Run history is the record of every time an automation has executed, so you can confirm it is doing the work you built it for and find the run behind a result you did not expect.
Where do you find an automation’s run history?
Speak AI puts run history one click from the automations list. Open Automations in the sidebar, then either select the run count in the automation’s row, or open the row’s menu and choose Run history.
The page opens at Run history for that automation, subtitled Every execution of this automation, what fired it, and what each step produced.
What do the tiles at the top tell you?
Five tiles summarize the window you are looking at, defaulting to the last 30 days, in this order.
- Runs: how many times the automation executed in the window.
- Success rate: the share of those runs that completed.
- Stopped: runs that ended early because a condition was not met.
- Failed: runs where a step could not finish.
- Median time: how long a completed run typically takes.
What does each column in the run list mean?
The list carries one row per run, with eight columns.
- Run: the run’s own identifier.
- Outcome: how the run ended.
- Fired by: which trigger started it.
- Subject: the recording or item the run worked on.
- Steps: how far through the automation it got.
- Produced: what the run wrote, such as fields or recordings.
- Duration: how long it took.
- Started: when the run began.
What do the outcomes mean?
An automation run ends in one of five states, and only one of them is a failure.
- Completed: every step ran to the end.
- Stopped: a condition was not met, so the run ended early. This is a working automation doing its job, not an error. A filter that stops a run is the filter succeeding.
- Failed: a step could not finish.
- Running: the run is still in progress.
- Pending: the run is queued and has not started yet.
The distinction between Stopped and Failed is the one worth learning. A sales-call scorer filtered to one folder will stop on everything outside it, and that is correct.
What can you filter the list by?
Four controls narrow the list, and you can combine them.
- Filter by outcome: show only completed, stopped, failed, or running runs. Pending is a state a run can be in, but it is not one of the choices here.
- Filter by what fired the run: pick a single trigger source, or leave it on Any trigger.
- Filter by time window: look back over the last 7, 30 or 90 days. It starts on 30 days, and it is the control that sets the window the tiles above summarize.
- Search runs by ID: paste a run or recording ID to jump straight to it.
The trigger sources you can filter by are Recording processed, Webhook or API, Re-analyzed, Scheduled, Translation, Test run, Field updated, AI chat completed, Recording received, and Clip created.
What does an individual run show you?
Selecting a run opens Run details, a step-by-step timeline of what the automation did.
Each step names what it was, such as AI analysis, Only continue if, Send notification, Call webhook, Translate, Upload to Speak, or Connected app action, and what it produced. A condition step reports either Conditions met, continued or Conditions not met, stopped here, so you can see exactly where a stopped run ended.
Outputs appear as references rather than full content. Open the recording itself to read what a step wrote.
A step can also report that it is waiting, with Waiting for the recording to finish processing or Waiting for audio extraction to finish. A run in that state has not stalled. It resumes on its own once the media is ready.
How long does Speak AI keep run history?
Speak AI keeps automation runs for 90 days, then removes them automatically.
That retention applies to the stats window too, so you can look back at most 90 days. If a run you are looking for does not load, it has most likely passed the 90 day limit. Export anything you need to keep for longer before it ages out.
Related guides
Start a run with the right trigger, narrow which media qualifies with filters, and choose what happens with actions.
Want a hand reading a run that did not do what you expected? Book a free consult.
Related: Automations · Triggers · Actions · Filters