A widely used Mac productivity app has added generative artificial intelligence in a way that keeps processing local to the device, protects user privacy, and lets people shape how it works for them.
Many major tech firms have been pushing AI into their products with flashy interface elements and email campaigns that urge users to try new features. That approach stands in contrast to the one taken by indie developer Omni Group, the Seattle-based team behind OmniFocus, a task manager long praised for deep flexibility and a clean interface. If you can picture a system for handling tasks the way you want it, OmniFocus usually lets you put that system into practice without creating clutter.
Omni’s plan for artificial intelligence follows that same philosophy: the tools run offline, keep data private, and are available for users who want to configure them. You won’t find intrusive prompts or banners in the average OmniFocus install asking you to try AI. Instead, the capability shows up as a resource for people who build automations or who install automations made by others. A handful of user-created automations are already available in the project’s directory.
Trying one out requires running one of Apple’s new “26” operating systems—macOS, iOS, or iPadOS are supported. Those releases include a fairly under-discussed capability in macOS 26 that allows third-party apps to call Foundation, the large language model that powers Apple Intelligence.
You’ll also need a current copy of OmniFocus; at the moment it’s the only Omni application that supports these AI features. Company representatives say similar functionality will be added to other titles in the suite, including OmniPlanner and OmniGraffle, down the line.
From there, visit the Omni-Automations directory and pick a tool that looks useful. The page displays the automation’s source code, and an Install Plugin-In button sits above the snippet for adding the plug-in itself. Depending on your system settings you may have to enable scripts from external applications before you can complete installation.
One automation, Help Me Plan, will take an item in your inbox and expand it into smaller actionable steps. I tested it with a task titled “Write about OmniFocus Automation features,” and the plug-in quickly created subtasks that ranged from research to drafting to final edits. Those particular steps didn’t match my normal writing routine exactly, but the point is to get momentum when you feel stuck.
My favorite automation, Clipboard Events, converts whatever is on your clipboard into one or more tasks, adding a clear description and suggested due dates. It’s an efficient way to turn a long block of text into a short checklist of next actions.
Users do not need to understand the inner mechanics to take advantage of these automations. What stands out is how much effort Omni has put into documentation: guides, a video walkthrough, and notes showing how the automation hooks into Apple’s new Intelligence APIs and how Omni’s own APIs tie into that system. The source for every automation is visible before installation, and most scripts are brief enough to review quickly.
This setup feels aimed at people who want to build on top of the AI to fit their own routines rather than being steered by a one-size-fits-all assistant. “This is ignition, not even liftoff,” said Naomi Pearce, who does public relations for Omni. “It's very early days.”
I’ve written before that chat-based interfaces may not be the primary way most people use AI long term, and I’ve pointed to indie work that treats machine intelligence as a background tool. Substage, for example, lives under the Finder and lets you manipulate files by typing plain text commands. That app puts the language model to work without making it the center of attention. Omni’s automation approach follows a similar idea: once a plug-in is in place, a single click runs it and the technology mostly disappears into the background.
I hope all AI becomes similarly invisible in time.