Node reference

A process flow is built from nodes on the canvas. This topic lists each node type, the fields it requires before the process flow can be saved, and whether it produces output that later steps can read. For how output passes between nodes, see How data moves through a process flow.

Node types

Node Purpose Required fields Produces output

Start event

The single entry point of the process flow.

Exactly one per canvas.

None. The run’s starting data is provided as interface data.

End event

The single exit point of the process flow.

Exactly one per canvas.

No.

Run Script

Runs a script from the Script Editor.

A selected script.

Yes, through Generate Outputs.

Send Email

Sends an email from a template.

A valid recipient address, and a template.

No.

User Task

Pauses the process flow until a person selects an action.

A title, an approver group, and at least one action.

Yes, the selected action’s value.

Agent Task

Sends the previous step’s output to a configured AI agent and returns the response. See Agent Task reference.

A selected agent.

Yes, the agent’s response.

Decision

Routes the process flow down one or more paths, based on conditions.

An expression in every condition.

No. It relays the incoming output.

Wait for Branches

Waits for every incoming path to finish before continuing.

None.

No.

Table Interaction

Reads or writes a table through a table definition.

A selected table, and an operation.

Yes, through Generate Outputs.

Save checks

Before a process flow can be saved, it must also pass these checks:

  • The process flow has a name and a development package.

  • The canvas has exactly one Start event and one End event.

  • There is a connected path from the Start event to the End event.