@arpit_bhayani yes ! The "you are absolutely right" statement given by the agent when it goes wrong, is actually an indication that its our prompts which were wrong/vague at the first place.
One thing I have noticed is, the code review AI does is good to catch an obvious bug but the code review that your senior teammate does is at a different level.
They will point out a very logical change in code design (for eg, promoting extendibility , structure etc) that AI simply wouldn't catch.
Hence my team follows the culture of AI review + human review , both non-negotiable. π
I remember having spent one whole day debugging failing tests. Getting the green check of "passed" had its own satisfaction in the pre AI era.
Now a single prompt and AI gets the tests passed. It is still satisfying, just differently π
#nostalgia#ai#coding#tech
@TTrimoreau I think not just startup, but this may apply to even smallest of the features in an existing product - finding the right idea and keeping the users >>>> others , in terms of difficulty.
@devops_prashant + 1
I recently saw someone say on X, that AI is a multiplier, so if one is at zero , it will just multiply it. couldn't agree more with that.
@0xlelouch_ misconfig in my files due to which my local API was calling stage DB instead of local DB , and I was wondering why it doesn't update DB ://
@rishikagupta__ Sometimes, honestly, I do miss that phase. It was slower, but had its own enriching experience. With AI as well, the experience is enriching , just that in a different direction.