
1pnpm dev
2# or
3npm run devAnnotate your agent's traces with your feedback. Highlight what works well and what needs improvement. Your insights become the foundation for making your agent better.
Pi automatically converts your feedback into evaluation judges. These judges continuously measure your agent's performance, giving you clear metrics on what's working and what's not.
Your feedback becomes actionable improvements. Pi uses your insights to generate online controls and refinements that make your agent smarter and more reliable over time.