To be specific, for the past year I've been having numerous long conversations about all the books I've read. I talk about what I liked, didn't like, the ideas and and plots I found compelling or lame, talks about the characters, the writing styles of authors, the contemporary social context the authors might have been addressing, etc. Every aspect of the books I can think off. Then I ask it for recommendations, I tell it given my interests and preferences, suggest new books with literary merit.
ChatGPT just knocks this out of the park, amazing suggestions every time, I've never had so much fun reading than in the past year. It's like having the world's best read and most patient librarian at your personal disposal.
My experience with plain Claude Code is that I can step back and get an overview of what I'm doing, since I tend to hyperfocus on problems, preventing me from having a simultaneous overview.
It does feel like being a project manager (a role I've partially filled before) having your agency in autopilot, which is still more control than having team members do their thing.
So while it may feel very empowering to be the CEO of your own computer, the question is if it has any CEO-like effect on your work.
Taking it back to Claude Code and feeling like a manager, it certainly does have a real effect for me.
I won't dispute that running a bunch of agents in sync won't give you an extension of that effect.
The real test is: Do you invoice accordingly?