Start
Triggers and webhooks
Begin workflows on schedules, application events, or incoming HTTP requests.
How the n8n workflow and agent ecosystem fits together
n8n sits between integration automation, low-code development, and agent building. Its strongest mental model remains a visible workflow that may contain carefully bounded AI decisions.
Trigger → Nodes and data → Optional code or AI → Sub-workflow → Observable result
Connect
Nodes and integrations
Call services, transform data, and execute defined operations.
Extend
Code nodes
Add custom logic when visual nodes are not enough.
Interpret
AI nodes and agents
Let models retrieve context, use approved tools, and participate in workflows.
Reuse
Sub-workflows
Package reusable logic and delegate execution without duplicating the whole flow.
Coordinate
Workflows
Hold the visible event-driven path across APIs, services, data, and AI steps.
How to choose within this ecosystem
- Deterministic shell, intelligent step
Keep validation, permissions, routing, retries, and final actions explicit. Use AI only for the part that genuinely requires interpretation.
- A strong learning platform
Its visible nodes make webhooks, APIs, JSON, branches, tool calls, and sub-workflows concrete without hiding the underlying data flow.
- Engineering still matters
Production use needs credential discipline, version control, test data, error workflows, scaling decisions, and careful handling of personal data.
- Learn first
Build a webhook-to-API workflow, inspect every input and output, then add one structured AI step with an explicit fallback.