I’m in ops and my director said they wanted product to be able to prompt engineer our system. I said we have a huge distributed cloud native system that interacts with services in other AWS accounts and all that. He was like, yeah, he wants me to at a whim spin up a new environment for them to do it.

I was like, what happens when they prompt something that fucks the VPC or security groups or configuration, who tracks that? what about the resulting PRs? who are they going to harass if something doesn’t work? am I going to be wasting my time debugging an AI slop environment? I don’t even have the time to fully implement the observability I want in prod and I’m supposed to waste my time supporting this bullshit?