What is a Node?
A node is a single unit of execution within a workflow. Each node performs a specific action—calling an AI model, executing a tool, transforming data, or controlling execution flow. Nodes are connected by edges to create complex automation pipelines.Self-Contained
Each node has its own configuration, inputs, outputs, and execution logic
Visual Interface
Nodes are represented as boxes on the canvas with input/output handles
Typed Connections
Nodes connect via edges that define data flow and execution order
Real-time Status
See execution status, outputs, errors, and costs as workflows run
Node Anatomy
Handles
Nodes connect to each other through handles - connection points that define data flow:Input Handles (Left Side)
Input Handles (Left Side)
Receive data and execution flow
- Located on the left side of nodes
- Accept connections from previous nodes
- Trigger node execution when activated
- Pass data from source node to target node
Output Handles (Right Side)
Output Handles (Right Side)
Send data to next nodesCommon output handle types:
- PARALLEL: Main execution path (success)
- ERROR: Error handling path (failures)
- TOOLS: Tool calling path (Agent → Tool nodes)
- IF/ELIF/ELSE: Conditional branching (Switch nodes)
Node Configuration
Every node has configurable properties:- Name & Description: Identify the node’s purpose
- Variable Mappings: Reference data from previous nodes
- Type-Specific Settings: Model selection, tool configuration, conditions, etc.
- Error Handling: How the node responds to failures
Detailed Documentation
Start Node
Workflow entry and trigger point
Agent Node
Autonomous AI agent with tool calling
Tool Node
Execute operations and actions
Switch Node
Conditional branching logic
Merge Node
Combine execution paths
Variable Mappings
Reference data with Pongo2
Node Lifecycle
Execution states and behavior
What’s Next?
Edges
Learn how nodes connect and how data flows between them
Workflows
Understand how nodes combine to create complete workflows
Build Your First Agent
Follow a step-by-step tutorial to build an AI agent