Use cases
Tagging
Every tag that applies, zero or more per input.
Topics, compliance flags and product attributes rarely come one at a time. A model trained on your tagged examples learns which combinations go together in your data.
The problem
Missed tags are silent
When a model leaves a tag out, nothing errors. You only notice later, in reports or audits.
Tag sets are yours
Your taxonomy is specific to your business, and long prompts listing every tag are slow and costly.
Volume adds up
Tagging usually runs on every item in a catalog or every message in a stream.
Where it applies
- Topic and theme tags
- Compliance and risk flags
- Product attributes
- Feedback and review themes
- Document metadata
How the switch is checked
Before you switch, the fine-tuned model and the LLM you use today answer the same held-out examples from your data.
- Tag F1 per example, for both models
- Exact tag-set matches
- Recall for every tag, so rarely-used tags are not lost
What you need
Each example's text, and its tags as a comma-separated list or a JSON array.