Reporting a run's outcome with the On finish lane

In CloudRadial AutomationAI, a workflow can report its own outcome after its main work is done, for example by posting a note back to the ticket that started the run. You build that report in the On finish lane, a small second flow on the Designer canvas that runs once after the main path ends. This article is for workspace users with the Admin role or higher who build workflows in the Designer.

• How the On Finish Lane Works
• Adding the On Finish Node
• Reading the Run's Outcome
• The AI-Written Summary
• Example: Posting a Note to the Ticket That Started the Run
• Lane Rules and Limits
• What You See in History

How the On Finish Lane Works

Every workflow has a main path: the nodes reachable from the Start node. When a step on the main path fails or times out, that branch stops there and the run's outcome is recorded as Failed or Timed out. The outcome is decided by the first step that failed or timed out, so a later success elsewhere in the workflow never hides a failure. If no step fails, the outcome is Succeeded.

The On finish lane runs once, after the main path ends:

• The On finish node has three outputs: succeeded, failed, and timed out. The lane runs from the output that matches the main path's outcome.
• An output with nothing connected to it does nothing. If you only connect failed, a successful run ends without running the lane.
• A run that someone cancels during the main path runs no lane.
• While the lane runs, the run shows the Finishing status. It changes to its final status when the lane ends, whether or not the lane's steps succeeded. To confirm the report was made, check the run's On finish result.
• The lane never changes the run's status. If the main path succeeded and the lane's ticket note fails, the run still ends Succeeded, and the lane's own result is shown separately as the On finish result.

Lane steps run on the deployment's runner like any other step, so they can use your installed extensions and the credentials stored with your runner.

Adding the On Finish Node

1. Open the workflow in the Designer.
2. In the Flow section of the palette, drag On finish onto the canvas. A workflow can have only one; once it is placed, the palette entry is disabled with the hint "This workflow already has an On-finish node."
3. Drag nodes for the report onto the canvas, such as an Extension tool node, a script node, or an AI Prompt node.
4. Connect the succeeded, failed, or timed out output to the first node for that outcome, and chain further nodes from there. You can connect each output to a different set of nodes.

The Designer draws a dashed region labeled On finish around every node the lane reaches, so you can see at a glance which nodes report the outcome and which do the main work. The On finish node itself has no settings; its outputs carry the meaning.

Reading the Run's Outcome

Any step can read the run itself through {{ run.* }} bindings, in the same places you use {{ nodes.* }} bindings. Unlike nodes.*, which exposes only steps that succeeded, run.* includes failed and timed-out steps, so reading it never fails the step. In a node's binding picker, choose the Run source to insert one.

A step reads the run as it stands when that step starts. A misspelled path, such as {{ run.nope }}, fails the step with a clear error, just like an unknown property on a node.

The first line of run.summary reads, for example, Onboard new user: FAILED after 2m 14s, followed by a line for each step, up to the 50 most recent, so a failed step such as "Add to groups" appears with the error it reported. The text is capped at 4,000 characters.

The AI-Written Summary

{{ run.aiSummary }} is a short, readable narrative of the run, written by AI. It is available only to nodes in the On finish lane:

• It is generated once per run, when the main path ends, and only if a lane node references it. Every lane step reads the same text.
• Generating it consumes Usage Credits, like any other AI call.
• If the AI call fails, takes too long, or returns nothing, run.aiSummary holds the same text as run.summary and the lane runs as normal.

Use run.summary when you want the full step-by-step detail, and run.aiSummary when you want a brief account a technician or end user can read quickly. You can post both.

Example: Posting a Note to the Ticket That Started the Run

This example adds an internal note to an Autotask ticket whenever an onboarding workflow fails. It assumes the Autotask PSA extension is installed and that the workflow is started by a webhook whose payload carries the ticket's internal id, for example in a field named TicketId.

1. Add the On finish node to the workflow.
2. Add an Extension tool node, choose the Autotask PSA extension, and choose the at_create_ticket_note tool.
3. Fill in the arguments:
   • ticketId: {{ nodes.trigger.output.TicketId }}
   • description: {{ run.aiSummary }}, or {{ run.summary }} for the full detail
   • noteType and publish: the note type and visibility values from your own Autotask instance. These ids differ per Autotask instance, so look them up rather than guessing; publish controls who can see the note.
4. Connect the On finish node's failed output to the Extension tool node. Connect timed out to it too if you want a note in that case.
5. Publish the workflow. Because the tool writes to Autotask, an Admin or Owner must approve the call when publishing. See the article The Extension tool node.

When the main path fails, the run shows Finishing while the note is posted, then ends Failed. Approved write calls require runner version 2.3.0 or later; runners that update themselves pick up the current version automatically.

Lane Rules and Limits

• One On finish node per workflow, and nothing can connect into it
• No connection between a lane node and a main-path node, in either direction. The Designer refuses the connection, and publish rejects a workflow that has one.
• For Each and Knowledge-Consolidate nodes cannot be used in the lane
• The lane has its own time budget, 10 minutes by default, separate from the run ceiling. Steps still running when it expires are marked timed out, and the run ends with its main outcome.
• A step in the lane that fails ends its own branch of the lane, the same as on the main path. It is recorded in the On finish result, not the run status.
• An Agent node in the lane cannot wait for a person. If it asks for input or approval, it fails with the reason "An On-finish agent cannot wait for a person", and no Inbox item is created.

What You See in History

• A run whose lane is running shows Finishing and can still be cancelled. Cancelling a Finishing run stops only the lane; the run ends with its main outcome.
• The On finish column in History shows the lane's result beside the run status, or a dash when no lane ran.
• The run detail shows the run's Summary, an On finish field with the lane's result, and the lane's steps after the main-path steps, each marked On finish.

For more on reading runs, see Running a deployment and reading run history.

If you are still having trouble, we're here to help! Submit a ticket here for assistance, and don't forget to check our status page to ensure there are no outages in your area.