Use AI Studio in a canvas

Open AI Studio from a canvas you can edit, give it canvas context, and review the generated object where it lands.

3 min read

Generate AI output directly on the canvas where the work and its reviewers already are, instead of in a separate tab. This page is the canvas-entry workflow; for the full catalog of output types, prompt patterns, credits, and troubleshooting, see Use AI Studio.

Before you start

AI Studio creates canvas content, so it needs edit access. It is available to workspace members who can edit the canvas, and only after the workspace AI credit check passes at submit time.

It is not available to guests, external collaborators, view-only or comment-only access, embedded canvases, or workspaces without AI Studio. If you are a guest or external collaborator, ask a workspace member to generate the output. If you are a member and the panel is missing, ask the canvas owner for edit access, or ask a workspace admin to confirm AI Studio and credit availability.

Generate directly from the canvas

AI Studio uses the current canvas, viewport, and selection, so open the canvas where the result should live before you start.

  1. Open the canvas and confirm you can edit it.
  2. In the AI Studio panel, leave the type on Auto or choose Doc, Image, Table, Audio file, or Web page.
  3. Add the prompt and the context the result should use.
  4. Submit. A placeholder appears near the visible center of your viewport while the work runs in the background.
  5. Review the finished object before moving it into shared work.

Desktop shows the canvas AI panel; on phones and narrow screens that support it, the entry point moves into the canvas toolbar panel. If you only need to ask a question about one existing object rather than create something new, use Object Chat instead. See Ask about a canvas item.

Add canvas context

The canvas is what makes in-canvas generation different from a chat box: AI Studio can draw on what is already there.

ContextHow to use it
Selected objectsSelect notes, files, images, tables, or earlier results before submitting. ALLO reads your selection as source material; the result still lands near the center of your view, not on top of the selected objects.
Attached filesAdd source material with paste, drag and drop, or the paperclip button; wait for uploads to finish before submitting.
Pasted linksPaste safe HTTPS links the server can reach.
Voice inputDictate the prompt, then skim the transcript before submitting.

On a freeform whiteboard, a smart arrow connecting a source object to the result tells AI Studio and Object Chat to treat that source as prior context. Keep attachment sets small and purposeful; a failed, oversized, or still-uploading file can make the request fail or come back incomplete.

Check generation status

The placeholder is the live sign that AI work is in progress. During processing you can select, move, comment on, or delete it, but preview, edit, download, and AI follow-up stay blocked until it finishes.

To cancel a run, delete its processing placeholder; deleting the whole canvas also stops any generation still running there. Longer jobs continue server-side if you leave, so return to the canvas to review the result instead of resubmitting the same prompt. Credits already spent are not refunded when you delete a finished result.

Review the generated result

Preview the result and check source accuracy, formatting, tone, and sensitive content before it enters shared work. Then edit it, move it near its source, add comments for review, share the canvas, or delete it if it is wrong or misleading.

Do not leave unreviewed AI drafts in a shared review area, and do not regenerate repeatedly when manual editing would be faster, since repeated AI work spends more credits. Before using AI output in client-facing or decision-heavy work, read Review AI output and share web pages.

Give feedback

Was this article helpful?