n8n uses a node based visual canvas where you connect nodes (steps) with wires. Each node represents an action, trigger, or data transformation. Data flows left to right through the nodes. You can branch, merge, loop, and add conditional logic visually.
Trigger types
App Event TriggersNative triggers for supported services such as 'New email in Gmail' or 'New row in Google Sheets.'
WebhooksReceive HTTP requests from any external service to kick off a workflow.
Schedule / CronTime based execution: every hour, daily, weekly, or custom cron expressions.
Chat TriggersStart workflows via conversational chat input for AI agent interfaces.
Sub Workflow TriggersNest workflows and trigger one from another for modular automation design.
Manual TriggersOne click execution for testing and debugging during development.
Node types
Trigger NodesStart a workflow from events, schedules, webhooks, or manual execution.
Action NodesPerform operations on 400+ integrated services including Google, Slack, GitHub, and Salesforce.
Core NodesIF, Switch, Merge, Loop, Set, Function, Code (JS/Python), HTTP Request, and Webhook.
AI NodesAgents, chains, LLMs, memory, tools, vector stores, and embeddings across 70+ AI specific nodes.
Community Nodes100+ user contributed integrations extending n8n beyond the official catalog.
Debugging is a key differentiator
You can re run individual steps without executing the entire workflow, pin mock data on nodes to avoid waiting for external systems, view logs inline next to code, and search through full execution history. This makes iterating on complex workflows significantly faster than competing platforms where you typically need to run the entire automation from scratch.
220
Workflow executions per second
A single n8n instance handles up to 220 executions/sec. For higher throughput, the Enterprise plan supports queue mode with multiple workers and a multi main setup.