The Monitoring section in AgentBuilder allows you to track and analyze your application's performance in production. Use the data analysis dashboard to monitor usage, latency, user feedback, and other metrics, helping you continuously improve your application.
Key Metrics
- Total Messages: Total number of user-AI interactions per day (excluding prompt orchestration and debugging).
- Active Users: Unique users with more than one Q&A exchange (excluding prompt orchestration and debugging).
- Avg. Session Interactions: Number of continuous interactions per session (e.g., a 10-round Q&A counts as 10). Available for conversational apps only.
- Token Output Speed: Tokens output per second, reflecting model generation rate and app usage frequency.
- User Satisfaction Rate: Number of likes per 1,000 messages, indicating user satisfaction.
- Token Usage: Daily token expenditure for language model requests, useful for cost control.