Google App Gets Gemini-Powered Search Live Feature in AI Mode With Voice Input Support

Google on Wednesday rolled out Search Live in AI Mode as a part of a small experiment in the US. This function lets customers have back-and-forth conversations in real-time in Search, just like Gemini Live however in the Google app. It was first introduced in the course of the Mountain View-based tech large’s I/O 2025 keynote session in May as an extension to the AI Mode. Search Live is powered by a customized model of Gemini and has voice enter capabilities, enabling the flexibility to not simply ask queries, but additionally follow-up questions.

Search Live in AI Mode

In a weblog publish, Google stated that Search Live is powered by the identical customized model of Gemini as AI Mode and options superior voice capabilities. It might be accessed by way of a waveform icon simply beneath the search bar in the Google app on each Android and iOS gadgets. Tapping this icon opens a full-screen interface for Search Live with two choices — mute and transcript.

The latter permits customers to view a textual content response and proceed the dialog by tapping as an alternative of verbal instructions. It may also present hyperlinks to related web sites on the display screen in case they want to entry extra particulars.

As per the corporate, Search Live works in the background and thus, customers can give up the Google app and nonetheless proceed their dialog with Search. Further, they will additionally revisit conversations by navigating to the AI Mode historical past. Google says that the function leverages a question fan-out method to current customers with a various vary of useful content material from the online.

The function is at the moment being examined as a part of the AI Mode experiment by way of opt-in in Search Labs. It is barely out there in the US and Google is but to broaden it to different areas.

Although Search Live in AI Mode is at the moment restricted to voice-based instructions, the corporate says it can additionally introduce the flexibility to level a digicam at an object and ask questions on it in real-time, in the approaching months.

A earlier report means that this function helps 4 totally different voices — Cosmo, Neso, Terra, and Cassini — which might be chosen from the three dot menu in the Search Live interface.