Claude Managed Agents vs. Claude Remote Routines

So I've been building out my own personal AI operating system, and I've been confused because I've been asking myself the question: what's the difference between Claude managed agents versus Claude remote routines? The way I have it set up right now is that I run a remote Claude routine that wakes up an orchestrator agent, which tells the other agents to get to work. I see Claude managed agents. You can create agents inside of there that also are able to run on Claude's infrastructure, so I'm not really sure which platform to use or what the differences are. Anyone run into this?

2 comments