Create with the AI chat assistant

Not everyone wants to fill in six prompt sections by hand. The AI chat assistant lets you describe an image in plain language and proposes a complete generation you can review before it runs. This guide covers how the assistant turns a request into a preview card and how to steer it toward what you actually pictured.

Describe what you want

Tell the assistant what you are after the way you would tell a collaborator: the character, the vibe, the setting, and the kind of shot. You do not need prompt syntax. The assistant reads your request, picks a suitable model, and assembles a prompt and settings on your behalf.

Review the proposed generation

Rather than firing off an image immediately, the assistant proposes a setup in a preview card with a short summary of what it will make, the model it chose, and an estimated credit cost. Nothing is spent until you approve. This keeps you in control: you see the plan first, then decide whether to run it.

Steer it toward the shot you want

1. Describe the image in plain language, including the character if you have one.
2. Read the proposed setup in the preview card and check the estimated cost.
3. Ask for changes in chat - a different model, a new wardrobe, a tighter crop.
4. Run the generation once the proposal matches what you pictured.
5. Iterate by replying with the next adjustment.

The assistant is a faster on-ramp than the manual composer, but the same caveats apply: results can still vary, so treat the first proposal as a draft to refine. When you want fine control over framing or lighting, switch to the section composer and write those fields yourself.