Google continues to explore ways to make Gemini more than just on-screen responses. The company is testing a feature that allows its assistant to make routine calls on behalf of the user , an idea previously seen in phone support tools and now coming to the Pixel mobile ecosystem. The feature, called "Call for Me," is currently in the experimental phase, so it's not intended as a stable feature or available to everyone.
The concept is simple: instead of dialing a number and waiting, the user describes a request, and the AI ​​takes care of the rest. Booking a table, checking if a store has an item in stock, or rescheduling an appointment are some examples of tasks Google aims to handle. The goal isn't to replace all human conversations, but rather to eliminate repetitive calls when they're inconvenient or don't feel like making them at that moment.
How Gemini's automatic calling works
To begin, the device must be compatible and the user must meet the eligibility requirements. The system relies on the Gemini app and the Phone by Google program , so the feature isn't available on every Android device or version. From there, the AI ​​dials the number, navigates through automated menus, selects options, and waits until a person answers.
When someone answers, the AI ​​identifies itself as an assistant calling on behalf of a user. It also warns that the conversation may be recorded , a common practice in many customer service systems. The user can follow a live transcript and, if needed, intervene to resume the conversation or correct its course.
User control and limits
Google emphasizes that the user remains in control. Before the call begins, a summary is displayed with the task, the number, and the data that will be shared , and no action is taken until permission is granted. During the conversation, the transcript allows users to check what is being said and join in when necessary. This oversight aims to build trust in an environment where the other party's responses can be unpredictable.
Not everything is allowed. Gemini cannot contact emergency services like 112 or 911 , nor can it make purchases with credit card details, or share sensitive information such as passwords, medical records, or Social Security numbers. Furthermore, since it's a test, there's a daily call limit. These restrictions aim to prevent misuse while Google studies how the tool behaves in real-world situations.
Availability: United States, Pixel 11 and closed beta
The initial rollout falls short of Spain and much of Europe. The test is limited to the United States and Gemini subscribers participating in the Phone by Google beta. Regarding hardware, the Pixel 11, Pixel 11 Pro, Pixel 11 Pro XL, and Pixel 11 Pro Fold are mentioned; some references also mention Pixel 11 models and higher.
For European users, therefore, the feature is not yet a real option. There is no confirmed date for its arrival in Spain, nor a public timeline for other countries. Google prefers to proceed cautiously because a real call can go in many different ways, and the system needs to learn from those scenarios before rolling it out to more people.
What does this mean for privacy and trust?
The proposal fits into a broader trend: assistants that act on behalf of the user. The company insists that the user remains in control and must authorize the process , which is why the feature displays the conversation and allows for intervention. Even so, the tool accesses personal data and conversations with third parties, so transparency will be key to avoiding backlash.
It also raises questions about how the person on the other end will feel. The AI ​​presents itself as an assistant and warns about the recording , but it's still an automated conversation. In Spain, for now, there's no option to try it, so any future arrival will have to comply with data protection regulations and address the understandable mistrust that these types of systems generate.
The race for assistants who do tasks
This test isn't happening in a vacuum. Google has been developing Gemini into a layer that interacts with apps, services, and devices for some time now , and automated calling is another step toward assistants that not only answer calls but also perform tasks. The company has already experimented with features for transcribing customer service calls and holding calls, and now it's taking things a step further by allowing AI to converse on behalf of the user.
The challenge is to make the system useful without being intrusive or annoying. The key will be in the limits, permissions, and the ability to cancel or resume the call . If it works, it could save time on daily tasks; if it fails, it could create confusion for both the user and the business receiving the call.
Gemini's "Call for Me" feature points to a future where our phones not only understand our requests but also handle tasks for us. For now, it's a US trial with specific requirements and clear limitations , so in Spain, we'll have to wait to see if it arrives and how it adapts to European regulations. The idea, in any case, is already on the table: delegating routine calls to an AI that speaks, listens, and waits on the other end.
