Google Opens Gmail to Voice and Gates Docs Live

Google is rolling out spoken Gemini tools in Gmail, Docs, and Keep this week for paying AI plan customers on phones. The names are Gmail Live, Docs Live, and Keep Live.

You can ask your inbox a question out loud, dump a thought into Keep, or talk a blank page into a first draft. Docs Live, the drafting tool, stays on Google AI Pro and Ultra. Gmail Live and Keep Live also sit on Plus.

Gmail Live Turns Search Into a Spoken Chat

Yulie Kwon Kim, vice president of product for Google Workspace, said the company is bringing Gemini Audio models into Workspace so daily tasks can be done with a voice. Google first showed the tools on May 19 at I/O. The conversational features rolling out this week skip the search box and let you talk.

Gmail Live is built for the moment you cannot pick through threads. You ask for a fact, the assistant scans mail, and you get an answer while a live transcript runs on screen. Kim’s post uses the same two examples Google has used since spring, because those are the jobs the product is aimed at: a gate number, and a school week.

PROMPTS GOOGLE USED IN ITS DEMOS

• Flight details: “What’s my flight’s gate number?”
• Family calendar: “What’s going on at my kid’s school this week?”
• A cooking list: Tell Keep to list ingredients for shakshouka, and it builds the list from the rant.
• A first draft: Talk through a proposal or summary in Docs Live, and the assistant structures the page while it works.

That is inbox search recast as a conversation, which is a different habit than typing a keyword and scrolling. It also means Gemini is reading the mail that already sits on Google’s servers, then speaking the result back. Sundar Pichai, chief executive of Google and Alphabet, framed the launch as one bucket of customers.

New Gemini voice capabilities are rolling out so you can search your Gmail inbox, organize thoughts and tasks in Keep, and create new Docs, all conversationally. Rolling out now for Google AI subscribers!

Sundar Pichai, CEO, Google and Alphabet, on X

The CEO line is broader than the SKU list. Plus customers can talk to Gmail and Keep. They cannot talk a document into existence unless they pay for Pro.

The $15 Jump to a First Draft

Google’s consumer plans page lists Google AI Plus at $4.99 a month with 400 GB of storage and 2x Gemini access. Pro is $19.99 a month with 5 TB and 4x access. Ultra starts at $99.99 a month with storage from 20 TB and up to 20x access.

The gap between Plus and Pro is $15 a month. That $15 is the price of Docs Live. Google calls Docs Live a hands-free thought partner and co-writer. You talk, it organizes the thought, structures the file, and, with your permission, pulls detail from Gmail, Drive, Chat, and the web.

That permission clause is the product. The assistant is not a blank-page chatbot. It is a spoken layer on top of mail, files, and chat you already gave Google, which is why the draft can mention a flight, a thread, or a Drive folder without you pasting anything. It is also why the company put that layer on Pro.

WHO GETS WHICH LIVE FEATURE

At I/O, all three tools were listed for Pro and Ultra only. By September 3, Gmail and Keep had dropped to Plus. Docs did not. The inbox and the note are the sample. The blank page is the upgrade.

Pichai’s own demo leans on that upgrade. He said Docs Live would be “super helpful in particular,” then posted a clip of the drafting loop rather than an inbox lookup.

https://x.com/sundarpichai/status/2095552297099198689

People who already pay $4.99 will feel that clip as a reminder that the useful tier is one plan above. Plus gets you talking to mail. Pro gets you a co-writer that can see mail, Drive, and Chat at the same time.

A Scratchpad That Files Itself

Keep Live is the quiet one, and Google put it on Plus with Gmail. Keep is where a thought goes before it has a title. The new mode treats that mess as the input.

Kim’s post tells you to brain dump. The feature “understands your stream of consciousness and processes it in the background,” then turns the talk into lists and notes with no typing. You can ask it to build a recipe list. You can ramble and let it sort the pieces after you stop.

That is a different job than Gmail Live. Gmail Live retrieves. Keep Live files. Docs Live composes. Only the third one needs Pro. For a $4.99 customer, the new habit is speaking a grocery list or a school reminder into Keep, then asking Gmail Live what the school actually sent.

Keep is also the easiest place for the speech model to look clever, because a list has a clear shape. A proposal does not. If the assistant mishears a spice, you delete a line. If it mishears a clause in a contract draft, you have a cleanup job. Google’s packaging follows that risk.

How Gemini Audio Reached Gmail and Docs

The Live names are new. The speech stack is not. Google has been moving Gemini Audio through phones, the Mac app, and developer APIs all year, then dropping the same listening layer into Workspace.

