Guide 09 · Developer docs

Automation, jobs and integration

A job is a trigger, conditions and an ordered list of actions, set up under Customization › Orchestration in the admin; jobs were called hooks before the module was renamed, and the API's routes keep that name. What starts a job, what it can do, branches and loops, email, scheduled work, integrations with other systems, and every event and action.

L2–L3

Unit 04 of 08 4 pages

Branches, loops and re-runs

Updated 23/10/2026 01:50

Conditions and loops turn a job from a trigger into a program. Branching that matches when you mean it to, loops that one bad row cannot take down, jobs that can run twice without doing the work twice, and where volume belongs.

Conditions: if, else if and else covers branching, including the two ways a branch matches without checking anything. Loops and failing iterations covers where a loop's list comes from, how it runs and waits, its time limit, and the two correct answers to a failing row: contain it, or abort.

Idempotency and replay is the page in this unit not to skip: every job that writes eventually runs twice, and the external-id pattern is what makes the second run harmless. Event job or scheduled job closes the unit with the decision that keeps event jobs quick.

In this unit

Read in order

01 Conditions: if, else if and else A condition runs the first of its branches that matches, or its else. The two branch types, the two ways a branch matches without checking anything, branching on a value, a comparison and the presence of a record, and how to keep a branching job readable. 02 Loops and failing iterations A loop runs its actions once per element of a list. Where the list comes from, what each iteration can read, the settings that decide how it runs and waits, the time limit, and what happens when one iteration fails: contain it and carry on, or abort. 03 Idempotency and replay Every job that writes eventually runs twice: a re-run after a fix, a partner sending the same call again, a retried loop element, an overlapping schedule. The external-id pattern that makes the second run change nothing, as job actions, and the test that proves it. 04 Event job or scheduled job An event job responds to one record or one request and should finish in seconds; a scheduled job works through volume and must be safe to resume. How to tell which you are building, and the symptom of choosing wrong: a job that works in testing and times out on real data.