Auto Mode

Auto Mode reads your request, selected target, project context, and references, then chooses the smallest useful image or video job.

What Auto Mode does

Auto Mode is the planning layer. It decides whether to create new images, refine existing versions, make more like one version, remix references, or create video.

• New directions: creates visual options from a prompt.
• More like this: makes more options from a selected image.
• Change this: edits one version or applies one instruction to all versions.
• Motion requests: routes explicit motion prompts to the right video mode.

Example plan:

• User request: "More like Version A, but calmer and more editorial."
• Target: Version A.
• Intent: variation.
• Planned job: Gemini image, 3 directions, selected image plus project instructions.

Use selected targets

Select a result before you type when the instruction is about a specific version. Auto Mode uses that target to keep the next job scoped.

1. Select a version.
2. Choose the intent: More like, Change, or All versions when the buttons fit.
3. Write the next move.
4. Review the new checkpoint.

Example targeted change: "Make Version B feel more premium and less chaotic. Keep the bright morning palette and the iced coffee can in frame."

Good requests

• "Make three poster directions for this launch."
• "More like Version A, but calmer and more editorial."
• "Animate this hero image as a slow product reveal."
• "Use the project references as style guidance for a new campaign direction."