Use AI Studio

That pile of meeting notes, screenshots, product links, and half-written copy does not need a detour through a separate AI tab. If the output belongs on a canvas, make it on the canvas, next to the work it came from and the people who will review it.

AI Studio is the canvas creation surface for that job. Use it to turn prompts, selected canvas objects, uploaded files, pasted URLs, and voice input into canvas objects you can move, edit, comment on, ask about, and share.

Treat the result as a draft. AI Studio is good at getting the first shape onto the canvas. You still own the facts, tone, dates, source permissions, and final handoff.

The short version

1. Open the canvas where the result should live 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. Write the prompt and add the selected objects, files, or URLs the result should use.
4. Submit, then watch for the placeholder near the visible center of the canvas while the server works.
5. Review the finished object before using it in shared work, or open Object Chat on that object when the next step is a follow-up question or edit.

AI Studio appears for workspace members who can create

AI Studio is for creating canvas content. It is available to workspace members who can edit the canvas. It is not available to guests, external collaborators, read-only access, comment-only access, or embedded canvases.

Low credits are handled at submit time. AI work starts only after the workspace AI credit gate passes. Workspace admins can buy credits in Billing. Other members need an admin to add credits. Guests and external collaborators do not use AI Studio.

Create from the canvas

Start from the canvas where the result should live. AI Studio uses the current canvas, current viewport, and current selection, so the page you have open matters.

1. Open the canvas where the result should live.
2. Confirm you can edit the canvas.
3. In AI Studio, choose Auto, Doc, Image, Table, Audio file, or Web page.
4. Add the prompt and any source context.
5. Submit.
6. Watch the placeholder on the canvas.
7. Review the finished object before you move it into a final section or use it in shared work.

If AI Studio is missing, check whether you are a workspace member and whether you have edit access. If you are opening an embedded canvas, open the canvas itself. If you are a guest or external collaborator, ask a workspace member to generate the output.

Choose the output type

Auto is the safest starting point. It lets AI Studio resolve the output type from the prompt and context, then starts the right flow.

For images, Count starts at one image and supports up to 10. Model, Style, and Ratio can stay on Auto unless you have a specific reason to trade cost, timing, or format. Auto is not a lazy setting here; it is the planned path that lets the backend pick a model and ratio from the prompt and references.

Change these controls when the result has to match a specific production format, such as a vertical social image, a square concept tile, or a wider presentation visual. If the format does not matter yet, leave them on Auto and spend the review time on the prompt and source context.

Give AI Studio the right context

The best prompt is not the longest prompt. It is the one that gives AI Studio the material it actually needs.

New placeholders are anchored near the visible center of your viewport, not on top of the selected objects. That is deliberate: selected objects can be off-screen, and generated work should appear where you can immediately see it, move it, or delete the placeholder to cancel.

On a freeform whiteboard, smart arrows can make source relationships clearer for both people and AI.

Use them when the source-to-result relationship matters:

• Reference image to generated draft.
• Research quote to requirement.
• Rough sketch to refined version.

When an arrow is attached from one real object to another real object, AI Studio and Object Chat can treat the source end as previous context for the target.

Attach files without turning the prompt into a junk drawer

AI Studio accepts up to 10 attachments in one draft. Supported attachment sources include images, PDF, Office and HWP files, Markdown, text and code files, CSV, JSON, SRT, and spreadsheets. AI Studio reads files for context with a 20 MB per-file cap, so split oversized source material before it becomes a failed or noisy request.

Completed uploads are included in the request. In-progress uploads are not. Failed uploads stay visible with an error state so you can remove them and try again.

Draft recovery stores completed upload metadata, prompt text, selected context, and options. It does not store raw in-progress file bytes. If you refresh while a file is still uploading, attach that file again before you submit.

