The early preview, announced on September 24, enables Gemini to contact shops, restaurants and service providers, navigate automated menus, wait on hold and converse with staff. Users can ask it to check whether an item is available, obtain business information or service quotes, make or alter reservations, and book, confirm or manage appointments.
Google is gradually making the capability available in the United States to eligible users aged 18 or over. Participants need a Pixel 11 with a US SIM card, a Google AI subscription, the public beta version of the Phone by Google app and the latest Gemini mobile app. Their device language must be set to English.
The company said the preview would roll out over two weeks to paid Gemini subscribers enrolled in the Phone app beta. Google described the programme as an early experiment, saying real-world conversations can be nuanced and that it is starting on a limited scale while refining the experience.
A user begins by telling Gemini what needs to be accomplished rather than manually dialling the business. Gemini can ask follow-up questions where necessary before initiating the call. Once connected, it identifies itself as an AI assistant and explains that the call is being recorded, before proceeding with the authorised request.
The call is made using the user’s phone number, according to Google’s documentation. Gemini may provide information that the user has supplied or approved when this is necessary to complete a task. That could include details required to move an appointment, reserve a table or ask a retailer to hold an item.
Google has built several controls into the experiment. A live transcript lets the user follow the conversation as it happens, while an option to take over allows the person to intervene and continue speaking directly at any point. The system also provides information about the call after it has ended.
The test extends Google’s existing suite of telephone assistance tools on Pixel devices. Hold for Me can remain on the line while a caller is placed on hold and alert the user when a representative becomes available. Direct My Call displays automated telephone-menu options on screen, reducing the need to listen through an entire recorded menu.
Gemini’s new capability combines those functions with an AI-generated voice that can conduct the conversation itself. That makes the experiment broader than tools designed merely to transcribe a menu or wait for a human operator, while placing greater emphasis on disclosure and user supervision.
Google has already deployed automated business-calling technology in other contexts. Its shopping service can contact nearby stores to check product availability after a user submits a request, while an earlier Ask for Me function has been used to seek information such as prices and services from businesses. The Pixel 11 experiment adds the ability for users to monitor and take over a live call.
The company says calls made through the new feature represent genuine customers carrying out genuine transactions, including booking appointments, placing products on hold and changing reservations. It has also acknowledged that automated calling affects the businesses and employees receiving those calls, a factor behind the deliberately narrow initial deployment.
The feature is not intended for every type of telephone interaction. Google’s guidance limits supported use cases and bars certain categories, while the initial eligibility requirements sharply restrict who can participate. The company has not announced a timetable for extending the feature to other Pixel models, countries, languages or users without a paid AI subscription.
