AI Image Generation in Design Agent

AI Image Generation lets Design Agent create the images your design needs, in your brand's style, and place them in the design for you. No more waiting on a visual before a campaign can go live: describe the design, and the images arrive with it.

Refine any image with a new prompt, or point at up to three images on the canvas and change them in one go. Every image from a saved design lands in your Design Agent image library, ready to reuse in other designs and variants.

Each generated image is clearly marked as AI-generated, so your team stays transparent with customers. You can restyle the visible label, or remove it once your organization takes on responsibility for disclosure.

Why marketers love AI Image Generation

No more waiting on visuals. Design Agent creates the images while it builds the design, so a popup or banner is ready to target the moment it's drafted.
On brand by default. Every image follows the palette and visual style in your Business DNA, so it fits the design around it.
Create once, reuse everywhere. Images from saved designs go to a shared image library, ready for any web or app design and variant.

Put it to work

Launch a first-order offer today

You ask Design Agent for a first-order discount popup. It writes the offer, generates a hero image in your brand's palette, and places it in the design. The popup is ready to target, with no visual request in the design queue.

The season changed, but the layout still works. You write "replace the hero image with a winter scene," and Design Agent swaps only that image. Your copy, styling, and every other image stay as they were.

Fine-tune three images in one request

Your multi-page gamification design needs a few tweaks. You leave a comment on each of three images, such as "Add a gift box," and send them together. Design Agent updates all three and tells you exactly what changed.

How it works

AI Image Generation works across popup, teaser, inline, gamification, and mobile app designs. Design Agent can generate both image elements and background images.

Every generated image is shaped by three things:

Your prompt or comment: what the image should show.
Your Business DNA: the palette and visual style, so the image fits the design around it. Keep your Business DNA up to date for the closest match.
Content rules: images of real people and copyrighted characters aren't allowed. Design Agent declines these requests and suggests describing the mood, setting, or style instead.

Design Agent generates an image only where the layout needs one, so a text-only design gets no image. If an image can't be created, the rest of the design is still delivered, and the reply lists what couldn't be done.

Generate images in a new design

Design Agent creates the images while it builds the design, so there's no separate image step.

Open Design Agent and describe the design you want in the prompt.
If you have a specific image in mind, describe it too, such as "a hero image of a gift box on a pastel background." If you don't, Design Agent decides what the image should show based on your request, or asks a follow-up question when it needs more detail.
Send the prompt. Design Agent builds the design, generates the images in your brand's style, and places each one in its image slot or as a background.
Review the design on the canvas. The reply lists the images Design Agent created and anything it couldn't do.

Note: A single prompt can create up to three images. If you ask for more, Design Agent still delivers the design and lists what it couldn't create. Ask for the rest in a new prompt.

Change images with the prompt

Write what you want to change and name the image you mean, such as "replace the hero image with a winter scene."
Send the prompt. Design Agent updates the images you mentioned.
Review the design on the canvas. The reply lists which images were updated and anything it couldn't do.

Images you don't mention stay as they are, and text or style changes leave your images untouched. A single prompt can change up to three images.

Change specific images with canvas comments

Use canvas comments to point at the exact images you want to change, then send them in one request.

Click an image on the canvas, then click its AI button.
Describe what should change, such as "Add a gift box," and confirm.
Repeat for up to three images, then click Send.
Review the design on the canvas. The reply lists which images were updated and anything it couldn't do.

Good to know about comments:

Each selected image shows a label, such as Image 1, and replies use the same labels. In multi-page designs, images are numbered per page.
A comment holds up to 500 characters. Edit or delete it from the image's numbered badge.
Click an empty image slot to open the comment box with a suggested prompt.
Send comments on their own or together with a written prompt.
Comments change what the picture shows. For layout or styling changes, such as moving, resizing, or adding a border, use the prompt instead.

Reuse generated images

When you save a design, its generated images are added to your Design Agent image library. Only images still in the design at that moment are added; images you replaced or removed before saving aren't.

Select an image on the canvas and open Edit Image.
Click Change and open the Image Library tab.
Select a generated image to place it in your design.

Shared across web and app. Reuse a generated image in any design or variant in web or app Design Agent.
The label comes back. A reused AI-generated image gets the AI Generated label again, even if you removed it in the design it came from.
Design Agent only. Generated images aren't listed in Action Builder or other Insider One products.
Unsaved designs aren't kept. Images from designs you never save aren't added to the library and are cleaned up automatically.

The AI Generated label

Every image Design Agent generates is disclosed as AI-generated. Each one gets:

A visible AI GENERATED label in the bottom-left corner, on by default.
Alt text stating that the image is AI-generated.
An invisible mark in the image file that identifies it as AI-generated. This mark can't be turned off.

