Monitoring is the dashboard that tracks a workflow's throughput, speed, and stability in one place. It surfaces the node that's becoming a bottleneck and lets you drill into the cause job by job (a job is the processing record of one run).
Dashboard layout
| Area | Contents |
|---|---|
| Filters | Workflow / Config / Source (Studio Web · API · Automation) / Time range (1 hour · 24 hours · 7 days · 30 days · Custom) |
| Key metrics | Total jobs · Latency (processing time) · Stability (success rate) |
| Per-node metrics | Throughput, time, and stability for each of Parse · Classify · Extract · Instruct. This is how you identify the bottleneck |
| Job list | Individual run history: job ID · config version · file · duration · status · source · created |
Working with the job list
- Click a job to open Job details and see what happened at each stage.
- Download Excel: export the results of selected jobs to Excel for reporting or reconciliation.
- Failure reason: failed jobs show the failure reason along with an explanation of the error code, so you can pinpoint the cause right away.
- Import to workflow: pull the documents of an API-processed job onto the workflow screen for a closer review.
- Statuses: Completed · Failed · In progress · Queued · Cancelling · Cancelled
Where to open it
- Monitoring in the account menu at the bottom of the sidebar (all workflows)
- Monitor in the view switcher at the top of the case panel on the workflow screen (this workflow only)
- The monitoring icon on each row of the workflow list (per-workflow entry)
A suggested routine: check the Stability (success rate) and Latency trends once a week, and use Import to workflow on failed jobs to find out why. Right after a deployment, filter by the new config version and compare it against the previous one.
Key Summary
"Read metrics as trends, chase problems job by job. Spot the anomaly on the dashboard, then narrow the cause in the job details."