Gemini for macOS Turns Voice Dictation Into a Screen-Reading Assistant

Google has expanded its Gemini app for Mac with a feature meant to replace classic keyboard typing with a conversation with an assistant that can see what's happening on screen. The update, announced on July 29, 2026, introduces single-key dictation activation and an optional mode in which Gemini reads selected files, images, and documents, then edits them itself on voice command.

How the New Dictation Works

The mechanism is simple in concept: the user holds down the Fn key in any desktop app, speaks, and a floating sound wave appears at the bottom of the screen to signal recording. Gemini automatically strips out "um" and "uh," catches corrections spoken mid-sentence, and pastes finished, formatted text exactly where the cursor sits.

Google compares the feature to Rambler, known from the Gboard keyboard on phones in the Gemini Intelligence lineup, where a similar speech-cleanup mechanism has been running for a while and is now arriving on desktop computers.

Screen-Aware Mode

The second, optional layer of the feature requires turning on "Gemini reasoning" mode in settings. Once activated, the app can select a file, image, text snippet, or an entire document open in another program, and on voice command summarize it, rewrite it in a different tone, or compose it from scratch.

In this mode, Gemini also generates and edits images without leaving the app the user is working in, just describe by voice what should change in the graphic or design. The feature can also be launched via the screen-sharing button in the "Ask Gemini" window, not only with the Fn key.

Competing With Apple Intelligence and ChatGPT Desktop

Reviewers note that this system-level integration, a single key that launches the feature in any window, puts Gemini closer to macOS's native capabilities than rival apps. According to the feature's description, Gemini can read the contents of entire folders directly from Finder and insert the resulting text without manual copying, something neither Apple Intelligence nor the desktop version of ChatGPT currently offers.

It's another step by Google toward turning the Gemini app from a classic chat window into an assistant layer spanning the entire operating system. Instead of switching between a browser window and a document, the user is meant to stay in their editor or spreadsheet, with Gemini adapting to the context of what's on screen.

What This Means for Users

At launch the feature works only in English, so Polish users will have to wait for full language support. Google says only that more languages are coming "soon," without giving a specific date. The Gemini app for macOS also needs to be updated to version 1.88 or later, downloaded from gemini.google/mac.

For businesses and freelancers working on Macs, this could mean faster note-taking, meeting transcription, or first drafts of text without manually retyping recordings. The screen-aware "reasoning" layer goes a step further than plain dictation, bringing Gemini closer to functions that previously required separate desktop automation tools.