Choose a workflow trigger
Decide whether a workflow starts manually, on a schedule, or from an event.
A trigger determines when a saved workflow starts. Choose it according to who needs the result, how quickly new input must be processed, and whether the steps can run unattended.
Compare trigger types
| Trigger | Use it for | Key consideration |
|---|---|---|
| Manual | Development, one-off processing, and interactive operations | A person explicitly starts the run. |
| Schedule | Repeated processing at known intervals | Source availability and timezone must match the schedule. |
| Event | Processing that follows a supported data event | Scope the upstream event and avoid feedback loops. |
| On file | Processing newly added files | Limit the store and file scope to the intended inputs. |
Set the trigger in the editor toolbar, complete its additional fields, and save. A trigger change affects future admissions; it does not rewrite the history of earlier runs.
Check handler compatibility
Not every handler supports every trigger. A job that requires an interactive decision can restrict the workflow to manual execution. Validation checks the complete graph against the selected trigger. Changing only the trigger cannot make an interactive job safe to run unattended.
App-managed recipes may protect the trigger alongside their job structure. Use the app's intended configuration controls for those workflows.
Introduce automation gradually
First run a small manual example. Verify the destination, record identity, and failure behavior. Then enable the automatic trigger and observe its first execution. An interval shorter than processing time can create overlapping or queued work; plan for that before choosing a frequent schedule.
Disabling a workflow is a management action. Cancelling an existing run is a separate execution action. If you need to stop work now, inspect the run and use cancellation rather than relying only on disabling the workflow.
Avoid a catch-all event workflow that writes the same kind of event-producing data it consumes without an explicit boundary. Read file and record events, scheduled triggers, and lifecycle management.