Scheduled Workflows (Closed Beta) and Workflow Run History
Workflows can now run on their own. Set a schedule once and your workflow runs automatically, with every run recorded in its history.
What's new
Until now, a workflow only ran when someone opened it and clicked Run. Anything that needed refreshing on a cadence, like a weekly trends analysis, depended on someone remembering. Now the workflow runs on the schedule you set, and run history shows what each run produced.
Schedule a workflow
Choose Schedule from a workflow's menu in the Workflows list, or set it up from inside the workflow. Pick daily, weekly, or monthly, or set up a custom repeat the way you would for a calendar event, choosing how often it runs and when it stops. Then pick a time and a timezone. A workflow can have several named schedules, such as a daily refresh and a monthly roundup. Each run uses the workflow as it is at that moment, so any edits apply from the next run onward.
Manage every schedule in one place
A new Scheduled section on the Workflows page lists every scheduled workflow with its cadence, status, and next run. You can search, sort, and filter by status, cadence, creator, or who scheduled it. Scheduled workflows also show a Scheduled label in your other workflow lists. You can edit, pause, resume, or delete a schedule at any time.
Workflow run history
Every run of a workflow, scheduled or manual, now appears in the run history panel in the editor. Each run shows when it ran, how many nodes it covered, and how many credits it used. Open any run to see the workflow exactly as it was for that run, outputs included, even if it has changed since. You can also rename runs and copy a link to one.
A few things to know
- Scheduling is in closed beta and is switched on per workspace on request. Workflow run history is live for everyone.
- Only people who can edit a workflow can schedule it or change its schedules.
- Scheduled runs use credits just like manual runs.
- Schedules are time-based only for now, and there are no notifications for failed runs yet, so check run history to see how each run went.