Google Launches Gemini Spark 24/7 AI Agent in India

NEW DELHI — Google has begun rolling out Gemini Spark in India, bringing its persistent, always-on artificial intelligence agent to domestic subscribers. Positioned as a major shift in Google’s consumer artificial intelligence strategy, Spark moves beyond the familiar chatbot model, operating continuously in the background to manage complex workflows and automate routine tasks across various applications.

The service is currently available to users subscribed to the higher-tier Google AI Pro and Google AI Ultra plans. Instead of waiting for individual prompts in an active browser window, Spark runs on Google’s cloud infrastructure, meaning it can execute multi-step instructions even when the user’s device is turned off or offline.

Shifting from Assistant to Agent

The transition from a reactive assistant to a proactive agent requires different underlying architecture. Google confirmed that Gemini Spark is powered by the Gemini 3.6 Flash model combined with the “Google Antigravity harness,” a new system specifically designed to manage and sustain long-horizon tasks.

Adam Coimbra, Director of Product Management for Gemini, stated that the system is designed to anticipate user needs rather than just following explicit, real-time commands.

“What’s really exciting about Gemini Spark is it goes beyond these sort of daily or weekly schedules to something that starts to be truly proactive,” Coimbra said during a media briefing. He noted that the agent can monitor ongoing projects, pull updates from internal chats, and automatically maintain trackers in Google Sheets without requiring manual intervention.

Core Mechanics: Tasks, Schedules, and Skills

Users interact with Gemini Spark through a dedicated interface in the Gemini app, relying on three primary components to build automated workflows:

• Tasks: These define the outcome a user wants to achieve. A task could be a simple request, such as organizing an inbox, or a complex, multi-step process like monitoring a competitor’s product releases and compiling a weekly research digest.

• Schedules: This feature dictates when the agent takes action. Schedules can be tied to a specific date and time, a recurring interval, or a trigger condition—such as activating whenever a specific type of email arrives in the user’s Gmail inbox.

• Skills: Instead of repeatedly typing out preferences, users can create “Skills,” which function as reusable instruction sets. These give Spark specific context on how to approach a recurring activity and which tools to prioritize.

Cross-Application Integration and Remote Browsing

Gemini Spark connects natively with Google Workspace applications, including Gmail, Docs, Drive, Calendar, and Tasks. For external software, the agent utilizes the Model Context Protocol (MCP) to communicate with third-party project management tools, customer relationship management (CRM) platforms, and communication applications like Slack.

For tasks requiring web navigation outside of API connections, Spark employs an auto-browse function via a remote browser instance. This allows the agent to interact with websites on the user’s behalf.

However, Google has implemented boundaries for security and privacy. The company stated that the agent will pause and explicitly ask for user confirmation before carrying out high-stakes actions, such as spending money or sending outbound emails. If a task requires sensitive input—like entering payment details or passwords—Spark halts the process and alerts the user to take over the session manually.

Pricing and Market Strategy in India

The rollout in India targets power users, founders, and enterprise professionals who are willing to pay a premium for productivity automation. Gemini Spark is currently excluded from the more affordable Google AI Plus tier.

In India, the Google AI Pro plan, which includes 5TB of storage, is priced at ₹1,950 per month, while the 10TB tier costs ₹3,250 per month. Users opting for the Google AI Ultra plans will pay ₹6,500 monthly for the 20TB option, with the highest 30TB tier—which includes higher usage tokens—priced at ₹19,500 per month.

The introduction of Spark aligns with broader industry projections regarding the commercial viability of agentic AI. Research firm Grand View Research estimates that the global AI agents market will grow from $7.6 billion in 2025 to nearly $183 billion by 2033, driven heavily by enterprise demands for workflow automation.

At launch, users in India must be at least 18 years old and logged into a personal Google Account to access Gemini Spark, as the feature is not yet available for work or school accounts.