Understanding Rule Execution Modes: Sequential vs. Parallel

Auto-Apply and Human Review

You can configure the auto-apply setting directly on enrichment and brand compliance agents:

• Auto-apply on: We apply results automatically, and the rule proceeds immediately to the next action.

• Auto-apply off: The rule pauses at that step and waits for a user to approve or discard the result in the Agents activity overview. The rule continues to the next step in both cases once the user submits their choice.

• Review Indicator: A human reviewer icon appears on actions where auto-apply is off. The tooltip reads "Requires review to continue".

Parallel Mode

Parallel execution runs all actions simultaneously. Each action operates independently, meaning a technical failure in one action does not affect or halt the others.

• We group all actions together within a single step on the canvas.

• A rule can contain a maximum of 5 actions inside that single step.

Multi-Asset Triggers

When a trigger matches multiple assets, such as a relative date trigger processing all assets matching your specific criteria, each asset receives its own independent path through the rule. An asset waiting for review or experiencing a failure does not block or delay other assets. This behavior applies to both sequential and parallel modes.

Technical Considerations

• Step and Action Limits: Rules are limited to a maximum of 5 steps in sequential mode (one action per step) or 5 actions in parallel mode (all inside one step).

• Condition Evaluation: We evaluate conditions exactly once at the start of the rule, before any actions run. Sequential mode does not support adding conditions between actions.

• No Retry or Resume: Sequential mode does not support retrying or resuming from a technical failure. If an action fails, remaining actions for that asset do not run.

• Visibility: The rule builder does not display active execution history or individual action statuses.

• Review Timeouts: If a sequential rule is waiting for human review and no user takes action within one year, the execution times out, and the remaining actions for that asset do not run.

• In-Flight Edits: You can edit a rule while an execution is in progress, but active executions are not guaranteed to apply your new changes.

• Loop Prevention: Transformation agent output does not fire the asset published trigger, which prevents infinite loops when you save a transformation result.

How to Configure the Execution Mode

Use the Details panel in the rule builder:

1. Open a rule in the rule builder canvas.

2. Open the Details panel by selecting Edit details from the rule's contextual menu, or by clicking an empty area of the canvas.

3. Under Rule flow, select either Sequential or Parallel.

The canvas updates immediately to reflect your choice. Sequential actions display with spacing between them and feature Then or And then labels. Parallel actions are grouped together with a single Then label between the trigger and the action block.

Why?

Choosing the correct execution mode ensures your automated workflows run efficiently based on your operational goals:

• Choose Sequential Mode when your automated actions depend on data from a previous step. For example, use this mode when one agent must generate metaproperty values that a subsequent agent needs to read.

• Choose Parallel Mode when your actions are completely independent of one another. For example, use this mode to run a metadata update and add an asset to a collection at the same time.