Google has started to make serious moves in the field of artificial intelligence on Mac computersThe company is testing it discreetly. a new native Gemini app for macOS with a small group of users, in what is its first major attempt to bring its assistant directly to the desktop, without relying on the browser.
This move comes at a time when OpenAI and Anthropic They have already taken the lead with ChatGPT and Claude's own applications for MacThis is something many users in Spain and the rest of Europe have been requesting from Google. Until now, using Gemini on a Mac required opening a tab in the browser, a functional solution, but less convenient for everyday use.
A private beta with essential features and the codename "Janus"
According to sources cited by Bloomberg and other specialized media outlets, including journalist Mark Gurman, Google has begun distributing a Preview version of the Gemini app for Mac to a very limited circle of external testers. Internally, the project is known by the codename "Janus"The goal of this early phase is to gather feedback, locate bugs, and fine-tune the experience before a wider release.
The company itself has warned participants that this is an incomplete softwareThe app currently only integrates features considered critical compared to other Gemini clients, leaving out some advanced capabilities that will arrive later. Even so, the fact that it has left internal labs and reached external users indicates that development is already at a fairly mature stage.
For those who use Macs daily in professional or academic settings in Europe, this move represents the first real step towards a More robust integration of Gemini into macOSAnd it suggests that Google doesn't want to fall behind its direct rivals on the desktop.
Until now, the Gemini experience has been limited to the web: practical for occasional queries, but less agile when looking for a tool to support the workflow in real time. With this private beta, Google is testing precisely that leap to the desktop.
Multimodal capabilities and a focus on productivity
Although the current version is limited, Google is already testing a set of features on Mac. multimodal and information analysis functions that go beyond classic text chat. Available options include generating images, video, and music, as well as creating tables and charts for working with data in a more visual way.
In addition, the app allows you to perform mathematical analysis and complex information processingThis is especially relevant for technical profiles, university students, or professionals who handle large volumes of data. All of this is complemented by the ability to search the web directly from the application, without needing to open a browser.
Another key is context management: Gemini for Mac can Review previous conversations and keep the history in mind from the user to offer answers more tailored to their needs. This helps ensure that the sessions don't feel disconnected, but rather like an ongoing relationship with the assistant.
The beta version also includes the option to analyze documents and multimedia files uploaded from the computer itselfIn this way, a user can upload reports, presentations, PDFs, or even audiovisual content stored on their Mac for Gemini to summarize, explain, compare, or restructure them in various ways.
All of this represents a significant leap forward compared to using Gemini in a browser, where uploading and managing files can be less convenient. On desktop, the app aims to become a kind of core productivity tool that accompanies the user while working with several programs at the same time.
Desktop Intelligence: Gemini takes a look at the rest of the Mac apps
One of the most striking features found in the code of the new application is something that Google has dubbed as Desktop IntelligenceUnder this name lies a set of capabilities designed to enable Gemini to interact with what happens on the user's desktopbeyond the app's own window.
In practice, this means that, with the user's permission, Gemini could view what's on the screen and extract information from other applicationssuch as the calendar, email, or documents currently open. With this additional context, the assistant would be able to offer more accurate answers and tailor its recommendations to what is actually being done.
For example, it could detect that a long report is being worked on in a word processor and to propose summaries, style revisions, or help in rewriting paragraphsOr, when faced with a busy schedule, suggest reorganizing appointments based on priorities, which is especially useful for those managing complex schedules.
According to the app's internal text, this feature would only activate when the user is actually using Gemini, which points to a an attempt to balance utility and privacyThe idea is that the assistant "sees what you see" only when you ask for help, and not permanently in the background.
This approach is reminiscent of similar proposals from competitors like Anthropic, whose desktop tools also aim to integrate AI into workflows without forcing users to constantly switch applications. The difference, in Google's case, is that Desktop Intelligence relies on... The company has extensive experience in services such as Gmail, Calendar, and Drive., which can lead to deeper integrations if extended to the desktop.
A change in philosophy for the browser company
Beyond the specific functions, the launch of a native Gemini app for Mac represents a striking shift in Google's historical strategyFor years, the company has championed the idea that almost everything can and should be done from the browser: email, office applications, video calls, and, until now, also its AI services.
On desktop, that focus on the web had remained fairly strong, while on mobile, Google had long offered native apps with advanced features. With Gemini for Mac, that line is starting to blur, driven largely by the user experience demanded by generative AI tools.
Professional and creative users in Spain and other European countries who have incorporated these models into their daily routine tend to prefer keyboard shortcuts, floating windows, and direct access to the context of what they are doingsomething that a simple browser tab cannot always offer with the same fluidity.
Another important detail is that the Gemini app arrives first in macOS, not WindowsDespite the fact that the latter remains the dominant system on desktop and laptop computers, OpenAI and Anthropic have followed a similar path, suggesting that the typical user willing to pay for advanced AI assistants is concentrated, at least for the moment, largely within the Apple ecosystem.
Regarding the design, the first leaks suggest that the app's interface for Mac will be... It maintains an appearance very similar to the iPhone and iPad versions.This visual continuity makes it easier to switch between devices and fits with Google's strategy of offering a consistent experience regardless of where Gemini is used.
Direct competition with ChatGPT, Claude and Apple's plans
Gemini's arrival at Mac is also understood within a race to dominate AI in personal devicesToday, ChatGPT and Claude already have native macOS applications, with keyboard shortcuts, system integration, and, in the case of tools like Claude Code, even a presence in the command line for developers.
Those who used Gemini on Mac until now were at a disadvantage, as they relied solely on the web version. The new app aims to to match, and in some cases surpass, that level of integrationespecially with features like Desktop Intelligence and advanced local file management.
This move fits into a broader internal reorganization at Google, which has unified software and hardware teams under one umbrella To accelerate the development of AI-based products, the company is trying to react more quickly to a market where progress is measured almost quarterly.
In parallel, Apple is working on its own evolution of Siri. and in new AI capabilities that will arrive in future versions of iOS and macOS. Previous reports suggest that some of these features could rely on Google models, which paints a curious picture: head-on competition in the user experience layer, but potential collaboration on the underlying technology.
For European users, the rollout of all these tools will also be subject to the data protection regulations and the specific regulation of AI which is being finalized in the European Union, a factor that large technology companies will have to take into account when activating sensitive functions such as access to screen content.
With no official release date yet announced, it seems likely that Google will continue refining the Gemini app for Mac in the coming months, expanding the number of testers and adding features until a public version is released. When that happens, macOS users in Spain and the rest of Europe will have access to it. a more integrated desktop alternative to work with generative AI, in a scenario where competition, far from slowing down, seems to be just warming up.
