Automations

Test and debug

Try automations safely before anyone depends on them, and see exactly what every run did.

#Test runs

Test run (on the automation's page) runs the draft of the automation on the app's test data (the records you add in the builder's preview), on its newest record.

• Emails, messages, posts, payments and calls are not sent: the run log shows exactly what would have been sent.
• AI steps don't call a model.
• Record steps change the test data only.

Run now runs the published version for real.

#Replay a real run

Replay as a test (on a real run's page) runs that run again as a test: the draft automation, the same input, and a copy of its record (as it is now) in the test data. Nothing is sent and your real data isn't touched; the copy stays in the test data so you can see what the run changed. Use it to fix an automation that failed on real data.

#Saved tests

An automation that starts from a table can have tests: given a test record with some values (Amount: 5000), then after the automation runs on it, these checks must be yes ([Status] = "Needs approval").

• Run all tests runs them on test data (nothing is sent; the test records are deleted afterwards).
• Publishing runs every test first and warns you about failures; it doesn't stop the publish.

#Run log

Every run (real and test) is listed with its status:

Open a run to see each step's input and result, and the error of a failed step.

#Retries

A step whose problem may pass (a timeout, a service that was down) is retried a few times, waiting longer each time. A problem with the settings (a missing field, a bad address) fails at once. Retry a failed run from its page: steps that already succeeded aren't done again, so nobody gets the same email twice.

#Who sees run logs, and for how long

The app's owners and editors see its run logs. Logs are kept for 7 days on the Free plan, 30 days on Builder and 90 days on Pro.