Generate images, not just text
Product mockups in context
Let users upload a photo—their bedroom, their living room—and the assistant generates a mockup of your products composited right into the scene. It edits the real photo, preserving the room's layout and lighting, so shoppers see exactly how your product looks in their space before they buy.
What you get
- Users upload a photo; the assistant edits it into a product mockup
- Preserves the room's real layout, lighting, and perspective
- Operator picks the quality tier in the portal—the model never controls cost
- Generated images render inline and persist in conversation history
- Built on file attachments—images, PDFs, and text, toggleable per application
- Billed per image with a per-plan allowance—no surprise costs
How it works
Enable it per application
Turn on image generation in the application's Behavior settings in the portal and pick a quality tier (Low, Medium, or High—High by default); while it's off, the assistant has no access to the image tool at all.
The user uploads a photo and asks
Through the widget's normal attachment flow the user uploads a photo of their space and asks for a mockup—e.g. "show me your charcoal quilt on this bed"—using one of your real products.
The assistant edits and returns the mockup
The model edits the uploaded photo, compositing your product in while keeping the room's layout and lighting, then renders the generated image inline so it persists in conversation history.