Google launches a native Gemini app for Windows 10 and 11

Google announced this week the global rollout of the Gemini app for computers running Windows 10 and Windows 11, available as a direct download from the company's website. The news, confirmed on September 10 by the official @GeminiApp account and picked up by Portuguese and Spanish tech outlets, places Google's AI assistant directly on the desktop, alongside Microsoft's own native operating-system apps. Until now, reaching Gemini on a PC always meant opening a browser and navigating to the service; from now on, the Alt+Space keyboard shortcut instantly opens a window ready to take a request, whether the user is working in a spreadsheet, a document, or in the middle of a game.

Little changes visually, but access is instant

Visually, the Windows app stays very close to the web version, with a lightweight window and a side panel giving access to Notebooks and recent conversations — a choice that favours familiarity for people who already use Gemini every day. The real novelty is the friction that disappears: there is no longer any need to keep browser tabs open just to consult the AI, or to interrupt an ongoing task to reach it. Google also says it designed the app to be unobtrusive, with optimised memory and processor use, so its background presence does not hurt the performance of other open applications.

Gemini Spark turns the assistant into an agent

The most significant part of the update is called Gemini Spark, a personal agent designed to carry out multi-step tasks autonomously — one that can keep working in the cloud even if the user's computer is switched off, always under permissions the user sets in advance. Rather than simply answering questions, Spark connects to apps such as Gmail, Calendar, Drive, Docs, Sheets, Slides, YouTube and Maps to review email, organise information or prepare documents. None of these connections switches on automatically: the user decides, app by app, which services the agent may access. Google structures these features around three concepts — Tasks, Skills and Schedules — which let people define concrete actions, teach specific procedures, and schedule their recurring execution.

• Activation shortcut: Alt + Space, in any application
• Gemini Spark: an agent that keeps working in the cloud with the PC off
• Available connections: Gmail, Calendar, Drive, Docs, Sheets, Slides, YouTube, Maps
• Creative generation: images with Nano Banana, video with Gemini Omni
• Availability: direct download from Google's website, Windows 10 and 11

Gemini's arrival on the Windows desktop is not an isolated addition: it is also a direct move into Microsoft's own territory, which already bundles its own assistant, Copilot, straight into the operating system and the Office suite. Unlike Copilot, which comes pre-installed, Gemini requires a manual download and installation — but it gains, in exchange, the promise of more native capabilities in the coming months, including deeper integration with on-screen context and local applications, which Google has not yet detailed. For the ordinary user, the immediate result is the option to run two AI assistants side by side on the same machine, each tied to a different ecosystem of apps.

For companies that standardise their workstations on Windows, this update forces a decision that did not carry the same urgency until now: deciding which AI apps can be installed on organisation-owned computers, and what data can be shared with agents like Spark. Integration with Gmail, Drive or Sheets makes the tool attractive for small and mid-sized businesses already inside the Google Workspace ecosystem, but it also means revisiting internal data-protection policies before allowing automatic connections between email, calendar and internal documents. In a corporate PC fleet where Windows remains dominant, the arrival of a second native assistant — alongside Copilot — turns AI licensing and permission management into a topic IT departments can no longer put off.