On Wednesday, Google introduced Search Live in AI Mode, marking a new experimental feature for users in the United States. This innovative tool allows for interactive, real-time conversations within the Search experience, resembling the functionality of Gemini Live but integrated into the Google app. The feature was initially revealed during the company’s I/O 2025 keynote presentation in May as a complement to the existing AI Mode. Powered by a tailored version of Gemini, Search Live incorporates voice input capabilities, enabling users to ask not only initial queries but also to pose follow-up questions seamlessly.
Search Live in AI Mode
In a recent blog post, Google highlighted that Search Live utilizes the same customized version of Gemini found in AI Mode and comes with enhanced voice features. Users can activate Search Live by tapping a waveform icon located just below the search bar in the Google app, available on both Android and iOS platforms. This action leads to a full-screen interface offering two options: mute and transcript.
The transcript feature allows users to view text responses and continue their interactions by tapping, instead of relying solely on voice commands. Additionally, the interface provides links to relevant websites for users seeking further information.
According to the company, Search Live operates seamlessly in the background, enabling users to exit the Google app while maintaining their conversation with Search. Furthermore, conversations can be revisited by accessing the AI Mode history. Google notes that the feature uses a query fan-out technique to deliver a variety of useful content from the internet.
This feature is currently in the testing phase as part of the AI Mode experiment through opt-in participation in Search Labs, and it remains exclusive to US users for the time being, with no announced plans for expansion to other regions.
While Search Live in AI Mode primarily focuses on voice-based commands at the moment, Google has plans to implement a future capability allowing users to point a camera at an object and ask real-time questions about it.
An earlier report indicated that this feature supports four distinct voice options—Cosmo, Neso, Terra, and Cassini—available for selection through a three-dot menu in the Search Live interface.