Why the label is there

The EU AI Act sets transparency rules for AI-generated content, in force since 02/08/2026. They work in two layers:

Machine-readable marking: The provider of the AI system must mark every generated image so it can be detected as AI-generated. For Design Agent, this is Insider One's obligation, and the invisible mark fulfills it.
Visible disclosure: The organization that publishes the content, in this case yours, must clearly disclose AI-generated images that resemble real people, objects, places, organizations, or events and could be mistaken for authentic when shown to audiences in the EU. Clearly artistic, creative, satirical, or fictional content has lighter requirements.

Insider One adds the visible label to every generated image by default. Whether a specific design needs it is your organization's decision. That's why removing the label requires your confirmation and makes disclosure your organization's responsibility.

Note: This section is general information, not legal advice. For the full requirements and how they apply to your designs, review Article 50 of the EU AI Act with your legal team.

Adjust the label

Select the image and open Edit Image.
In the AI Generated Label section, adjust any of these settings:
- Badge Color Theme: Dark (default) or Light.
- Transparent Background: Shows the label text without its background. Off by default.
- Scale: The label's size. 1 by default.
- Position: Any of the four corners. Bottom-left by default.
- Margin: The horizontal and vertical distance from the image edge. 8 px by default.

If a value would make the label too small or push it outside the image, Design Agent adjusts it automatically so the label stays readable. You can also change these settings by prompt, such as "move the AI label to the top-right corner." Settings are kept per image, and the AI Generated Label section appears only on AI-generated images, not on images you upload.

Remove the label

Select the image and open Edit Image.
Turn off the AI-Generated Label toggle.
In the confirmation window, select "I confirm my organization is responsible for disclosing AI-generated content in this campaign."
Click Remove Label. To keep the label, click Cancel.

Edit Image only. You can't remove the label by prompt. If you ask, Design Agent points you to Edit Image.
Every removal is logged. When you save the design, Insider One records who removed the label, when, and from which image.
The invisible mark and alt text stay. Removing the visible label doesn't remove either one.
Turning it back on is instant. Turning the label on again needs no confirmation.

Example prompts and tips

Try these to get started:

Create a first-order discount popup with a hero image of a gift box on a pastel background.
Replace the hero image with a winter scene.
Move the AI label to the top-right corner.

For sharper images:

Describe the image you want. Name the subject, setting, and mood, rather than leaving it to Design Agent.
Name the image you're changing. "The hero image" or "the background" tells Design Agent exactly which one to update.
Describe moods, not people. Instead of a celebrity or a well-known character, describe the feeling, setting, or style you're after.
Keep your Business DNA current. It's what keeps every image on brand.

Good to know

Up to three images per request: by prompt or by canvas comments.
500 characters per comment.
No real people or copyrighted characters: Design Agent declines these requests.
Images only where the layout needs them: a text-only design gets no image unless you ask for one.
A failed image never blocks the design: the image slot keeps what it had, and the rest of the design is delivered.
Image request limit: Your panel has a limit on image requests. If you reach it, the reply tells you.

Troubleshooting

Why didn't my design get an image?

Design Agent adds images only where the layout needs one, so a text-only design gets none. To add an image, ask for it in the prompt. If generation failed, the placeholder stays, the rest of the design is still delivered, and the reply lists the image as not created.

I asked for several images, but only some were created. Why?

A single prompt can create up to three images, and some images can also fail to generate. The reply lists which images were created and which weren't. Ask for the rest in a new prompt.

An image couldn't be created. What happened?

The image slot keeps what it had: a new design keeps the placeholder, and an existing image stays as it was. When you change images with canvas comments, the reply also explains why and what to do next. Common causes:

The image service was busy or slow to respond.
The image was too large.
Your panel reached its image request limit.

If nothing in your request could be applied, click Try Again in the reply.

Why was my image request declined?

Images of real people and copyrighted characters aren't allowed. Retrying the same description gives the same result, so change what you ask for instead, such as describing the mood, setting, or style.

The image doesn't match my brand. What can I do?

Images follow the palette and visual style in your Business DNA, so first check that it's up to date. Then describe the style you want in the prompt, or leave a comment on the image to refine it.

I used a comment to move or resize an image, and it didn't work. Why?

Comments change what the picture shows. Ask for layout or styling changes, such as moving, resizing, or adding a border, in the prompt.

How do I remove the AI GENERATED label? Asking in the prompt doesn't work.

Removing the label needs your confirmation, so it isn't possible via prompt. Select the image, open Edit Image, and turn off AI Generated Label. The prompt can only change how the label looks.

I can't find my generated image in the image library. Where is it?

Generated images are added only when you save the design, and only if they're still in the design at that moment. Open Edit Image, click Change, and go to the Image Library tab. Generated images aren't listed in Action Builder.

Next steps

AI Image Generation puts on-brand visuals inside every Design Agent draft, with transparency built in. Ready to try it? Ask Design Agent for your next popup and describe the hero image you have in mind. To learn more about Design Agent, see Design Agent.