Writing, experiments, prototypes, and notes—thinking by building.
Pattern by pattern, force every implicit design rule into words — then measure whether an agent actually follows them.
A framework for evaluating AI in creative work that treats evaluator disagreement as signal, not noise.
The gap between attempt and correction is where expertise lives. That's what AI can actually compress.
Generating UI faster is fine. Understanding what to build in the first place — that's the part AI can actually change.
I'm doing in a day what used to take a week. But figuring out what to build — that hasn't moved at all.
If leadership thinks design is mostly 'making screens,' of course they'll believe prompting a model replaces the role.
A single prompt that reviews any flow through three lenses and returns one consolidated set of findings.
Dump what you know first—PRD, research, gaps—then use AI to poke holes in your thinking, not to design.
“Should this exist?” As AI speeds up execution, understanding what to build becomes more valuable than shipping it faster.
What if skills became shareable superpowers? Not a list of abilities, but executable capabilities others can run, test, and combine.
Instead of handing over static snapshots, we're handing over runnable intent — the way a product should behave.

A single-file PWA that turns any QR or barcode into a pass you can scan, save, and reopen without a browser address bar in sight.

Baby care tracking for exhausted parents—quick logging, logical days, and plain-language insights.

There's so much to remember. We password manage. You forget about the tricky things.

Imagine to know what your industry leaders are reading or watching? Now it's possible. Favbox is a place to discover new perspectives and share your own.