Runs and held runs

A trigger requests an automation run. The run works through the steps once it starts; run limits can hold it before then. This page covers what a run’s status means, how to stop or retry a run, and what to do when runs are held before they start.

Who can use it

Owners and admins on a plan with automations (see Automations) see an automation’s runs, start held runs and retry failed runs. Who can open a single run depends on who it ran as; see Runs as.

How a run works

• One trigger requests one run. Every trigger on an automation uses the same steps.
• A run uses the version of the automation that was live when it started, even if you publish an edit while it’s going.
• Steps run one after another, and later steps see what earlier ones produced.
• A run goes to the end without stopping for input. You can’t approve or redirect it partway; a person reviews its drafts afterward in For review.
• If a run is interrupted, for example by a server restart, it resumes from where it left off. That doesn’t count as a failure, and finished steps don’t run twice.
• Anything a step tries outside its permissions is blocked and recorded on the run.

Open a run from the automation’s Runs tab to see its status, duration, trigger, who it ran as, the credits it used, and what each step did. You can ask the agent what went wrong on a run and have it propose a fix.

Check results after launch

Check the Runs tab after the first few triggers, then regularly while the automation is active. Inspect completed runs as well as failures: a completed run can still produce the wrong result. Read the step’s inputs and output, explain the problem in the automation’s chat, test the change, and click Publish changes.

Before starting a group of held runs, check one trigger record. Starting a large group can create many tasks or drafts. Use Usage to check spending before releasing it.

Run statuses

A run’s status shows on the Runs tab and at the top of the run page.

Status	What it means
In progress	The run is working through its steps.
Completed	Every step finished.
Failed	A step failed, so later steps didn’t run.
Skipped	A step found nothing to do for this trigger and ended the run early.
Not started	The run hasn’t started, usually because it’s held. See Held runs.
Dismissed	A held run was dismissed or expired, so it never started.
Cancelled	The run was stopped before it finished.

Stop a run

Stopping can’t be undone. Steps that haven’t started are skipped.

1. Open the run from the automation’s Runs tab.
2. Click Stop run.

The run shows Cancelled.

Retry a failed run

There’s no retry button. The agent replays the run for you as a new run that picks up from the failed step.

1. Open the failed run from the automation’s Runs tab.
2. On the failed step, click Review with agent. A chat opens titled Review run #number, with the step and its error filled in. You can also ask in any chat, for example “Replay run #42 of the renewal reminder automation.”
3. The agent checks what the failed run already did, such as whether an email went out, and tells you whether a retry is safe. If the automation caused the failure, it fixes the automation first.
4. Confirm. The agent starts a new run at the failed step, or at an earlier step if that step’s output was the real problem. Finished steps before it are carried over, not run again.

When it worked, the new run shows Re-run of Run #number, and the failed run shows Marked resolved by Run #number with a View re-run link.

The agent can’t replay a run that didn’t fail, one already marked resolved, a run of a paused or deleted automation, or a sequence run. While your workspace is at its daily automation limit, replays wait until the limit resets at midnight in your workspace’s time zone.

Close out a failed run without a retry

If you fixed the cause and later runs succeed, open the failed run’s options menu and choose Mark as resolved. You can also ask the agent to mark it resolved.

Held runs

A held run was triggered but hasn’t started. It shows Not started until someone starts or dismisses it. The label on each group explains why it is held:

• Surge protection. One source, such as a webhook or record changes from a CSV import, triggered the automation more times in 10 minutes than its run limit. Surge protection holds only webhook and record-change triggers.
• Daily limit. Your workspace reached its daily automation limit. Runs from any trigger wait until the limit resets at midnight in your workspace’s time zone. Test runs are never held by this limit.

The interface may also show Daily spend limit. Hover over the group to read Why these runs are held, and check Usage before starting more runs.

Start or dismiss held runs

1. Open the automation’s page and find Held runs. Runs are grouped by why they’re held (Surge protection, Daily spend limit or Daily limit) and by what triggered them, with a count such as 12 runs held.
2. Hover over a group to see Why these runs are held, the source, and when the group expires.
3. Click Run all to start the group’s runs, or Dismiss all to drop them.

Runs start in batches as capacity frees up. Dismissed runs show Dismissed and never start. The agent can tell you which runs are held and why, but only you can start or dismiss them.

• A group still filling up shows Collecting runs… until the 10-minute trigger count falls back under the limit. To act on it sooner, open Rate limit options and choose Reset rate limit.
• Run all is blocked while the automation is paused (resume it first) and while the workspace is at its daily limit (run them after the reset).
• A group nobody acts on expires 14 days after it stops collecting, and its runs are dismissed.

Turn off surge protection

Surge protection is on by default. To turn it off for one automation, switch off Run protection on the automation’s page. Runs past the limit then start right away instead of being held. Keep it on for automations fed by imports or webhooks, since one bulk change can otherwise start hundreds of runs.

Troubleshooting

My automation didn’t run when I updated a record

Check Status first: a Paused or Draft automation does not run on triggers. If it is Active, compare the fields under Triggers with the field you changed. A record-updated trigger fires only when a watched field changes. Ask the agent to add the field if the trigger should watch it. If the settings match, inspect the run history or ask the agent to investigate.

No run appears after another process changes a record

First establish whether the automation has a record-created or record-updated trigger. Ask the agent to inspect the trigger, watched fields, and the source of the change, such as an import, API call, automation or workflow. An absent run is different from a failed run. Do not assume a rule for record-created triggers also applies to updates or to built-in workflows.

If the cause is still unclear, give the agent the automation, record and change time so it can narrow the investigation. Review matching and duplicate prevention before adding a scheduled fallback.

My runs show “Not started”

Open Held runs and read Why these runs are held. The group label distinguishes surge protection from a daily limit. See Start or dismiss held runs for the actions available.

A run failed

Open the run, read the error on the failed step, and click Review with agent. The agent can inspect the error, propose a fix and, when the run is eligible, replay it. See Retry a failed run.

Common questions

Can I rerun a failed automation run from where it stopped? Yes. Ask the agent to replay it. It starts a new run at the failed step, linked to the original.

Why are my automation runs held? The automation was triggered more times in 10 minutes than its run limit, or your workspace reached its daily automation limit. Start them with Run all or drop them with Dismiss all under Held runs.

What happens if a run is interrupted? It resumes where it left off. An interruption doesn’t count as a failure or start the run over.

Does editing an automation change runs already in progress? No. Each run keeps the version it started with.