Can your phone finally handle hold music for you?

Google is rolling out an experimental feature called Call for Me for users of its newest phone, the Pixel 11. If you pay for a Gemini subscription and use a specific test version of the phone app, you can ask the AI to call a business on your behalf. Whether you need to move an appointment, check if a store has an item in stock, or make a reservation, the AI handles the dialling, the phone menus, and the time spent on hold. You can watch a live, written transcript of the call on your screen as it happens and jump into the conversation yourself at any moment.

This feature relies on a large language model, which is an AI system trained on vast amounts of human text to understand and mimic natural speech patterns. When you give the command, the AI treats the phone call like a conversation it is having on your behalf. It uses its language skills to introduce itself and understand what the person on the other end of the line says. Because it is designed to recognize the logic of automated phone systems, it can navigate through menu prompts—like pressing four for customer service—without you having to listen to the recorded list yourself. When it reaches a real person, it holds a back-and-forth dialogue using the same voice technology you might already use for other AI interactions on your phone. It is essentially an apprentice that you have trained to handle the awkward social parts of a professional phone call.

The reality of this technology is a mix of convenience and uncertainty. While it promises to save us from wasted time, real-world calls are rarely perfect. A human on the other end might hang up because they suspect it is a spam call, or the AI might struggle with a thick accent or a noisy background. Google is running this as a test because these interactions are messy and unpredictable compared to simple computer tasks. This is a step toward a future where our devices act more like personal assistants than mere tools, but it also highlights that we are still in the early stages of teaching software how to handle the nuances of everyday human life.