Scheduled work
For recurring or one-time work, use a schedule trigger inside an automation. The schedule sets the time; the automation carries everything else.
Configure the schedule
Give the schedule a name that identifies its purpose and owner. Choose:
- A recurring cadence or a one-time future run.
- Fresh-run or persistent-monitor behavior.
- The operating package that should apply every time.
- The repositories, documents, data, and systems in scope.
- The frequency, local time, and timezone.
- Email or Slack notifications.
- The user or organization identity that owns each run.
- The prompt, completion checks, exception owner, and approval gates.
Recurring schedules run hourly, daily, weekly, or on a custom cadence. A one-time schedule disables itself after execution and keeps its run history.
Manage the schedule
A schedule can be active, paused, or in an error state. Pausing stops new invocations and keeps the history. Deleting removes future execution, while completed sessions remain available.
The schedule detail links to past sessions. Compare outcomes, failures, evidence, and reviewer corrections there before changing the recurring procedure, so the change is informed by what actually went wrong.
Good scheduled work
- Daily pipeline, cash, operations, or incident summaries.
- Weekly competitor, account, policy, or market research.
- Monthly reporting and reconciliation packs.
- Periodic Knowledge cleanup and Skill review.
- Recurring data quality, dependency, or security checks.
- Nightly browser, application, and end-to-end tests.
Define automations and schedules only after a manual run has a dependable completion test.