Google is pushing Gemini further into the agentic AI era, testing a capability that allows the AI assistant to make and handle phone calls on a user’s behalf—moving beyond answering questions toward completing real-world tasks.
The new “Call for Me” capability can contact businesses to check product availability, obtain service quotes, book or modify appointments and reservations, navigate automated phone menus, and even wait on hold.
Users first tell Gemini what they want accomplished and review details—including information that may be shared—before authorizing the call. Gemini then uses the user’s own phone number to place it.
Importantly, users remain in control. They can monitor a live transcript, listen to the conversation or take over the call whenever necessary. Gemini also identifies itself as a Google AI assistant calling on the user’s behalf.
The initial rollout is limited. Google says the feature is gradually becoming available to eligible U.S. users aged 18 or older using a Pixel 11, a Google AI subscription and the public beta version of Phone by Google.
There are safeguards: Gemini cannot make emergency calls, conduct financial transactions, share sensitive financial or health information, or perform telemarketing. Google also says call recordings and transcripts are not used to train its AI models.
The bigger significance is the transition from AI assistants to AI agents. Instead of simply recommending what users should do, Gemini can increasingly execute tasks for them—making identity, consent, transparency and human oversight critical as autonomous AI enters everyday communications.
