I use AI to shorten the distance between thinking and testing.
Not to replace product judgment. To get to something we can react to faster.
Research → patterns
Turn interview notes, feedback, and messy qualitative data into themes worth investigating. I still read every transcript. AI just stops me from losing the third theme because I got attached to the first one.
Idea → prototype
Move from product conversation to interactive concept while the context is still fresh. A stakeholder describes something on a call. By the next call there is a thing to click.
Prototype → code
Explore actual interaction behavior instead of stopping at a polished Figma frame. Scroll behavior, empty states, loading, error handling, the parts a static file quietly skips.
Complexity → clarity
Pressure-test requirements, edge cases, flows, and documentation before they become engineering problems. Ask what breaks, early, when it is still cheap to change.
The point isn’t to make more things faster.It’s to learn faster.
Where it actually helps, and where it doesn’t.
Helps
- Getting to something clickable in a day
- Finding the edge case nobody wrote down
- Turning a decision into documentation engineering can read
- Testing whether an interaction feels right, not just looks right
Doesn’t
- Deciding what is worth building
- Knowing which user complaint is the real one
- Sitting with a customer while they explain their week
- Telling a stakeholder no
What I actually reach for.
Claude for research synthesis and pressure-testing requirements. Figma and FigJam for design, Claude Design when I would rather move something by hand than describe it, Figma Make when an idea needs to be clickable before it needs to be right. Claude Code for prototypes developers can start from, and Conductor when I want several running in parallel on the same problem. FullStory and session recordings for watching what people actually do. Kling for motion.
None of it decides what is worth building. That part is still the job.
Have a complicated problem?
Those are usually the interesting ones.