AI lets you try more ideas than you'll ever ship. Each one can teach you something. Combine what you learn and that's what ships. Ideally something of value to users.
Thariq
@trq212
Sep 23
the right way to use model capabilities is not to ship 10x more features to prod it's to spend more time understanding your users, trying experiments, building prototypes, learning about things you don't understand so that you can ship things that actually work

Sep 23, 2026 · 5:00 PM UTC

29
11
210
25,193
Sort replies: Relevant Recent Liked
Replying to @addyosmani
what's the kill criteria for an experiment when building it cost nothing
1
1
282
At what point has it got diminishing returns (not providing more signal to your original question)
2
2
349
more replies
Replying to @addyosmani
more ideas tried isn't the flex people think, most of mine are still trash, just cheaper trash now
1
152
Replying to @addyosmani
Most of my tries never leave a branch and that's fine
79
Replying to @addyosmani
💯 thats how I’ve been using it. hv it generate learning notes (with sources) as static HTML, then I spend the time on layout and feel. Same loop for app mockups. Most of it never ships. The useful part is the taste you pick up from the ones that almost work.
82
Replying to @addyosmani
idk man, it lets everyone else try my ideas too. with models getting so good they can probably one shot my microsaas lol
39
Replying to @addyosmani
The advantage is testing ideas before spending months building something users might not want.
57
Replying to @addyosmani
The throwaway experiments are where the learning actually happens.
38
Replying to @addyosmani
the tricky part is writing down what each experiment taught you before starting the next one. do you keep those notes somewhere or just carry them in your head?
1
3
Replying to @addyosmani
More ideas tried also means more that quietly die without a check on them. The lesson only counts if something verifies what you learned, not just what you remember.
3
Replying to @addyosmani
true
14
Replying to @addyosmani
How do you know if you ship too fast? Feynman after every major step?
34