Most designers use AI as a finishing tool, to polish copy, to rename layers. I use it earlier, closer to the thinking. Before a single frame exists, I'm already using it to stress-test a concept, to generate the edge cases I haven't imagined, to pressure an assumption that feels too comfortable.

On the HDFC consent project, I used Claude to simulate what a customer with limited literacy would misread in a data-permission flow before I built anything. Not to replace research, but to move faster to the right research questions. On the AI extraction engine in Dubai, I was processing competitor workflows, architectural constraints, and trust models simultaneously, and AI became the environment I thought inside of, not a tool I opened afterward.

For feature exploration I rely on v0 and Lovable to turn rough ideas into something you can click through. The value isn't the fidelity, it's the speed at which a speculative idea becomes something stakeholders can react to. A prototype someone can click through closes arguments that wireframes leave open.

Figma stays at the center of anything that goes to engineering. But the hours before that, the ones where I'm deciding what to build and why, that's where AI earns its place. I care about the decision more than the pixel, so I spend my most careful attention there, and I use every available tool to get there faster.

I also built this portfolio end to end using Claude Code, which is part of the same conviction. I don't think tools are beneath a designer. I think knowing which ones extend your thinking and which ones replace it is the judgment that matters.