{"id":48507,"date":"2025-07-16T16:48:28","date_gmt":"2025-07-16T20:48:28","guid":{"rendered":"https:\/\/mjtsai.com\/blog\/?p=48507"},"modified":"2025-07-16T16:50:54","modified_gmt":"2025-07-16T20:50:54","slug":"shortcuts-in-macos-tahoe","status":"publish","type":"post","link":"https:\/\/mjtsai.com\/blog\/2025\/07\/16\/shortcuts-in-macos-tahoe\/","title":{"rendered":"Shortcuts in macOS Tahoe"},"content":{"rendered":"

Jason Snell<\/a>:<\/p>\n

\n

In macOS 26, there’s a built-in clipboard manager that can be accessed from the Spotlight interface, and a new set of Shortcuts triggers let you run automations when events occur on your Mac or at specific intervals.<\/p>\n<\/blockquote>\n\n

Simon B. Støvring<\/a>:<\/p>\n

“Folder” and “File” seem like interesting automation triggers in Shortcuts for Mac. You can do things like “When a file is moved to this folder, process it with on-device AI”.<\/p><\/blockquote>\n\n

Jason Snell<\/a>:<\/p>\n

\n

Of all the features I’m excited about using in macOS 26, the one that most intrigues me is the Use Model action in Shortcuts. Use Model does exactly what you think it does: you toss data into it, and an AI model somewhere (on your Mac, on Apple’s Private Cloud Compute servers, or even at an OpenAI server farm) will take that data and turn it into… something.<\/p>\n

The other day, I realized that this new feature would allow me to expand my existing automation that uploads images to the Six Colors web server by adding a description of the image.<\/p>\n<\/blockquote>\n\n

Dan Moren<\/a>:<\/p>\n

But pulling information out of a document—especially information that might appear anywhere in a variety of forms—seems like something an AI model would be good at, so I decided to take another crack at it with Shortcuts’s new AI capabilities.<\/p>

I started out my workflow by grabbing all the text from a PDF or web page, then passing it to the Private Cloud Compute model. (I attempted to use the On-Device model at first, but it was both very slow and not quite as good at formatting the response in the manner I wanted.)<\/p>

[…]<\/p>

And therein lies the rub with all of this. The results are neither reliable nor necessarily repeatable. The same data run through this shortcut multiple times provides different answers: I’d think that anathema (not to mention madness inducing) to the sensibilities of any programmer. Given the same data, the algorithm should yield the same thing every time, but the non-deterministic nature of AI models throws that out the window.<\/p><\/blockquote>\n\n

Federico Viticci<\/a>:<\/p>\n

So, the ChatGPT integration in Shortcuts’ Apple Intelligence action for iOS\/iPadOS\/macOS 26 appears to be the really old GPT 4 Turbo with a knowledge cutoff date of November 2023…?<\/p><\/blockquote>\n\n

Federico Viticci<\/a>:<\/p>\n

The greatest threat to Apple Intelligence’s App Intents adoption isn’t the underlying lack of an LLM (Apple can fix that sooner or later): it’s web apps and the rise of MCP. The automation and inter-app model is shifting from local extensions to web-based ones.<\/p>

I’m imagining an Apple-made MCP bridge that runs in Private Cloud Compute 🤔<\/p><\/blockquote>\n\n

Previously:<\/p>\n