Hey family.
This week I sat with three pieces by Herajeng Gustiayu, a product designer who leads a team in travel tech. I think they're the clearest writing I've seen on what our job turns into when prototypes get cheap.
I think about it like a walkthrough at practice. A play can look perfect at half speed with nobody guarding it. That doesn't make it the play you call in the game.
Her first point: how finished a prototype looks and how decided the team is used to move together. They don't anymore. An AI flow can look ready for engineering before anyone has actually chosen it. Her fix is one question in every review: what would make us abandon this direction? If the room shrugs, nobody wrote the assumption down.
Her second: AI made exploring cheap, so the skill moved from making options to choosing them. Bring two or three to a review, not six. And when a stakeholder asks for something you disagree with, build their version plus two alternatives, then say which one you'd run and why.
Her third hit me hardest. Her team built a tool that turns a brief into a first-draft screen that fits their design system in under 30 minutes, no designer needed. The list of places where designers still had to be involved turned out to be a list of where judgment lives. In her words, "Some people who thought they were senior will realise they were just fast."
Fast at half speed isn't the same as knowing which play to call. That's the side of the line I want all of you on.
Try this.
Next prototype you share, yours or one an AI made, put one line under it: "We're assuming ___. We'd drop this if ___." Do it three times this week and post your favorite in the comments.
Her three pieces: