Every row. A little less work.
Bring your CSV. Add a column for each question: yes or no, a category, a score, or tags. Watch answers land, then export your original data with them. Connect a HolySheet key or your own TypeSafe API key above.
Enter your pilot key to see your monthly allowance.
Results appear as each chunk finishes. Keep this tab open until the run completes. Model cost estimates use returned token counts and exclude failed attempts, gateway and infrastructure charges.
Each call uses Jev’s native parallel question fan-out. Duplicate and blank rows are skipped.
Run telemetry
- Route
- —
- Model
- —
- Batch plan
- —
- Execution
- —
- Provider limit floor
- —
- Summed batch time
- —
| Batch | Rows | Time | Input tokens | Route |
|---|---|---|---|---|
| No batch data yet. | ||||
Results are model estimates, not facts about a person. Review uncertain rows before acting. Runs handle up to 50,000 rows while this tab stays open. Category and score confidence comes from the model’s answer distribution, not measured accuracy.