product update

Google launches Gemini Spark AI agent for Ultra subscribers in US with automated task execution

TL;DR

Google has launched Gemini Spark, a 24/7 AI agent for Google AI Ultra subscribers in the US. The service automates tasks across Google Workspace apps with remote browser control, supporting up to 15 concurrent tasks with compute-based usage limits.

2 min read
1

Google launches Gemini Spark AI agent for Ultra subscribers in US with automated task execution

Google has released Gemini Spark to Google AI Ultra subscribers in the United States, according to 9to5Google. The AI agent automates tasks across Google Workspace and connected apps with remote browser capabilities and code execution.

Core capabilities and integrations

Gemini Spark accesses Google Workspace (Gmail, Calendar, Drive, Docs, Sheets, Slides), connected apps, personal intelligence data, location, websites users are signed into, and operates a remote browser with automatically saved information. The service can navigate websites and interact with pages, including adding items to shopping carts.

In Gmail, Spark searches email threads, summarizes conversations, drafts replies, forwards emails, and organizes inboxes with labels. Calendar functions include checking availability, RSVPing to invitations, scheduling events, rescheduling meetings, and suggesting time slots. Drive capabilities cover file search, reading file contents, checking metadata, creating and editing documents, building spreadsheets with formulas, and generating slide presentations.

Architecture: Tasks, schedules, and skills

Gemini Spark operates through three components. Tasks define high-level goals or projects. Schedules determine when Spark runs automatically, either at specific times or in response to events (example: "Every day at 8 AM, give me an update on AI news"). Skills are reusable instruction sets accessed via @ or /, teaching Gemini how to complete specific tasks and which tools to use.

Google provides this example: Task - "Plan and manage my business trip to London." Schedule - "When my flight is delayed, notify me and propose an update to my itinerary." Skill - Using "Travel Booking" and "Gmail Writing" skills simultaneously to rebook a room and send confirmation.

Usage limits and access

Users can run up to 15 tasks concurrently. New task requests require waiting for existing tasks to complete. Schedules will not trigger if 15 tasks are already running. Gemini Spark shares the same compute-based usage limits as the broader Gemini platform.

The interface appears as a "Spark" tab in the web side panel opposite "Chat." On Android and iOS, Spark sits between Search chats and Daily brief, marked as Beta. The task thread and work panel display planned, current, and completed steps, with users able to take control of the remote browser when needed.

What this means

Gemini Spark represents Google's entry into autonomous AI agents, competing directly with capabilities announced by Anthropic (Claude with computer use) and OpenAI (Operator). The 15-task concurrent limit and compute-based restrictions indicate Google is cautiously scaling the service. The remote browser capability with page interaction pushes beyond simple API integrations, though real-world reliability and error handling remain to be tested. For Ultra subscribers, this adds tangible workflow automation value beyond conversational AI—assuming the agent executes reliably across complex, multi-step tasks.

Related Articles

product update

Google releases Magic Pointer app for unreleased Googlebook device to Play Store

Google has released Magic Pointer to the Play Store, an app designed for its yet-to-be-announced Googlebook device. The app allows users to select on-screen content to receive contextual AI suggestions powered by Gemini, including search, image creation, and shopping features.

product update

Amazon Nova Act Brings Vision-Based Web Navigation to UX Testing, No Hard-Coded Scripts Required

AWS has released a cloud-deployed UX testing platform built on Amazon Nova Act, a multimodal foundation model that navigates web interfaces through visual understanding rather than hard-coded selectors. The solution processes documentation with Claude 4.5 Sonnet to generate test scenarios, executes parallel testing via ECS, and analyzes results automatically, addressing the scalability limitations of manual testing and maintenance overhead of traditional automation tools.

product update

Google AI Mode adds chat history search on Android, file uploads on iOS

Google AI Mode now supports searching previous chat conversations on Android and iOS devices. The update also brings file upload capability to the iPhone app, matching features already available on the web version.

product update

Apple releases iOS 27 public beta with AI-powered Siri overhaul built on Apple-Google Foundation Models

Apple released the iOS 27 public beta, making its AI-powered Siri overhaul available to all users for the first time beyond developers. The assistant leverages Apple Foundation Models built in collaboration with Google Gemini, running on-device with Private Cloud Compute across Apple's 2.5 billion active devices.

Comments

Loading...