Taking the wraps off its new additions, Google has introduced Gemini Spark at the Google I/O 2026 keynote. Spark is a personal AI agent designed to help users navigate their digital lives by taking actions on their behalf under their direction. It operates on dedicated Google virtual machines hosted on Google Cloud. The platform is powered by Gemini 3.5 and Google’s “Antigravity” framework, allowing it to run tasks seamlessly in the background. Spark will also integrate with third-party tools through MCP support.
Users will soon be able to access Gemini Spark through the Gemini app, Google services, and email windows.
Gemini Spark is designed to access multiple databases and compile information into streamlined, well-articulated summaries. It can pull information from drafts, chats, emails, and other sources to generate organized outputs.
For more complex prompts, Spark can break tasks down into multiple steps and execute them sequentially. It keeps users informed about its progress while handling everyday tasks such as drafting emails, scheduling, and updating documents. Deep integration across Google’s ecosystem also allows Spark to update Sheets, track records, and manage workflows more efficiently.
The AI agent can also pull relevant information from external websites if it affects a user’s planning or workflow.
Functional Across Android and Apple Devices
With new voice-assisted AI capabilities, Gemini Spark can understand details and nuances from voice notes and prompts before acting on them. Even when tasks involve multiple layers, Spark can divide them into manageable chunks and work on them simultaneously.
As Google prioritizes safety, Gemini Spark will first roll out to trusted testers this week, followed by a beta release for Google AI Ultra subscribers next week. To expand accessibility, Google AI Ultra has also introduced a new starting plan priced at $100, while the top-tier Ultra subscription has been reduced from $250 to $200 per month.
Google also revealed that Gemini Spark will eventually work directly within Google Chrome, effectively acting as an agentic browser assistant capable of navigating and interacting with the web on behalf of users.
In addition, Google announced Android Halo, a new home base for AI agents on Android devices, which is expected to launch later this year.
