Putting the job on a schedule
The step where it stops being your job: it runs because it is the first of the month, not because you remembered.
A scheduled task is the same four things you already have. A name, the prompt, the approval mode, and how often it runs. The cadences are hourly, daily, weekly, on weekdays, or manually when you want to run it on demand.
Where a scheduled task runs
Remotely, which is the useful part: it runs on its cadence even when your computer is asleep or the desktop app is closed. The exception is a task that needs local files or local apps, which can only run locally, so your machine has to be on for that one.
That is the difference that decides how you set your job up. A job on files in a connected cloud store can run at 6am on the first of the month while you sleep. The same job on a folder on your laptop cannot.
Managing them
| You can | From |
|---|---|
| See upcoming and past runs | Scheduled, in the sidebar |
| Pause one, or resume it | The same place |
| Run one now, without waiting | The same place |
| Delete it | The same place |
Reading past runs is the part people skip. A scheduled task that has quietly produced the wrong thing for three months is worse than one that failed loudly on day one.
What to schedule, and what not to
- Good: a report nobody reads until Monday. If it is late or odd, nothing breaks.
- Good: a monthly clean-up whose output you check before using.
- Careful: anything that writes where other people are working.
- No: anything that sends, publishes or pays without a person in the loop.
- Schedule your job monthly, then run it on demand once to see it work end to end.
- Put a reminder in your own calendar to read the past runs after two months.
Slow is fine. Stopping is the only problem.