Using ChatGPT to create Time Machine Backups

From John Carter.

When my M4 MacBook Air was running macOS 26, I spent several days back and forth with ChatGPT developing the Shortcut code to perform unattended nightly backups. It is incredible how little ChatGPT knows about coding for the Mac. I have a M4 Mac mini running macOS 26. The Air was to be backed up remotely up to the mini Time Machine drive, which was partitioned for the two backups. The same Shortcut code worked for both the Air and the Mini with only minor changes needed for the Air due to way it needed to communicate with the mini.

Then along came macOS 27 Beta that I installed on the Air, but kept the mini at macOS 26. And the backup for the Air stopped working.

The logical thing to do was to ask Siri AI to inspect the error and make a recommendation. It basically said that the method being used to communicate from the Air to the mini was now failing because of new security measures in macOS 27. It took a whole day with Siri AI to try different methods, but finally one of them worked.

Being curious, I asked about the Shortcut for the mini. Would the existing code work with macOS 27? It suggested minor changes, which I tested while the mini was still on macOS 26, and it worked. So, with that change, the mini will continue to backup nightly.

However, when the mini is upgraded to macOS 27, the Shortcut for the Air might stop working due to the change in security on the mini. I’m not concerned, because I feel confident that Siri AI will find a way to make it work.

The other thing that happened when the Air was upgraded to macOS 27 was that the automation to run the Shortcut stopped working. Siri AI to the rescue again! Two methods were suggested, and I chose the one that required using Terminal. Siri AI provided the code and the detailed steps to instantiate it for nightly operation. No debugging needed. And no special programming talent needed - just followed instructions.

Anyone interested in knowing the Shortcut code and the automation to run it, let me know.