Sandbox demo
Combining the Basics
Watch a vague classifier prompt become production-grade: label space, output contract, few-shot edge cases, and a hostile-input test.
Sandbox transcript
Press Play to replay this transcript step-by-step.
0 / 14
The transcript is canned — no real API calls. The point is to internalize the loop: user → plan → tool → result → answer.
Check your understanding
Q1. In the transcript, refund emails bounced between billing and churn_risk. What actually fixed it?
Q2. The sarcastic email ('Wonderful. Third invoice...') was mislabeled how_to. Why did the assistant reach for examples instead of another rule?
· Score 100% on the quiz.