Google Launches New Windows and macOS Apps, Expanding Its Desktop Ecosystem

Google’s Windows client, now out of beta, replaces the makeshift launcher that debuted last September with a fully integrated search overlay that can be summoned at any moment by pressing Alt + Space. The floating UI sits atop the active window and accepts both web queries and, with user permission, local file‑system searches. According to Ars Technica, the app can also ingest screen context: a Lens button lets users highlight any portion of the display to trigger an instant web lookup, while a “share screen” option can feed an entire window or the whole desktop into the query engine. Results appear in a browser‑like pane, complete with Google’s AI Overviews and the optional “AI Mode” that augments standard snippets with generative responses.

The Windows version is limited to Windows 10 and 11 and operates exclusively in English, but it can be run without signing into a Google account. Ars Technica notes that unsigned usage strips away personalized features, mirroring the experience of an anonymous web session. The installer is hosted on a dedicated landing page rather than the Microsoft Store, reflecting Google’s preference for direct distribution. Internally, the app leverages the same search infrastructure that powers the Chrome omnibox, extending it to index local documents and applications when the user grants the requisite permissions.

On macOS, Google’s first native desktop offering is a Gemini client that brings the full suite of its generative‑AI tools to a Swift‑based application. Ars Technica reports that the app was assembled in under 100 days by a small team and is built entirely with Google’s “Antigravity” framework, a cross‑platform UI layer that the company has been promoting for internal tooling. Users invoke the Gemini prompt bar with Option + Space, after which they can pose natural‑language questions, upload files, or open notebooks just as they would on the web version. The Mac client also supports contextual queries by accessing open windows, a capability that mirrors the Windows search app’s screen‑share feature but is oriented toward AI‑driven assistance rather than pure keyword lookup.

Feature parity with the web service is a central design goal. As Ars Technica details, the Gemini Mac app includes “Deep Research” and “Canvas” tools, as well as image, video, and music generation models that were previously only reachable through a browser. The application is packaged as a native Swift binary, which should yield smoother performance and tighter integration with macOS security mechanisms such as notarization and sandboxing. While the initial release already supports more than 100 features, Google has signaled that additional capabilities will be rolled out incrementally, though no timeline is provided in the source material.

Both desktop clients underscore Google’s broader strategy to extend its AI and search services beyond the browser sandbox. By exposing local‑context search on Windows and a full‑featured Gemini environment on macOS, the company aims to create shortcut‑driven workflows that compete with Microsoft’s Copilot integration and Apple’s native Spotlight enhancements. Ars Technica points out that the Windows app’s reliance on a keyboard shortcut and floating overlay is reminiscent of classic launcher utilities, while the Gemini Mac app’s deep integration with Google’s generative models suggests a more ambitious push to make AI a first‑class desktop experience. The rollout marks the first time Google has offered native, non‑browser entry points for its core consumer AI products, a move that could reshape how users interact with search and generative tools across the two dominant desktop platforms.