Google is stepping up its AI game on desktop, with a dedicated Gemini app for Mac now in testing, and it could bring a major new capability called Desktop Intelligence. According to a report by Bloomberg, Google has started privately testing an early version of the Gemini Mac app with select users. The move is part of a broader push to compete more directly with desktop AI apps like ChatGPT and Claude, both of which already offer native Mac experiences.
The current version is still early and doesn’t include all planned features, but testing suggests Google is aiming for a near-term release. For now, Mac users primarily access Gemini through the web, so a dedicated app could significantly streamline the experience.
What would set it apart?
Over the past few months, on-device and cross-app AI workflows have really exploded in popularity. Tools like Manus and Openclaw are geared at power users, but for an average user, AI that dips into their local data (such as email, calendar, photos, etc.) is what they seek, without any overt technical hassles.
That’s where Gemini comes into the picture. Google recently introduced Personal Intelligence for Gemini, which lets it pull information from sources such as Gmail, Photos, YouTube, and Search. Google is also integrating it across Workspace tools, such as Drive, too. Anthropic’s Claude Cowork is doing something similar within Microsoft’s Office 365 apps.
Broadly, the vision is to let AI access the on-device (and on-cloud) personal data and seamlessly get work done without opening all these apps separately. Google apparently has a similar vision for its Gemini app on Mac. Think of it as an AI that can see all the information and has the right context to answer your queries. If you ask it about vacation ideas, it can look up your Gmail for upcoming tickets, bookings, and itineraries to answer accordingly.
“When you enable apps for Desktop Intelligence, you are enabling Gemini to see what you see (such as screen context) and pull content directly from these apps to improve and personalize your experience only when Gemini is in use,” the in-development app’s code notes, as per Bloomberg.
What is Desktop Intelligence?
The standout feature here is something Google calls Desktop Intelligence, and it’s arguably the biggest differentiator for the Mac app. Instead of relying only on prompts, Gemini can see what’s on your screen and pull data directly from Mac apps, like your calendar, documents, or browser, to deliver more context-aware responses.
In practice, this means Gemini doesn’t just answer questions; it understands what you’re doing. Whether it’s summarizing a document you’re working on or helping plan something based on your existing data, the AI can respond with far more relevant, real-time context.
Even in its early form, the app already looks fairly capable. Google is testing features like image and video generation, coding support, document analysis, and web search, along with the ability to process uploaded files and recall past conversations, thus putting it on par with other advanced AI assistants. Google hasn’t officially confirmed a release date yet, but with external testing already underway, a public rollout may not be too far off.
