Artificial Intelligence

Gemini's Windows App Arrives, but Hands-On Impressions Are Poor

PCWorld's hands-on look at Google's new Gemini app for Windows finds a chatbot front end that struggles to perform even basic tasks within Google's own services.

Featured image for Gemini's Windows App Arrives, but Hands-On Impressions Are Poor

Google has brought its Gemini app to Windows, following a macOS release earlier this year. But a hands-on review from PCWorld concludes the new desktop client offers little beyond what the Gemini web app already provides.

According to PCWorld's Mark Hachman, the Windows version is essentially a front door to Gemini's generative features. Signing in with a Google account unlocks stored chats and tabs for images and video, along with a media library. Users can optionally connect a Workspace account to receive a daily brief of important emails.

Those integrations sound useful, but Hachman reports they did not translate into action. When he asked Gemini to delete old emails from his Social tab, the chatbot said it lacked the API access needed to delete messages or modify category tabs, directing him to do it manually. A request to identify the largest file in Google Drive also failed; Gemini said it could not see file metadata or sort files by size.

That limitation stands out against rival desktop apps. PCWorld notes that ChatGPT's Windows client can connect to a user's desktop or services like Slack and take actions on the user's behalf. By that comparison, Gemini's inability to operate on Google's own Workspace apps is a notable shortfall.

Hachman also encountered instability. A request to generate a video of a bunny and a squirrel playing tennis apparently vanished after he navigated away from the tab, and his saved media library later showed as unavailable.

The review praises some settings. A Personal Intelligence tab lets users toggle Gemini's memory about them, and a separate import option allows them to have other chatbots summarize what they know and feed that into Gemini. Hachman notes that Anthropic's Claude offers a similar capability, as reported by his colleague Ben Patterson.

Google's announcement mentioned access to Spark, the company's personalized AI agent, according to PCWorld. Hachman says he did not see it in the app, though he notes he does not pay for the $100-per-month Pro plan that may gate the feature. The app instead pointed him toward Scheduled Actions under Settings, which offers suggestions like a weekly recipe or a daily motivational quote.

Hachman's verdict is blunt: no app is ever truly finished, but for now he considers the Gemini app for Windows a waste of time. His colleague had given the Mac version a similarly middling reception.

What remains unclear is whether the gaps reflect temporary launch limitations, account-tier restrictions, or deeper design choices. PCWorld's assessment is based on one reviewer's experience, and Google has not responded publicly in the piece. Readers weighing the app should treat the feature comparisons to ChatGPT and Claude as context rather than a definitive ranking. The more practical takeaway: a desktop client tied to Google's ecosystem still cannot manipulate that ecosystem's core apps, which undercuts the main reason to install it. Until that changes, the web interface may serve most users just as well.

Sources and further reading