Google is widening access to Gemini Spark, and the feature stands out because it is designed to keep working even when the user is not actively using a device. The AI agent is now rolling out to Google AI Ultra subscribers in the United States, extending access beyond the earlier trusted-tester phase.
Spark is available on Android, iOS, and the web, but only for Ultra customers at this stage. On the web, it appears in the side panel next to the Chat tab, while on mobile it sits between Search chats and Daily brief.
Google positions Spark as a background agent that can take on tasks and continue running for 24 hours a day. According to Google’s Gemini page, the tool is meant to help users manage digital life by handling assigned work even if phones and laptops are off.
That level of automation does not mean Spark acts without limits. Google says the feature must be enabled first, and it is built to ask for approval before taking major actions.
Built around Tasks, Skills, and Schedules
Spark is structured around three core parts: Tasks, Skills, and Schedules. Each one handles a different layer of automation, from completing work to shaping behavior and timing actions.
Tasks connect Spark to Google Workspace services, including Gmail, Calendar, Docs, Sheets, and Slides. Google says this allows the agent to handle everyday productivity work and keep monitoring information over time.
One example Google gives is a task that searches for and tracks interior design internship openings in New Orleans for the summer. That example shows Spark is not limited to one-time answers and can continue watching for updates.
Skills are meant to let users define how Spark should behave during repeated work. Google says a user can, for example, have Spark read the last 50 emails they wrote and turn that into an email style guide.
That guide can then become a custom skill, such as a “ghostwriter” mode that Spark uses automatically when Gemini is asked to draft email. The goal is to reduce repetitive prompting and make the experience more personal.
Schedules add another layer by letting Spark run on a specific timetable or in response to a condition. Google’s example is a Monday 9 a.m. inbox scan that produces important updates, a weekly priority list, and focused calendar blocks.
A more active role for Gemini
Spark also signals a broader shift in how Google wants Gemini to work on everyday devices. Instead of acting only as a chatbot that waits for a question and responds, Spark is designed to keep monitoring, following up, and carrying out work in the background.
That makes the feature more than a simple conversational tool. Its value is tied to real workflows across Gmail, Calendar, and other productivity apps, where ongoing action can matter as much as answers.
The initial rollout is still limited to a premium audience, which keeps Spark positioned as part of Google’s higher-end AI offering. Google AI Ultra starts at $99.99 per month, a price point Google places among its best options for heavy AI users.
Google has also said Spark will keep expanding this summer. One feature previously mentioned is the ability for Spark to spend the user’s money, and Google has also indicated that Spark will come to the Gemini desktop app during the summer.
That could broaden the agent’s use cases for Ultra subscribers who work across phones and computers. For now, the rollout marks an early step in Google’s effort to turn Gemini into an always-on assistant rather than a tool that only responds when asked.
Source: www.androidpolice.com
