Generate images with AI

Describe the image you want to an AI agent in a mark chat, and it generates one in that same thread.

Visualise the idea quickly, review the result with your team, and carry the revisions straight on.

Generating an image

1. Type / in a mark chat.
2. Choose an AI agent that can generate images.
3. Describe the image you want.
4. The generated image appears in the same thread.

You can keep asking for changes in that same thread afterwards.

Getting the result you want

Including these in your description makes the result land closer:

• What you want made
• What the image is for
• The mood and style
• Colour, composition, aspect ratio
• Anything that must be included or left out

For example:

Design a shoulder-mounted water bottle for cyclists, with a hose running to the wrist. A sporty product concept that still looks stable during hard riding.

Describing both the purpose and the characteristics is enough.

Revising a generated image

Generated images can be revised in the same thread. For example:

Make the hose shorter. Switch to a side view so the structure reads clearly.

Ask for the change and it generates a new image from the previous result. Because the conversation and every generation stay in the thread, the background never has to be re-explained.

Reviewing it together

Click a generated image to open the image viewer. Zoom in on the detail, or leave a comment on the spot that needs changing.

A comment can become a to-do immediately, with an owner and a date, so the work carries on. See the guide on reviewing images and PDFs.