Google is accelerating its move into the desktop AI space, developing a dedicated Gemini app for the Mac to better compete with OpenAI and Anthropic.
According to a report from Bloomberg’s Mark Gurman, Google began private beta testing of the software this week with a select group of consumers.
The upcoming app, which carries the internal code name “Janus,” looks to move Gemini right into the macOS workflow. Early versions of the software were shared with Bloomberg by app researcher M1Astra, revealing that Google is testing a feature dubbed Desktop Intelligence.
Desktop Intelligence is designed to allow Gemini to tap into other Mac programs, such as the calendar, to provide more personalized responses. Much like Anthropic’s Claude Cowork, the software can essentially “see” what is on a user’s screen to pull context directly from active apps. Code within the app suggests this functionality only triggers when Gemini is open due to privacy.
While the current beta looks similar to the existing iPad and iPhone versions, it is focused on core features like image and video generation, music creation, and complex data analysis. Users can also search past conversations and upload documents for the AI to summarize or critique.
The launch is a clear attempt to catch up to OpenAI’s ChatGPT and Anthropic’s Claude, both of which already offer native Mac applications. Currently, Mac users can only use Gemini through a web browser, which is not as seamless versus a standalone app.
Apple’s upcoming AI overhaul, code-named “Campo” is expected to debut with macOS 27 later this year. Ironically, Apple’s upcoming Siri revamp is expected to rely on Google’s models to help power some of its new features.