When Auto sees completed attachments or prompt URLs, AI Studio runs an intent check before generation:

This matters in real work. If you paste a product URL and ask "what does this say?", you want a source object plus Object Chat, not a random new document. If you paste the same URL and ask for a client-ready launch brief, you want a generated Doc.

Write prompts that leave less cleanup

Good prompts make the review easier later. Give the model the decision shape you want, not just the topic.

If the result is bland, do not keep regenerating the same vague request. Add the missing audience, format, source object, or constraint. Manual editing is often cheaper than asking AI to guess again.

Voice input is prompt writing, not generation

Voice input turns speech into prompt text. It checks AI credits before browser microphone capture starts, then records, uploads the audio, transcribes it, and appends the transcript to the prompt.

Empty or no-speech recordings are treated as empty input. They should not behave like completed AI work. Browser microphone denial, missing microphone devices, unsupported recording, and transcription failure surface as prompt-composer errors.

Use voice input for rough instruction capture, then skim the transcript before submitting. A small transcription mistake in a prompt can become a large mistake in the generated object.

Watch the placeholder, then review the object

After AI Studio accepts a generation request, work continues in the background. The placeholder on the canvas is the live sign that work is in progress.

While a placeholder is processing, output actions such as edit, preview, download, duplicate, copy, lock, and thumbnail mutation stay blocked. Treat selection and deletion as the safe controls during processing.

Deleting the processing placeholder is the cancel gesture. Deleting the canvas also cancels matching running tasks. If generation has already finished or credits have already been consumed, deleting the final object does not undo the credit spend.

Longer jobs, especially file-heavy, audio, image, and web page work, continue server-side after you leave the canvas. Come back to the canvas to review the result rather than resubmitting the same expensive prompt.

Use Object Chat for follow-up work

Use AI Studio when the next step is to create or save something on the canvas. Use Object Chat when the next step is to ask about one existing object or continue working on one generated result.

That boundary keeps the canvas cleaner. AI Studio creates the first artifact. Object Chat asks, explains, rewrites, critiques, or edits the artifact after it exists. Generated and AI-edited objects can show an AI badge as their object-level AI entry point; that badge is separate from human comments.

For item-level work, read Ask about a canvas item. For review and webpage sharing, read Review AI output and share web pages.

Credits are workspace spend

AI credits belong to the workspace. The actor is recorded for audit, but the balance is checked at the workspace level before provider calls.

AI Studio generation, Object Chat messages, recommendation buttons, and voice transcription all pass through the workspace credit gate. Admins recover by buying credits in Billing. Non-admin members, external collaborators, and guests cannot buy credits from the blocked AI flow; they need a workspace admin.

Successful AI work can spend credits even if the output later turns out to be wrong, gets deleted, or needs manual cleanup. Before you submit a large image, audio, or web page request, check that the prompt and source context are worth the spend.

For billing details, read Understand AI credits and AI credit billing.

When AI Studio does not behave as expected

AI Studio is missing. Check workspace membership and edit access first. AI Studio is not available to guests, external collaborators, embedded canvases, or view/comment-only access.

The request is blocked by credits. If you are an admin, use the Billing recovery path. If you are not an admin, ask a workspace admin to buy credits or enable Auto-reload.

A file did not get used. Confirm the upload finished before submit. Remove failed upload tiles. Split oversized files, convert password-protected documents, and keep the attachment set focused.

A URL did not work. Use HTTPS links that the server can access. Private, blocked, expired, very large, or unsupported pages are not reliable source material.

The placeholder stays pending. Do not submit the same expensive prompt again. Refresh the canvas, check whether the placeholder still exists, and review credits and attachment failures. If you intend to abandon it, delete the processing placeholder.

The result is wrong or too generic. Add source context, audience, format, and constraints. Select the objects the result should use. Regenerate only when manual editing would take longer.

Support needs details. Send the workspace name, canvas link, output type, prompt summary, attachment types, approximate time, and whether credits were blocked or spent. Do not send passwords, API keys, private customer data, or unnecessary source files.