AI Process
AI runs through every step of my process. I use it to move faster and think wider, with a human-in-the-loop review at each step to keep the work consistent. The tooling changes constantly; the judgement doesn’t.
My three principles for designing with AI
Intent before output: Strategy and context are the foundation of every project. Defining scope, references, and constraints up front keeps AI working inside deliberate guardrails instead of guessing. The output is only as good as the setup.
Keep humans in the synthesis: I deliberately don’t hand qualitative research and analysis to AI. Sitting with the interviews and the messy detail is how empathy forms and how the real problem surfaces. Automating it strips out the emotion the work depends on — keeping the emotion in the loop.
Prototype fast: The long, frontloaded UX process is mostly over. Rather than documenting for weeks before anything is visible, I prototype early, get something real in front of people, and let feedback steer direction. Momentum beats polish in the early stages.
Tooling
What I reach for at each step, from intent to ship. The stack shifts every few months; these are the ones in use now.
- Claude Code · Intent, build and verify
- Notion · Docs and specs
- Mobbin · UI references
- Figma · Foundations and design
- TSTokens Studio · Design tokens
- Paper · Frames and states
- Pencil · Frames and states
- Cursor · Editing in code
- React · Screens in code
- Tailwind CSS · Styling from tokens
- shadcn/ui · Component base
- HeroUI · Component library
- Base UI · Headless primitives
- Motion · Interface animation
- anime.js · Timeline animation
- GitHub · Repo handover
- Vercel · Live URL
Resources
