Google is developing a new feature for Gemini, its artificial intelligence, which will allow users to create personalized AI agents for various tasks. The discovery was made by experts who analyzed a version of the Google application, pointing to a feature that promises to optimize the experience with the platform. This news comes shortly after the official arrival of the Gemini Spark in Brazil on July 31, 2026.
Custom AI agents promise new possibilities for users
The functionality under development will allow Gemini to transcend its role as general-purpose artificial intelligence, acting in specific scenarios. Unlike the “personas” (Gems) already existing in Gemini, which adjust the response style of artificial intelligence, these new agents would be designed to perform autonomous and focused tasks. This customization aims to meet users’ more specific and practical demands.
Discovery reveals internal tests in Google app
Experts have identified the “New Agent” option in the Gemini sidebar within version 17.45.14 of the Google app, located next to the “New Conversation” shortcut. The detection indicates that the company is advancing in testing for the public availability of these tools. Clicking the button starts a guided process for configuring a new agent.
Learn more: Artificial intelligence expands Alexa+ control in smart home systems
Discover the categories of virtual assistants already being tested
The initial virtual agent options being tested cover several areas of everyday and professional life. Although creating an agent completely from scratch is not yet available to the general public, the pre-defined suggestions promise a high level of usefulness:
On the same topic: ChatGPT releases app for computers, promising to revolutionize the use of the operating system
- Family coordinator:assists in planning household tasks, sharing events and organizing family activities.
- Tutor:offers support for the study of complex topics, help with academic work and in preparing quizzes.
- Group Organizer:facilitates the coordination and planning of social or collaborative events.
- Marketing Agent:supports the creation of publications for social networks, newsletters and marketing campaigns or creative projects.
- Finance agent:helps you monitor and manage expenses, as well as remind you about bills to pay.
Current restrictions and the path to official launch
Currently, the full customization freedom to create an entirely new agent appears restricted to Google Workspace or Enterprise account users. For the general public, choices are limited to the pre-defined options during the testing phase. Given the experimental nature of the functionality, there is no confirmed release date for these custom agents to reach all Gemini users. More features can be incorporated throughout development.
The arrival of Gemini Spark on the Brazilian market
The Gemini Spark platform, the basis for these advances in artificial intelligence agents, was officially launched in Brazil on July 31, 2026. The national version offers support for web browsing and is accessible to subscribers of Google AI Pro, with a monthly fee of R$96.99, or Google AI Ultra, which costs R$779.90 per month.