On August 26, 2026, Google announced a major update to Gemini Live, the real-time voice mode in the mobile Gemini app, that folds several agentic capabilities into a single conversation. Four pieces anchor the release: integration with the Spark agent, a spoken Daily Brief, hands-free Gmail management, and Personal Intelligence for recalling past context. A feature built for talking now does the work you hand it.

Requests made mid-conversation keep running in the background

The headline change is that Gemini Live can now invoke Spark directly. Spark is an agent that works across Google Docs, Sheets, Drive, and the web, holding on to your goal as it proceeds. It handles long-running and scheduled jobs that stretch across days or weeks, and it keeps working even when the Gemini app is closed.

Google's examples are concrete. Talk through the loose fragments in your head while you are on the move, and Spark will sort out the themes so a structured outline is waiting in Google Docs by the time you sit down. Or have it maintain a weekly meal plan for your family and build a grocery list from recipes already saved in Docs. Neither task finishes the moment you ask for it.

Voice interfaces suit short queries but have long been a poor fit for multi-step work. Putting Spark behind the microphone is Google's way around that limit.

Answering "what should I care about today?" out loud

Daily Brief can now be pulled up from within Gemini Live. Ask "what's my daily brief?" and Gemini gathers the important messages and appointments from Gmail and Calendar, then reads back a summary of what the day holds.

The point is to grasp the shape of your day without looking at a screen, which fits the moments when you are getting ready. Compared with working through notifications one at a time, it flips the order in which information reaches you, from the order it arrived to the order that matters.

Sorting Gmail using only your voice

Inbox handling has also moved into Gemini Live. Beyond answering questions like "any new emails?" or "anything urgent from the kids' school today?", it can search, summarize, star, archive, and delete messages by voice.

Because there is no need to open a mail app and tap through it, triage can happen while your hands are busy. Narrowing things down conversationally also feels different from typing terms into a search box.

Remembering "that restaurant we went to last summer"

Personal Intelligence ties together your past chat history and the information in your connected Google apps, including Gmail, Photos, Search, and YouTube.

Google's examples are questions such as "what was the name of that restaurant we went to when we visited New York last summer?" or "can you pull up the recipe we talked about yesterday?" The system takes the vague phrasing people use when reaching for a memory and finds the matching item across history and connected apps. Not having to re-explain yourself from scratch each time is a large part of what makes a voice assistant usable.

Gemini decides which tool to use

None of these four features requires the user to switch modes explicitly. Ask "what does my day look like?" and Daily Brief answers. Follow up with "grab those event invites from my email, add them to the family calendar, and include the driving times," and Gemini hands the job to Spark on the spot. The right tool is chosen automatically within a single conversational thread.

Removing the need to memorize feature names addresses a problem shared by every AI assistant that has accumulated multiple agentic capabilities. The more features there are, the harder the entry point becomes to find, so narrowing that entry point down to voice alone makes sense.

Getting started means checking your plan and your connections

Access differs by tier. Spark requires a Google AI Pro subscription or higher, and Daily Brief requires a Google AI Plus subscription or higher. Not everything is available within the free tier.

The setup itself is simple: enable your app connections in Gemini's Personal Intelligence settings, then tap the Live icon in the Gemini app and start talking. Leaving those connections off means Gemini cannot reference Gmail or Calendar, so the behavior may not match expectations.

Google also disclosed alongside the announcement that 63 percent of Gemini users speak to it out loud. Devoting resources to voice-led feature expansion looks like a decision grounded in that figure.

Summary

This update widens the scope of the voice assistant from answering on the spot to taking a job and finishing it later. The Spark integration allows long-running tasks to be delegated, Daily Brief and inbox controls let the morning's information triage happen entirely by voice, and Personal Intelligence carries context forward. Together with the design choice of leaving tool selection to Gemini, it raises the practical utility of operating AI by voice a notch.

※画像はイメージです