Classification labels
The categories the model chooses between. Classification is zero-shot, so anything you add here works on the next complaint — no retraining, no restart.
Add a label
The name is what the model matches against, so keep it a short, plain category. The description is a keyword hint layered on top — list the concrete phrases your complainants actually use and it will break ties the model gets wrong.
Configured labels 8
Edit inline, press ↵ or Save. Toggle off to exclude without deleting.| Name | Description / example phrases | Active | |
|---|---|---|---|