The AI does not send a file blindly: it sells content it must be able to describe, tease and defend in negotiation. To do that, it relies on text at two levels.
The media description
Each media in the library can carry a description (editable from the media's detail card). It helps put the content in context: what it shows, the mood, the standout details.
Generate descriptions
Instead of writing each media description by hand, you can have the AI write it: in the media library (Models → your model → Media tab), click Select, tick the media, then Actions → Generate descriptions.
Price: €0.10 per media item, deducted from your credits.
Images and videos only: audio files are not described.
A video longer than 60 seconds is described from 8 frames taken from the video.
Descriptions are written in the language chosen for the model in Spoken languages.
You can then edit each description from the media's detail card.
Media tab: the ticked media (1), Actions (2), then Generate descriptions (3).
The script step description: the source of truth
This is the most important level. When you add a media to a script step, the step description is required: position, outfit, colors, context, situation. The AI treats it as the reality of the scene:
its teasing is grounded in these concrete details (that is what sells, far more than generic teasing);
if the fan asks a question about the content ("what are you wearing?", "where are you?"), it answers with the facts from the description, never with an invention;
the facts of the scene outrank the profile's tastes: blue dress in the description, blue dress in the conversation, even if the profile's favorite color is black.
Media sent by fans
When a fan sends a photo or a video, Desirely automatically generates a description of it so the AI can react naturally. There is nothing you need to do.
In short
The more precise your descriptions, the more credible the AI and the better it sells. One minute spent describing a step properly is worth more than ten media files with no context.