Google Announces Gemini Spark AI Agent for Ultra Subscribers

Google has introduced Gemini Spark, a proactive AI agent designed to manage digital tasks autonomously across various platforms [1]. Announced at the Google I/O 2026 developer conference, the tool is built to function 24/7 in the cloud, allowing it to complete multi-step workflows even when a user's devices are powered off [2].

Key takeaways

• Gemini Spark is a cloud-based agent powered by the Gemini 3.5 Flash model and built on Google’s Antigravity infrastructure [2, 3].
• The agent integrates with Google Workspace apps like Gmail, Docs, and Slides, with additional support for Canva, OpenTable, and Instacart [1, 3].
• Spark is scheduled to reach trusted testers this week, followed by a beta launch for US Google AI Ultra subscribers next week [1, 2].
• Google plans to introduce an Agent Payments Protocol (AP2) in the coming months to allow the agent to securely make purchases on a user's behalf [1].

Proactive automation in the cloud

Unlike traditional chatbots that require constant user input, Gemini Spark is designed to operate as a background assistant [2]. By running on dedicated virtual machines, the agent can handle complex, long-running tasks such as parsing credit card statements to identify subscriptions, monitoring school emails for deadlines, or compiling meeting notes into polished documents [1, 3]. Google demonstrated the agent’s ability to draft emails in a user's voice and automatically update documents based on incoming inbox activity [2].

To maintain user control, Google says the agent is designed to request permission before executing high-stakes actions, such as spending money [1]. While the initial rollout focuses on Workspace and select third-party integrations, the company plans to expand Spark’s capabilities later this year to include Chrome browser integration, custom sub-agents, and the ability for users to communicate with the agent via text or email [1, 2].

Why it matters

The introduction of Spark marks a shift in Google’s strategy, evolving Gemini from a reactive chatbot into a proactive agent capable of navigating a user's digital life [1]. This development places Google in direct competition with other personalized AI agent offerings currently emerging in the industry [2]. As the project moves into beta, Google is also adjusting its subscription structure, with plans for a $100 lower-end Ultra tier and a $200 per month high-end Ultra plan, both of which will include access to Spark [2]. Future updates are expected to bring deeper integration with macOS and local machine workflows, further expanding the agent's reach across the user's computing environment [1].