The "five-minute AI task" that wasted 40 minutes of my life
by•
I used to plan my lunch breaks around AI tasks. Pick the one task that's easy enough to just hand off, start it, go eat.
Came back more than once to find the agent had stopped two minutes in over some dumb question. Not even a permissions thing, it just sat there waiting for me to say "go ahead". Didn't even know it was stuck until I checked.
Anyone actually solved this or are we all just checking constantly?
112 views
Replies
too relatable
It's still checking constantly...
Yep. The task itself takes 5 minutes, but babysitting the AI takes 40.
i run most of this in yolo mode now. approve the plan once, scope permissions tight, let the agent work. checking constantly defeats the whole point
yolo mode plus 'continue without confirmation' in chat settings cuts most of the dead waits. you trade some safety for it so set up scoped permissions first (no commits to main, no destructive shell, no env file access).
the surprisingly bigger win for me was a one-line addition to the system prompt: 'never ask clarifying questions. instead make the best guess and clearly label every assumption you made.' you can correct on the next turn if it guessed wrong. one wrong guess plus one correction is way faster than three back-and-forths.
bonus: at the end of every loop ask the agent for a 3-line 'what i changed and why' summary. ten seconds to read, catches drift before it compounds.
Pfff, not me. I'm casually taking my laptop with me just in case...
We all are still checking constantly.
@alex_dochioiu This is the annoying gap between automation and delegation. If the agent stops on a low-risk question and doesn’t notify you clearly, you’re not really saving time — you’re just moving the waiting around. I think agents need better “continue unless risky, pause only when it matters” rules.
Upvoting this. It's actually becoming frustrating
I use a remote control app on my iphone when I need to connect to my machine on the go