Google 's Gemini Call for Me is getting ready to expand beyond business calls, bringing artificial intelligence to our personal communications. According to a discovery by The Verge and Android Authority , Google appears to be testing a new feature that would allow Gemini to make short personal calls on your behalf — for example, alerting your mom that you'll be 15 minutes late. The revelation came from an APK teardown rather than an official announcement, so the final release remains uncertain.
Android Authority spotted an introductory screen called “Gemini Calling” inside an APK , which includes example commands like “Call Mom and tell her I will be 15 minutes late” and “Call John and ask if he is coming for dinner .” These examples suggest that the feature will be limited to short, structured communications — essentially automated voice messages — rather than extended social conversations. It also found granular permissions for Gemini ’s actions on individual apps, including settings for phone call permissions .
See also: WhatsApp and Slack notifications can hack Google Gemini on Android
The development is part of a broader strategy by Google to transform Gemini from a simple text-producing assistant into a full-fledged AI agent that takes action across apps and services. This journey began in 2018 with Google Duplex, which first demonstrated that AI could conduct natural phone conversations with businesses for bookings and appointments — while also sparking heated debates about transparency and consent.
Gemini Call for Me: From business to personal calls
The current version of Gemini Call for Me focuses on business calls: Gemini navigates automated menus, waits on hold, speaks to agents, and produces a transcript or summary of the conversation. Availability is currently limited to select users in the US, Google AI , and supported Pixel. The potential expansion to personal calls represents a quantum leap: instead of communicating with anonymous service representatives, Gemini will call people who know the user personally.
This raises important issues. The recipient of an automated call from Gemini has not necessarily consented to receive such a call or to have the conversation recorded. Google seems to have taken this issue into account: the granular permissions found in the APK will allow users to choose which app capabilities they want to grant to Gemini, and there may also be an opt-out for receiving automated incoming calls. In the current implementation for business calls, there are already daily call limits and an opt-out for businesses and individuals.
Technically, the feature requires multiple layers of processing: speech recognition to convert the user’s command and the recipient’s response into text, natural language understanding to extract the recipient, purpose, and message, contact recognition (mapping “Mom” or “John” to the correct contact), dialog management, telephony integration , and real-time inference for natural pauses. Each of these steps introduces potential errors — an incorrect name or arrival time can create issues.
See also: Gemini 4 Argon: Google unveils new frontier AI model
Benefits and challenges
The benefits for users are obvious: help for people who are driving, busy, have mobility or hearing difficulties, or just want to send a quick message to someone who prefers phone calls or still has a landline. But the risks are just as real: misidentified contacts (“Mom” could be the wrong person), incorrect message content, social friction from the cold feel of a robocall, exposure of private information through transcripts, and concerns about impersonation — the recipient might think the user made the call themselves.
See also: RatHat Android Malware Console uses Gemini to locate victims
In conclusion, Gemini Call for Me could represent an ambitious expansion of Google’s AI-powered personal communications. While the findings from the APK teardown don’t guarantee a public release, the direction is clear: Google wants Gemini to take on more and more communication tasks on behalf of users. Success will depend on the reliability of the technology, transparency towards recipients, and respect for privacy — issues that Google knows well from its experience with Duplex in 2018.
