Comment by ChrisMarshallNY

I’ve become comfortable with using LLMs as “trusted advisors.”

I am not [yet] ready to just let an agent write a whole app or server for me, but I am increasingly letting them write a whole function for me.

They are also great “bug finders.” I can just feed some code, describe the symptoms, and ask for an observation. I often get great suggestions, including things like finding typos and copy/pasta problems.

I find that just this limited application has significantly increased my development velocity, and, I believe, the quality of my work.