Yeah, second that. This is just release management best practice.

And it makes total sense to me, it’s a way of organizing your dependency graph by the lifetimes of your components.

This will also simplify testing, development, and audit. You won’t need to recompile autogen schemas as often as the business logic implementation anyway. Depending on artifacts pushed through a release pipeline is even more reliable. You can QA everything in guaranteed isolation while keeping it conveniently a workspace monorepo.