Engine Performance Monitoring
Monitoring your engine performance ensures your AI workflows, scheduled tasks, and automated processes run efficiently. GoSiteMe provides a comprehensive monitoring dashboard with real-time metrics, historical trends, and alerting capabilities.
Accessing the Monitoring Dashboard
- Go to Dashboard → Engines → Monitoring.
- The overview shows all active engines with status indicators (green/yellow/red).
- Click any engine to drill into its specific metrics.
Key Metrics
| Metric | Description | Healthy Range |
|---|---|---|
| Response Time | Average time to process a request | < 500ms for RAG, < 2s for PLAYBOOK |
| Throughput | Requests processed per minute | Varies by plan |
| Error Rate | Percentage of failed requests | < 1% |
| Queue Depth | Number of pending requests | < 50 (sustained) |
| Token Usage | AI tokens consumed per engine | Within plan allocation |
| Memory Usage | RAM consumed by the engine process | < 80% of allocation |
Real-Time Graphs
The monitoring dashboard displays live graphs for each metric with adjustable time ranges (last hour, 24 hours, 7 days, 30 days). Hover over any data point to see the exact value and timestamp. Use the comparison feature to overlay metrics from different engines or time periods.
Setting Up Alerts
- Navigate to Engines → Monitoring → Alerts.
- Click Create Alert.
- Select the engine, metric, and threshold (e.g., "Error Rate > 5% for 5 minutes").
- Choose notification channels: email, Slack, Discord, webhook, or SMS.
- Set the severity level: Info, Warning, or Critical.
- Click Save.
Log Analysis
Engine logs provide detailed request-level data:
- Filter logs by engine, time range, status code, or keyword.
- View individual request traces showing each processing step and its duration.
- Export logs as CSV or JSON for external analysis tools.
- Use the search bar to find specific error messages or request IDs.
Optimization Recommendations
The monitoring dashboard includes an Optimize tab that analyzes your engine usage patterns and suggests improvements:
- Identify underutilized engines that can be downscaled.
- Detect frequently repeated identical requests that benefit from caching.
- Highlight slow workflow steps that could be parallelized.
- Recommend plan upgrades when you are consistently hitting resource limits.
Using Alfred for Monitoring
Ask Alfred for quick insights without opening the dashboard:
- "Alfred, what is the current error rate for the RAG engine?"
- "Alfred, show me engine performance for the last 24 hours."
- "Alfred, why did the PLAYBOOK engine slow down yesterday?"