Nodes are the building blocks of AI Workflows in Canvas. Each node performs a specific function — from LLM calls and API requests to data transformations and control flow.

Node Categories

Input & Output

Node	Purpose
Input Node	Define workflow input parameters
Output Node	Define workflow output structure

AI & LLM

Node	Purpose
Universal LLM	Call any configured LLM model
Run AI Worker	Prompt an AI Worker with your input and receive its response

Data & APIs

Node	Purpose
API Call	Make HTTP requests to external APIs
Read URL	Fetch and parse web content
Browser	Headless browser automation
CSV to JSON	Convert CSV data to JSON format
PDF to Images	Convert PDF pages to images
Render Markdown	Convert Markdown to HTML, PDF, or DOCX

Storage & Memory

Node	Purpose
Storage Upload	Upload files to storage
Vector Save	Save content to vector memory
Vector Search	Semantic search in vector memory
Save to Collection	Save data to MongoDB collection
Find in Collection	Query MongoDB collections

Control Flow

Node	Purpose
Run AI Workflow	Execute another workflow with specific inputs
Route Workflow	Evaluate conditions and execute different workflows based on the result
Repeat Workflow	Execute a workflow for each item in a list, in parallel
Combine Workflow Results	Combine results from multiple workflow executions into one
Repeat Workflow Until	Execute a workflow repeatedly until a condition is met

Foundation

Node	Purpose
Custom Nodes	Execute JavaScript/Python code

Common Patterns

Parameter Templating: Use {{nodeId.field}} syntax to reference outputs from previous nodes.

Example: {{1.response.data.items}} references the items array from node 1's response.