Yash Ukalkar

[Yash Ukalkar @yashUkalkar]

@oxcelato @catalinmpit Also, it's obviously not expected for them to, the million dollars should be put into making the experience as fast as possible. That's what big techs do, but at their scale shaving every milliseconds off is a huge effort.

[Yash Ukalkar @yashUkalkar]

@oxcelato @catalinmpit What I mean is difference in redundant code, something like the model writing code making 3 separate db calls where 1 is needed. This shouldn't be done irrespective of the usage metrics, but happens when you over rely on AI.

[Yash Ukalkar @yashUkalkar]

@oxcelato @catalinmpit Agree partially here, they earn with laggy and buggy products, but theirs face the issues because of volume, be it a lot of data or a lot of features to ship.
It shouldn't be laggy/buggy when you're working with 30 customers and 10-15 features of which 3-5 are used.

[Yash Ukalkar @yashUkalkar]

@catalinmpit @sandislonjsak I also use similar simple conversational prompts when the task at hand is fairly simple. It should be sufficient. If I need to prompt a 100 lines for the model to do what is asked for a scope this small then 🫠....

[Yash Ukalkar @yashUkalkar]

@catalinmpit @sandislonjsak Please do reply to this. I don't mean it in any sarcastic way or trying to put either points down, just want to see if there's something I'm missing or can do better.