n8n: what it is and how to start

A visual workflow builder that connects your apps and moves information between them automatically, with optional AI steps — and you can run it on your own server for free.

Who it is for

Small businesses and solo operators who are copying information between tools by hand and want it to stop. Also the best option for anyone whose automation volume makes per-run pricing painful, since self-hosting costs the same whether you run a hundred workflows a month or a hundred thousand.

What it costs

The self-hosted Community Edition is free with no execution limits — you provide the server. n8n Cloud is paid: Starter is EUR 20 per month billed annually for 2,500 workflow executions, Pro is EUR 50 per month for 10,000, and Business is EUR 667 per month. Starter and Pro offer a trial with no credit card required; the Business trial requires one. Note that Cloud has a free trial rather than a permanent free tier.

Quick facts

Category: AI Automation & Chatbots — Workflow Automation
Pricing model: freemium
Runs on: Web (cloud), Self-hosted (Docker, Linux, Windows, macOS)
Difficulty for a beginner: intermediate
Official site: https://n8n.io

How to get started with n8n

Decide between cloud and self-hosting first: Cloud means signing up and building in minutes, on a paid plan after the trial. Self-hosting is free forever but you need somewhere to run it and some comfort with a terminal. If you are unsure, start on the cloud trial to learn the tool, then move if the pricing bothers you — your workflows export and import as files.
Write down the process before opening n8n: List the trigger, every step, and every decision, on paper. n8n will happily let you build something complicated before you know what you want, and untangling that later is harder than starting from a clear list.
Build the smallest possible first workflow: Two nodes: a trigger and one action. A form submission that adds a spreadsheet row is ideal. Getting one trivial thing working end to end teaches you more about how n8n thinks than any tutorial.
Connect your accounts as credentials: n8n stores each app connection as a reusable credential. Use a shared or service account rather than a personal one, because a personal account taken away later silently breaks every workflow that depended on it.
Run it manually and read every step: Click Execute Workflow and inspect the data going in and out of each node. This panel is the single most useful thing in n8n — most problems are visible immediately as a field with a different name or shape than you assumed.
Add error handling before going live: Set an error workflow, or at minimum a notification, so a failure reaches you. Automations fail silently by default, and a broken workflow can go unnoticed for weeks.
Only then consider adding an AI step: Get the workflow correct with fixed logic first. Add a model step only where the task genuinely needs judgement — classifying a message, drafting a reply — and give it explicit instructions on what to do when it is unsure.

Should you use this instead of something else?

Zapier and Make are easier to start with and connect to more apps out of the box, and if you are running a handful of automations they are the sensible pick. n8n wins on cost at volume, because self-hosting does not charge per run, and on control, because your data and credentials stay on your server. It costs you a steeper learning curve and, if self-hosted, the job of keeping a server running.

Common beginner mistakes

Building a large, branching workflow before getting a two-node version working. Debugging complexity you never verified in pieces is the slowest way to work.
Using a personal account for credentials, so the automation dies when that person changes a password or leaves.
Skipping error handling, which means failures are silent and get discovered by a customer rather than by you.
Putting an AI model into a step that a simple rule would handle, which adds cost, latency and a new way to be wrong.
Self-hosting on day one without server experience. Learning the tool and learning to run a server at the same time turns one problem into two.

Last verified: 2026-08-07. Source: https://n8n.io/pricing/, https://github.com/n8n-io/n8n

Read the full AI Automation & Chatbots guide for how this fits into the wider process.