Eval-driven development

1. 01Collect real inputs before writing the first prompt☐
2. 02Label 50 examples with the answer a good employee would give☐
3. 03Score every change against the same set☐
4. 04Track failures by category, not only a single score☐
5. 05Set the pass bar with the customer before the pilot☐
6. 06Keep the eval set in version control☐

Continue reading