Google announced Gemini Spark on May 19, 2026 [1], a personal AI agent designed to complete digital tasks autonomously.
The release marks a shift from reactive chatbots to proactive agents that can execute complex workflows without constant user prompting. This move places Google in direct competition with other emerging AI agents, such as OpenClaw [3, 4].
Unveiled at the Google I/O developer conference in Mountain View, California, Gemini Spark operates 24/7 [2, 5]. The agent is powered by the Gemini 3.5 model [6]. It is designed to work across Google Workspace apps to monitor emails, and create study guides [2, 4].
One of the agent's primary functions is the ability to identify hidden subscription fees within a user's accounts [2, 4]. By proactively scanning data and executing tasks, the system aims to reduce the manual effort required for digital administration. Google said the agent is intended to improve overall user productivity [3, 4].
Access to the tool will be limited initially. Google said a beta version will be available soon for Google AI Ultra subscribers [7]. This rollout strategy allows the company to test the agent's autonomous capabilities with a high-tier user base before a wider release.
Gemini Spark represents a broader industry trend toward "agentic AI," where software does not just provide information but performs real-world work [8]. The integration into the existing Google ecosystem gives the agent a significant advantage in data access—allowing it to sync calendars, documents, and communications seamlessly.
“Gemini Spark operates 24/7”
The transition to a 24/7 autonomous agent signifies a move toward a 'zero-UI' experience, where the AI manages the interface for the user. By integrating Gemini 3.5 deeply into Workspace, Google is attempting to lock users further into its ecosystem through utility that extends beyond simple search or content generation into active digital labor.
