against real behavior. It also gives me a practical understanding of the tradeoffs: where complexity is worthwhile, where a simpler approach will do, and which details deserve more time.

Today, AI is central to that process. I use agents to explore approaches and implement them, and put care into the environment they work in: repo structure, configs, lints, and domain context. I want the workspace to handle as much routine correctness as possible so I can give more attention to architecture, visual weight, and interaction feel.

I stay responsible for the result. That means reading the code, using the product, and questioning choices that look reasonable at first glance. Generating more software makes that judgment more important.