Google has introduced Continued Conversation for Gemini on Home devices, allowing users to issue follow-up commands without repeating the wake word after each query [source]. The feature, announced April 21, keeps the microphone active for several seconds after an initial voice command, enabling multi-turn exchanges that more closely resemble natural dialogue.

According to the company blog post, users can now say "Hey Google" once, then ask a question, receive a response, and immediately ask a follow-up without re-invoking the assistant. Google states the feature is designed to reduce friction in common tasks such as setting multiple timers, asking related questions, or controlling smart home devices in sequence.

The implementation differs from previous voice assistant behaviour, where each discrete command required a new wake word. Google has not specified the exact duration the microphone remains active between turns, nor detailed what signals terminate a conversation session. The post notes that Continued Conversation is rolling out to Gemini-enabled Home devices and can be managed in assistant settings.

Privacy implications centre on extended microphone access. While Google has historically positioned always-listening devices as activating only after wake word detection, Continued Conversation extends that window. The company has not published technical documentation on how the feature distinguishes intentional follow-ups from ambient speech, or whether conversation audio outside confirmed commands is processed or retained.

The feature represents a shift in interaction model for Google's voice assistant ecosystem. Users who prefer single-command interactions can disable Continued Conversation through device settings. Google has not indicated whether the feature will expand to mobile Gemini implementations or other surfaces.