THE SPEECH STACK SINCE SPRING

1. April 2026: Adds cross-app dictation to the Gemini app on Mac, so you can speak into the window already in front of you.
2. May 19, 2026: Previews Gmail Live, Docs Live, and Keep Live at I/O for Pro and Ultra, and promises a Workspace business preview for summer.
3. August 12, 2026: Ships Pixel 11 Gboard Rambler dictation, which turns rambling speech into cleaned text and strips filler words.
4. August 26, 2026: Releases Gemini 3.5 Transcribe, the model behind Rambler, with a 70 percent faster time to final transcription than Chirp 3.
5. September 3, 2026: Rolls out Gmail Live, Docs Live, and Keep Live for consumer AI plans, still listing Workspace business as coming soon.

Diego Melendo Casado, senior director of engineering for Gemini Audio, and Luke Leonhard, chief of staff for that team, said 3.5 Transcribe converts raw audio into polished, formatted text. It handles mid-sentence corrections, drops “ums” and “ahs,” and formats as it goes. Artificial Analysis measured an average word error rate of 4.0 percent for streaming and 2.6 percent for non-streaming. On the FLEURS benchmark, streaming word error is 5.50 percent and non-streaming is 5.04 percent. The model detects more than 85 languages. For recorded audio it can label up to three speakers.

Rambler on Pixel 11 was the consumer proof. The Mac Gemini app already uses the same cleanup on a desktop cursor. Chrome is next, Google said, so you can talk into any web field. The Workspace Live features still run on Android and iOS. The model is spreading faster than the named Live buttons.

Four Billion Workspace Users Are Still Waiting

On May 19, Kim wrote that more than 4 billion people rely on Workspace apps like Gmail, Docs, and Drive. The same post said the voice tools would roll out to Pro and Ultra “and in preview to Google Workspace business customers this summer.”

Summer ended. The September 3 post still says all of these features are coming soon for Workspace business customers. Consumer Plus, Pro, and Ultra get the buttons first. The work accounts that sit behind an admin console do not, at least not as a finished product Google is willing to date.

Google’s AI plans page is also tighter on maps than the launch copy. It lists Gmail Live (Plus, Pro, Ultra) and Docs Live (Pro, Ultra) as rolling out on Android and iOS in the United States. Kim’s launch post does not name a country. It does say the consumer features are rolling out this week, in English, on phones.

WHAT WE KNOW

• Consumer timing: The three Live features are rolling out the week of September 3 for paying Google AI plans.
• The split: Gmail Live and Keep Live are on Plus, Pro, and Ultra; Docs Live is Pro and Ultra only.
• The next SKU: Workspace business is still listed as coming soon, with no date.

WHAT IS UNCONFIRMED

• Keep on iOS: The plans page names Android and iOS for Gmail Live and Docs Live, and is silent on Keep Live’s phone list.
• Where it runs: The plans page says United States for Gmail Live and Docs Live; the launch post does not.
• The business preview: Google has not said whether any Workspace tenants already have a quiet preview, only that the features are coming soon.

That delay is the part a consumer launch post treats as a footnote. For a company that counts more than 4 billion Workspace users, the money is in the admin-controlled seat, not the $4.99 personal plan. Personal Gmail is the demo. The demo is live. The fleet is not.

Voice on the Desktop Arrived From Microsoft First

Microsoft already runs spoken Copilot inside the apps that compete with Gmail and Docs. Voice chats in Microsoft 365 Copilot work on the web and in the Microsoft 365 app on iOS, Android, Windows, and Mac. Outlook has a voice catch-up path for mail and calendar. Word and PowerPoint got in-app voice chat on desktop and web in late 2025. Microsoft has said users interact with Copilot twice as much when they use voice instead of text.

Google’s Live buttons, as shipped on September 3, are a phone conversation with a single app at a time. Gmail Live talks to mail. Docs Live talks to a document, then asks permission to read mail, Drive, Chat, and the web. Keep Live talks to notes. That is not a “Hey Gemini” wake word on a laptop, and none of it is the Chrome field dictation Google has already promised for 3.5 Transcribe.

The company can close that gap. The Mac app already takes cleaned dictation. Chrome is on the Transcribe roadmap. Workspace business is on the Live roadmap. Until those land, the person who pays $19.99 and works in a browser still drafts with a keyboard, while the person who pays $4.99 can only talk to the inbox and the scratchpad.

Pichai’s clip will get the taps this week. The test is whether Docs Live is good enough that Plus customers pay the extra $15, and whether the same voice loop shows up in a work account before the next I/O.