How Threadify works - Threadify Documentation

Concept	What it is
Thread	One run of a process, such as one order or one agent session
Step	Something that happened in that run, with its outcome and data
Contract	The rules a run must follow
Execution wait	Asking for permission before a step, or waiting for its check after
Decisions	Asking what the contract allows next, without claiming anything
Reference	A business ID, such as `customer_id`, attached to a thread
Entity profile	History and metrics for one customer, agent or partner across many threads
Notification	An event your code can react to

Recording is not the same as passing

Threadify keeps three facts about every step separate, and it helps to keep them separate in your head too:

Was it recorded? Reporting a step returns an acknowledgement as soon as Threadify has stored it.
Did it succeed? That’s the outcome you reported: success, failed or error.
Did it follow the contract? Threadify checks this separately, usually a moment after recording.

A failed payment can follow the contract perfectly, and a successful one can break it, for example if it ran without approval. An acknowledgement doesn’t tell you the check passed. If your code needs the result, ask for it with waitFor: true.

Threadify doesn’t run your code

Your services and agents do the work. Threadify records it and tells you whether it was allowed. If a step must not happen without permission, your code has to ask first and stop if the answer is no. Recording a step after the fact can flag a problem but can’t undo it.