I tried putting a MacBook Air next to my main machine and working across both at once. I expected a second monitor. I got something more useful.
Not a second screen, a second machine
The usual way to use a spare device is to mirror or extend the picture onto it. That is what Sidecar does with an iPad and AirPlay does with a Mac: the second screen becomes a passive surface while all the computing stays on the main machine.
What caught my attention is the other mode — Universal Control. Here the MacBook Air remains a full computer with its own processor, its own memory and its own set of running apps. Only the mouse and keyboard are shared. You work across two machines as if they were one desk, but the load is genuinely split between them.
One mouse and keyboard, two computers
The cursor reaches the edge of one screen and continues onto the next. The keyboard follows it. No hotkeys, no KVM switch, no extra software.
It feels exactly like working with a regular attached monitor, right up until you remember there is a separate operating system with separate windows over there.
What moves across, and what doesn't
The main limitation: application windows do not move between machines. This is not an external display, so you cannot drag an open editor from one Mac to the other. That is the part I miss most.
Files, on the other hand, do move, and it works better than you would expect:
- images, PDFs and other documents;
- ordinary folders;
- a selected chunk of text.
The catch is that you usually have to drop the content straight into a compatible app — from the desktop of one Mac right into an open messenger or notes window on the other. Transfers go over Wi-Fi and Bluetooth, so a heavy video will keep you waiting, while working documents and images land almost instantly.
Where the real gain is
The setup pays off most when a single project has several tasks or several AI agents running at once. Long builds, test suites, indexing, background agents — all of it can move to the second machine while the primary one stays responsive.
You put a second computer's resources to work without loading the first one at all.
Instead of splitting one processor across everything, you get two independent pools of resources under a single set of controls. For agent-heavy work that turned out to be close to an ideal trade-off.
If you are building a similar workflow or automation around AI agents, get in touch — I am happy to compare notes.