REVLAI in PracticeAI can keep talking while it works

AI BRIEF / 006

Still talking.
Already working.

Gemini 3.8 Live, explained through two animated lanes: the conversation and the work happening alongside it.

Conversation and background work illustrated as two separate lanes
Original REVL concept illustration. No product recording or independent model test.

Google describes Gemini 3.8 Live as able to continue a conversation while tools work in the background. Its Extended Thinking variant adds spoken progress updates during more complex tasks. The announcement was published on 15 September 2026 and updated on 17 September. Read Google's announcement.

Two lanes, one interaction

Our animation turns that idea into two visible lanes. The upper lane carries the conversation. The lower lane shows an illustrative task. Keeping them separate helps explain an important distinction: hearing an acknowledgement and receiving a completed result are different events.

Consider a fictional calendar search. You ask for available times, then clarify that you prefer the afternoon. We animate the request, the extra constraint and a set of possible results. This is a conceptual example created by REVL, not a recording of Gemini, a real calendar connection or proof of a particular integration. Nothing is booked.

Progress is useful. Completion is different.

A spoken update can help someone understand that work is continuing. Our editorial question is what happens at the end. Does the returned result match the latest instruction? Is it clear whether an action was merely proposed or actually performed? Could you inspect and correct the outcome?

For our fictional search, a useful result would show afternoon options and leave the final choice to the user. A friendly acknowledgement on its own would not answer the original request. That distinction is the central point of the illustration.

What this piece establishes

The capability descriptions above are attributed to Google. REVL has not independently tested these models, measured their latency or verified account-specific availability. The graphics, timing and example messages are original editorial material. They do not reproduce a product interface. The narration is synthetic and is not Gemini's voice.

This is an explanation of a September release, not a breaking-news claim. Source checked: 29 September 2026.