

I want to hate on this guy, but at the same time this is just the reality of a lot of workflows these days. Most programmers I know professionally will do boilerplate work with Claude Code (like the chainsaw analogy he gave) and then do the more meaningful work themselves. Same for automating commit schedules.
Especially for small low-impact projects like Lutris (linux community loves it, but its just 3 people), if you’re going to maintain development speed, involving AI automation is probably going to be more of a positive to you than a negative. The real issue would be if the AI use snowballs and they unrealistically increase the scope of their project as that’s when most projects actually start to die.

Because development isn’t exactly asynchronous by nature. If you are waiting on placeholder assets, you are blocking everything dependent on “what comes next”. Even at the cost of going back to repopulate your assets with non-placeholders, you save a tremendous amount of time.