It's both amazing and painful to watch codex use browser + computer use to open Chrome, go to my PR, tap on comment and wrangle with the macOS picker - all TO UPLOAD AN IMAGE. GitHub has no API doesn't stop anyone.
I let my codex run in VMs so they don't steal app focus.
I can't help but imagine how incredibly productive, fast and more accurate than AI, humans would be if people got off our balls and let us focus on one thing at a time. Interrupt your agents withj something that came through on a heartbeat while mid-task. See what 2 subjects do.