Workflow nodes overview

The CloudRadial AutomationAI Designer builds a workflow from nodes you drag onto the canvas and connect into a flow. This article introduces the palette, the node types in each group, the fields most nodes share, and how nodes pass data to one another. It is for workspace users with the Admin role or higher who build workflows, and it serves as the index for the per-node articles that follow.

In this article:

• The palette
• Scripts
• AI
• Integrations
• Knowledge
• Logic
• Flow
• Common fields
• Connecting nodes and passing data

The Palette

The Designer palette is the left rail. It groups the node types you can add into sections: Scripts, AI, Integrations, Logic, Knowledge, and Flow. Drag a card onto the canvas to add a node, then select it to open its settings. Every workflow also has a Start node and an End node; the Start node is created automatically and the End node terminates the flow. A workflow can also have one On finish node, which starts a separate lane that reports the run's outcome.

Scripts

Script nodes run code on a runner. They share one set of fields and differ only in the language and the runner flavor that executes them:

• PowerShell Script, run a script step in PowerShell
• Python Script, run a Python step
• C# Script, run a C# step

AI

AI nodes call a language model and consume Usage Credits:

• AI Prompt, a single prompt that produces one output
• Agent, a goal-directed node that reasons and uses tools, adapting at runtime

Integrations

Integration nodes call your installed extensions directly:

• Extension tool, make one call to one tool of an installed extension, such as adding a ticket note, with arguments you set as literals or bindings. No AI model is involved. A tool that writes to another system must be approved by an Admin or Owner when the workflow is published or tested.

For detail, see The Extension tool node.

Knowledge

Knowledge nodes let a workflow author your knowledge base as it runs:

• Knowledge-Write, write content into your knowledge base
• Knowledge-Consolidate, merge and tidy existing knowledge

For detail on both, see Authoring knowledge in workflows: Knowledge-Write and Knowledge-Consolidate nodes.

Logic

Logic nodes control the shape of the flow:

• Condition, branch on a rule (if/else)
• For Each, run a step once per item in a collection (loop or fan out)

Flow

The Flow section holds the End node, which terminates the workflow. The Start node is the entry point and is not in the palette because it is added automatically and cannot be deleted. For both, see The Start and End nodes.

The Flow section also holds the On finish node, one per workflow. It has no input and three outputs, succeeded, failed, and timed out. The nodes connected to it form the On finish lane, which runs once after the main path ends, from the output that matches the run's outcome; for example, to post the outcome to the ticket that started the run. The Designer draws a dashed region around the lane, and lane nodes cannot connect to main-path nodes. For detail, see Reporting a run's outcome with the On finish lane.

Common Fields

Most nodes share a small set of fields:

• Name, the node's display label on the canvas. Renaming a node never breaks a binding, because expressions reference a node's stable id, not its name.
• Timeout (seconds), how long the step may run before it times out. Script nodes accept 1 to 1800 seconds (default 1800).
• Retry count, how many times a failed script step is retried, from 0 to 5 (default 0).

The AI, Integrations, and Logic nodes carry their own additional fields, which the per-node articles describe. The Condition node runs in the control plane and has no timeout or retry. The On finish node has no fields; its outputs carry the meaning.

Connecting Nodes and Passing Data

Connect nodes by dragging from one node's output handle to the next node's input handle. A node runs after the node connected into it. When one node's output connects to more than one node, those branches run at the same time, in no set order, so don't rely on one branch finishing before another. A step that fails or times out ends its branch: the nodes connected after it do not run, and the run ends Failed or Timed out. To act on a failure, connect nodes to the On finish node's failed or timed out output. Each node writes a JSON output object that later nodes read with binding expressions. Any node can also read the run itself, including failed steps, with {{ run.* }} bindings such as {{ run.status }} and {{ run.summary }}. For the full data contract, output shape, binding syntax, the trigger payload, filters, and default versus explicit inputs, see Passing data between nodes: input and output JSON.

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.