How do you actually orchestrate your AI agents?

Hey everyone,

Right now, I usually start separate Claude Code or Pi sessions and manage each agent individually, deciding what to delegate and when to jump in.

I'm curious how other people are doing this in practice.

Do you have some simple way of orchestrating multiple agents, delegating tasks, keeping track of what they're doing, etc.?

I'm specifically not looking for huge workflows or overly complicated agent architectures. I'd rather keep things simple and actually useful.

What does your setup look like?