THESE ARE THE TWO QUESTIONS YOU SHOULD ASK CLAUDE AT THE END OF EVERY AI SESSION TO CATCH THE STUFF THE MODEL SILENTLY SKIPPED the whole point is forcing the ai to admit what it rushed past instead of pretending everything is fine 1\ "what are you least confident about right now" it'll list 6 or 7 things it didnt properly check about one in four times, one of them is a real problem it took action on without ever understanding. then you just have it go investigate each one. 2\ "whats the biggest thing im missing here, what dont i realize" this ones from sam altman. it pulls back and catches the blind spot in how youre thinking about the whole thing, not just the code. some other ones that people added on that are just as good: > "if this breaks in 3 months, whats the most likely reason" catches assumptions that wont survive scale > "if you could add one unrequested thing that would make this industry leading, what would it be" for actual creative ideas > "what could i have done differently to make this session more efficient" to fix your own prompting the funny part is we now have to literally ask ai what it skipped past but it works