Google adds prompt-built mini-apps to Sheets

• Google’s September subscription update puts Gemini inside the work surface, bundling voice controls, generated Sheets interfaces, image editing, and browser tasks in Google's overview.
• Sheets canvas turns a spreadsheet into a prompt-built interactive mini-app, according to Google's Sheets canvas announcement.
• Voice input now reaches Gmail, Docs, and Keep for drafting, inbox retrieval, and structured note capture in Google's voice announcement.
• Google Pics brings Nano Banana image generation and editing into Workspace, as Google's Pics rollout describes.
• Gemini Spark can take on Chrome errands and Photos work, including edits and album curation, in Google's Spark update.

The Workspace release note says a Gmail prompt can create a formatted Doc or structured Sheet in the background and save it to Drive. The same page limits launch support to English, subjects advanced AI features to usage limits, and excludes scheduling from consumer AI Pro.

Canvas is a generated, read-write interface layered over a Sheet, with edits synchronizing with the underlying data.

Google’s Sheets canvas guide describes four mechanics:

• Entry point: select the Gemini icon in Sheets, then choose Create canvas.
• Data model: the canvas and its source spreadsheet stay synchronized in real time.
• Iteration: follow-up prompts change the layout, design, and functionality.
• Sharing: the canvas lives in a normal Sheets tab.

The guide lists global English availability for Business and Enterprise Standard and Plus, the Google AI Pro for Education add-on, and consumer AI Pro and Ultra.

Google calls its cross-app routing layer Workspace Intelligence, and the rollout note says Gemini can be prompted from Chat, Drive, Docs, Slides, or Gmail.

The documented action set has five parts:

1. Create formatted Docs, structured Sheets, or stylized Slides in the background, then save them to Drive.
2. Research large folders or long threads into a report with source attributions.
3. Draft an email from notes or a document, then open it in Gmail or send it.
4. Find time, schedule meetings, and resolve calendar conflicts.
5. Turn context into Google Tasks reminders or to-do lists.

For email sends and calendar commitments, the same update says Gemini presents an editable preview card before execution. Google also says the agent inherits the authenticated user’s existing document permissions, and Workspace data is neither human-reviewed nor used to train Gemini.

Gmail Live handles conversational inbox lookup, Docs Live helps organize spoken drafts, and Keep Live turns spoken brain dumps into notes and lists.

Google’s feature details say Docs Live can pull context from Gmail, Drive, Chat, and the web with user permission. Gmail and Keep are available on AI Plus, Pro, and Ultra, while Docs requires AI Pro or Ultra; business access arrives later in the rollout.

Pics is Google’s embedded image tool, built on its Nano Banana generation and editing model.

The [initial Google post](src:2|Google's initial Pics announcement) lists AI Pro and Ultra access, while [Google's rollout post](src:12|its later Pics update) adds most Workspace business customers. Pics is rolling out in Docs and Slides, with Drive next, according to Google's integration update.

Google’s Pics feature page lists object-specific edits, in-image text editing and translation, multiple generations per prompt, and real-time collaborative editing.

Spark moves the agent boundary past Workspace. Google says it can work through Chrome and Google Photos for AI Pro and Ultra users in the U.S.

With permission, Spark can use logged-in accounts and saved passwords to schedule an apartment viewing, research flights, or start a booking, according to Google's Chrome integration details. That page names two controls: prompt-injection safeguards for browser work, and a handoff to the user for sensitive actions such as payments.