Loop monitoring
Monitor predict and train traffic for each deployed loop. Usage tracking starts after you generate an API key and call Loop endpoints from the UI or your application.
Per-loop usage (recommended)
- Open Loop and click your loop.
- Generate API key if needed (deployment guide).
- Scroll to the Usage section on the loop detail page.
You will see:
- Summary cards — total inferences, requests, average response time, endpoints used
- Charts — daily inference volume and response time, split by Heimdall UI vs API
- Request log — sortable table with endpoint (predict, train, etc.), inference count, response time, channel, and user agent
- Filters — 7 / 30 / 90 day windows and endpoint filter
Use this to verify your continuous learning loop is receiving both predictions and training feedback from production.
Account-level monitoring
Open Usage (/usage) → Data Intelligence for workspace-wide Loop traffic alongside ML, Forecast, Read, and Vision.
See Production monitoring for daily charts, channel mix, and top assets.
What gets recorded
| Field | Meaning |
|---|---|
| Endpoint | REST path called (predict, train, definitions, etc.) |
| Inference count | Predictions or training examples processed |
| Response time | Milliseconds to complete |
| Channel | Heimdall UI or API |
| User agent | Client string when present |
Next steps
- API integration guide — predict and train from your app
- Production monitoring — account usage and platform health
- Loop modeling — define schemas and bootstrap training data