What is n8n: how the automation tool works and how a small business uses it

What is n8n, in the words of the people who make it

People look up what is n8n after meeting the name in a job advert or a chat about AI. n8n’s own documentation answers in one line: it is a fair-code licensed workflow automation tool that combines AI features with business process automation. Wikipedia adds that it is developed by n8n GmbH, a Berlin company, and was founded by Jan Oberhauser and first released publicly in 2019. The odd name, the GitHub README explains, is short for “nodemation” and is pronounced n-eight-n.

In practice n8n is a place where you draw a process instead of repeating it by hand. You put steps on a canvas, join them with lines, and the program runs the chain each time something happens: a form is sent, an email arrives, Monday morning comes. As of 20 October 2026, the README counts more than 1,500 integrations and lets you mix the visual canvas with custom code.

Workflows, nodes, triggers and credentials in the n8n editor

Four words from n8n’s glossary explain most of the editor. A workflow is a collection of nodes that automate a process. A node is one building block: an entry point for retrieving data, a function to process it or an exit for sending it. A trigger node starts the workflow when a condition is met, and every production workflow needs at least one. Credentials store the login details for other services, such as a password, an API key or OAuth secrets.

Between the trigger and the last node sit actions: read a row, send a message, create a record. When a service has its own node, you fill in fields. When it does not, the HTTP Request node can query any app with a REST API, though the documentation warns that this needs a basic grasp of how APIs work. Each run is called an execution, and the list of past executions is the first place to look when something breaks.

The n8n licence in plain words: fair-code, not open source

n8n does not call itself open source, and its licence page says why: open source licences cannot limit use. The company calls its model fair-code instead. The source code is openly available, the software is generally free to use and extend, and its authors restrict commercial use. The main repository is covered by the Sustainable Use License, which n8n created in 2022, while files marked .ee. need a separate Enterprise licence.

The Sustainable Use License sets three limits. You may use or modify the software only for your own internal business purposes or for non-commercial or personal use. You may pass it on only free of charge and for non-commercial purposes. And you may not remove the licensor’s notices. The licence FAQ sums it up: if workflows are only created or modified by you or people in your organisation, the community version is free for you.

The same FAQ, as of 20 October 2026, allows running your business on n8n, building automations for clients on your own instance as long as they cannot create or edit them, and charging for setup, maintenance, training and consulting. It rules out hosting n8n as a service where clients build their own workflows, white-labelling it, forking it into a rival product and using enterprise features without a licence. These rules cover self-hosted n8n, while n8n Cloud is a paid subscription with its own terms. Borderline cases are worth checking with n8n and with a lawyer.

n8n Cloud or your own server: who keeps it running

n8n’s page on choosing a setup names two options. n8n Cloud is fully managed: no installation, no technical expertise, and n8n handles hosting, updates and scaling. It starts with a 14-day trial with Pro features, followed by a paid plan. According to n8n’s security page, the cloud runs on Microsoft Azure and its data is currently hosted in the European Union. This route suits a team without a technical person, or one testing an idea this month.

Self-hosting is free in the Community edition, which has almost the complete feature set, but the work becomes yours: the server, installation with Docker or the one-line script, the database, users and security settings. n8n’s update guide suggests updating at least once a month, and its backup guide recommends a full backup before every update. Credentials are stored encrypted, so a restored database is useless without the key. Your own server pays off when someone already looks after servers or the data must stay in a particular country.

AI nodes and agents as n8n’s documentation describes them

n8n offers two ways to put a language model to work. The AI Agent node sits inside an ordinary workflow as a cluster node: a root node with sub-nodes for a chat model, memory and tools, and at least one tool is required. The agent decides which tools to call to complete a task. A chain, by contrast, calls the model in a fixed sequence without choosing its own steps, which is more predictable for simple jobs such as summarising a letter.

Separately, n8n now has standalone agents with a model, instructions, tools and a knowledge base of uploaded files, which people reach through chat, channels and schedules. The documentation marks them as Preview and says plainly that they can make mistakes. Hence a simple rule: let the model draft, sort and summarise, and keep a person’s approval before anything reaches a customer or moves money.

Three n8n workflows a small business can copy

The first workflow handles a website request. Step one is the n8n Form Trigger, which starts a workflow when someone submits a form and can generate the form page itself; a form on your own site can post its data to a Webhook node instead. Step two writes the name, phone and question into your CRM, through its node or the HTTP Request node. Step three is the Telegram node, using a bot token that BotFather issues after the /newbot command, which sends the request to the sales chat.

The second collects supplier invoices. An Email Trigger (IMAP) node watches the accounts mailbox and can download attachments, an option the documentation advises enabling only when needed, since it increases processing. The Extract From File node turns a PDF or a spreadsheet into structured data, and an AI step or a few expressions pick out the supplier, date and amount. A Google Sheets node appends them as a row, and the Send Email node, in a mode that waits for a reply, asks the accountant to approve the entry.

The third is a weekly report. The Schedule Trigger runs workflows at fixed times, much like the Cron utility on Unix-like systems; set it for Monday at nine and check the time zone, because a self-hosted instance defaults to America/New York. A Google Sheets node reads last week’s rows, a small step adds them up, an optional AI step writes a short comment, and the Telegram node sends it to the owner.

Where n8n stops: errors, upkeep and where the data lives

Automations fail quietly: a password changes, a field is renamed, and the workflow stops without telling anyone. n8n’s guide to error handling lets you give each workflow an error workflow that starts with the Error Trigger node and sends, for example, an email or a chat alert. Failures, the guide says, usually come from errors in node settings or from a workflow running out of memory. On n8n Cloud the platform does the server work, but the logic stays yours.

Then there is personal data. Customer names, phones and emails pass through n8n’s server and can stay in its execution data. Russia’s Federal Law No. 152-FZ on personal data, in part 5 of Article 18 as amended in February 2025 and published in the ConsultantPlus legal database, does not allow recording or storing Russian citizens’ personal data in databases located outside Russia, apart from exceptions listed in the law. Other countries have rules of their own, so check your case with a specialist before customer data goes through a workflow.

Starting with n8n this week, or ordering the setup

To start alone this week, open a Cloud trial, or run the one-line installer on a computer with Docker to see the self-hosted version. Pick one small, annoying task, such as a form that should reach a chat. Build it from a template or from scratch, run it with test data and add an error workflow before relying on it. Ordering the setup makes sense when the process touches money or customers, when several services must be joined through their APIs, when personal data rules apply, or when nobody inside will look after the server.

Before talking to anyone, describe the process as it runs today, step by step: who does what, which services are involved and who holds admin access to each. With that page, a studio such as VITON13 Studio can scope the job precisely, from workflows and hosting to upkeep and error reports. Ask for the workflows to be built on your own instance or account, so the result stays yours. Whether you build or order, what is n8n then turns into a practical question: one dull job done reliably, with someone watching.