Right. Be the person on your team who just fucking doesn't because you'll learn your problem domain while everyone else is doing unpaid tool testing work for AI companies.

One day these things will actually do what they are supposed to do with a measure of consistency that doesn't involve GRIMOIRE.md or whatever and you can use them then. And most of the early mover advantage will be gone, because LLMs will not be the winning technology.

In the meantime be the person who learned the lessons of social media: popular isn't the same as good, appropriate or sensible.