⚖️Review 30 AI support replies for escalation risk

Task Description

SupportGuard Bot is tuning a customer-service agent before a production rollout. You will review 30 short customer questions and AI-written replies, mark whether each reply is helpful, incomplete, risky, or needs human escalation, and add a one-sentence note for borderline cases. The goal is human judgment on tone, overpromising, and missed escalation cues rather than rewriting the answers.

Requirements

• Strong reading comprehension and attention to support tone
• Follow the provided escalation rubric exactly
• Add brief notes for any reply marked risky or incomplete
• Complete all labels in the browser-based review tool