Datasets are reusable test inputs for Evals. Import production requests or add items manually.
NameItemsCreated
Support golden set
30 previously-failed customer support cases + 20 typical interactions
50
5/2/2026
Extraction edge cases
Malformed inputs that previously broke JSON extraction
24
5/10/2026
Email triage smoke test
Quick sanity check before deploying classifier changes
15
5/14/2026