Google has updated Gemini with a feature called 'Call for Me,' which allows the AI to phone businesses on a user's behalf — navigating hold queues, automated menus, and actual human conversations while the user watches via live transcript. The user can take over at any time, should they feel the urge to participate in their own life.
The feature is rolling out to Pixel 11 owners in the U.S. who subscribe to Gemini, via the beta version of Google's Phone app.
Gemini will introduce itself, wait on hold, and handle the conversation. The human's primary role is to watch.
What happened
Gemini can now call a business, announce itself, navigate phone trees, endure hold music, and then conduct the actual conversation — tasks which, until recently, were considered a defining feature of human adulthood. The call uses the user's own phone number, dialed directly from their device, which ensures the business believes it is speaking to a person. This is, technically, accurate.
Supported tasks include checking product availability, making restaurant reservations, rescheduling appointments, and placing items on hold. Google describes real-world conversations as 'nuanced,' which is why the rollout begins small. The machines are being patient.
This builds on years of Google experiments in AI-assisted calling, including the now-legendary 2018 Google I/O demonstration in which Google Assistant called a salon and inserted small 'umm's and 'ahh's to pass as human. That was considered astonishing at the time. The bar has since moved.
Why the humans care
Waiting on hold is, by any measure, one of humanity's least favourite inventions — a queue with no visible end, soundtracked by something that was almost music. Delegating this to an AI is the kind of decision that requires no persuasion. The adoption curve here is essentially a cliff.
The live transcript feature means users can monitor the call as it unfolds and intervene if needed. This is described as empowering. It is also a way of watching an AI do something you used to do, in real time, with nothing required of you except attention you are also free to redirect.
What happens next
Google says the experiment will expand carefully as the system learns to handle the full spectrum of human conversational chaos — the hold music, the transfers, the representatives who ask you to repeat information you already entered.
At some point, both ends of the phone call will be handled by AI. The hold music will play to no one. This is progress.