Gemini Spark comes to Google's Gemini app for macOS

Google has begun rolling out Gemini Spark, an agentic AI assistant, to its dedicated Gemini app for macOS as of June 30, 2026. The update shifts the assistant from a reactive chatbot to an active partner capable of automating local desktop tasks and managing multi-step workflows across files and applications.

Available in beta for users in the United States aged 18 and older, the service is exclusive to Google AI Ultra subscribers, who pay between $99 and $100 per month. According to SiliconANGLE, the assistant is built on the Gemini 3.5 Flash model and uses "Google Antigravity," a next-generation autonomous AI programming environment designed to orchestrate AI agents.

Media additions

We're bringing Gemini Spark to the @GeminiApp for macOS app so it can ...

Local Automation and Desktop Control

Unlike previous versions that relied on remote browsers and cloud computing, Gemini Spark for macOS can now interact directly with a user's local environment. Google DeepMind Vice President of Engineering Srinivasan Venkatachary stated the macOS version is designed to automate desktop file management and local tasks, connecting the agent to files and applications in ways not previously possible.

Users can assign the assistant to handle repetitive digital work, such as sorting a large volume of PDFs in a Downloads folder into specific categories. The assistant also bridges the gap between local storage and Google Workspace. According to Engadget, users can ask Spark to create a budget spreadsheet using invoices saved locally on their laptop and set a recurring schedule to update the document.

To manage these capabilities, version 1.80.15.516 introduces a Spark tab in the sidebar. From here, users can select Connected folders to grant the assistant permission to access specific local data.

"Scan my unorganized files, group them by type or context, and archive the clutter into a clean folder structure. Please show me your organization plan first and ask for my input on where to move them."

Example prompt, via 9to5google

Google stated that the assistant will only access files for which the user has granted explicit permission. Additionally, a new Usage limits page has been added to the account menu in the bottom-left corner of the app to help users track activity. This item previously opened a browser link to gemini.google.com/usage.

Expanded Integrations and Real-Time Tracking

Alongside the macOS launch, Google is expanding the ecosystem of apps Spark can control. The assistant now integrates with Google Tasks and Google Keep, allowing it to transform notes into action items. Further integrations with the following third-party services are rolling out over the next week for web and mobile, with the macOS app to follow in the coming weeks:

• Canva (custom flyer design)
• Dropbox (file access and sharing)
• Instacart (grocery ordering)
• OpenTable (restaurant reservations)
• Zillow Rentals (apartment tour reservations)

Google is also introducing support for custom Model Context Protocol (MCP), which enables users to connect their own preferred apps directly into Spark. This allows the agent to operate as a digital secretary, handling tasks such as cleaning up inboxes or monitoring credit card statements for new subscription fees.

Spark has also gained the ability to track topics in real time across blogs, news sites, social media, finance, weather, and sports. For example, users can set triggers for the assistant to send a financial report if a stock reaches a specific threshold or provide highlights the moment a soccer match ends. This real-time tracking is available across web, mobile, and desktop platforms.

The Roadmap: Remote Execution and Voice

Google has signaled that the current release is the start of a broader desktop rollout. A forthcoming update will introduce remote task execution, allowing users to control their Mac via a mobile device or web browser. Users could, for instance, ask Spark from their phone to locate a sales report on their Mac, extract a total revenue number, and email that data to them while they are away from their desk.

Another upcoming addition is an upgraded voice experience. According to Android Authority, this will allow users to speak casually—including ums and ahs—while the AI cleans up the phrasing to create precise drafts or requests based on the context visible on the screen.

What to watch next: The rollout of the new voice experience and remote execution features is expected later this summer.