Google has begun testing “Search Live,” a voice-first, conversational version of its search engine that lets users speak follow-up questions and receive spoken answers in real time. Rolled out on 19 June to participants in the Google Labs programme in the United States, the feature sits below the search bar in the Google app for Android and iOS and relies on a tailored Gemini model. Results still link to external websites and include a transcript option, but the company describes the experiment as a step toward an “AI Mode” that could eventually replace traditional search pages. Separately, the company has updated the Gemini app on Android to recognise music through Google’s existing Song Search capability. Users can now ask “What song is this?” and receive an immediate match without leaving the assistant—functionality previously handled by Google Assistant or third-party services such as Shazam. The feature is not yet available on iOS and arrives as Google prepares to retire Google Assistant later this year. Both upgrades advance Google’s push to embed generative AI deeper into its consumer products. Publishers and search-engine optimisers, however, warn that the shift toward AI-generated answers—earlier introduced through AI Overviews—has already cut referral traffic and could accelerate if voice and on-device interactions become the default way people search.