- Gemini Spark can perform tasks automatically in the background.
- The feature is currently available to Google AI Ultra subscribers in the US.
- Google says users can run up to 15 active tasks at the same time.
Google has begun rolling out Gemini Spark, a new AI-powered agent that can carry out tasks on a user's behalf without needing constant instructions, marking a significant shift in how the company sees the future of artificial intelligence.
The feature, currently available to Google AI Ultra subscribers in the US, was unveiled at Google's I/O developer conference alongside Gemini 3.5 and several other updates to the Gemini platform. Unlike traditional AI chatbots that wait for user prompts, Gemini Spark is designed to work continuously in the background, handling tasks, monitoring events and responding to changing conditions automatically.
Beyond chat: Google's AI takes action
At its core, Gemini Spark is intended to function more like a digital assistant than a chatbot. Users can assign tasks, set schedules and create reusable instructions that allow the system to perform actions across different apps and services.
The platform is built around three key elements: Tasks, Schedules and Skills. Tasks define what the user wants accomplished, whether that's organising emails, researching a topic or planning a trip. Schedules determine when those actions should take place, while Skills act as reusable instruction sets that teach the system how to complete specific jobs.
According to reports, Spark can access a range of Google services, including Gmail, Google Calendar, Drive, Docs, Sheets and Slides. It can search inboxes, summarise email conversations, draft messages, organise files, update documents and manage calendar appointments.
The system can also interact with websites through a remote browser environment. This allows it to navigate webpages and perform actions on behalf of users, such as filling out forms or adding products to online shopping baskets. Google has also equipped the platform with a remote computer environment capable of handling more advanced workflows and code execution.
A glimpse of the AI workplace
One example highlighted in reports involves travel planning. A user could ask Spark to organise a business trip, monitor flight updates and automatically react if plans change.
If a flight is delayed, the system could potentially update an itinerary, make alternative arrangements using pre-configured workflows and prepare email notifications without requiring further input from the user.
The tool may also appeal to professionals who spend large amounts of time managing information. Spark can reportedly summarise newsletters, create personalised news briefings, monitor developing stories and compile research from multiple sources into structured reports with citations.
Users can monitor ongoing activities through a dedicated dashboard that displays scheduled tasks, work in progress and completed actions. Manual control remains available, allowing users to take over whenever necessary.
For now, Google has imposed usage limits. Reports indicate users can run up to 15 active tasks simultaneously, with additional requests waiting until capacity becomes available.
While Gemini Spark remains in beta, its arrival signals Google's growing focus on AI agents that can complete work independently rather than simply answer questions. Whether users are comfortable handing over more responsibility to AI systems may ultimately determine how quickly tools like Spark become part of everyday digital life.
