Blog · Automation

What is n8n and what is it used for in business

Updated: 2 October 2026 · reading time about 7 minutes

A website form lands in an inbox, someone copies it into the CRM, then sends an offer, and finally enters a date in the calendar. Each of these steps takes a moment, but together they add up to hours every week and plenty of chances for mistakes. n8n is one of the tools that let you automate such chains of tasks. Below we explain what it is, how it works and when it makes sense in a small or mid-size company.

What is n8n

n8n (pronounced "n-eight-n") is a workflow automation tool. It lets you connect apps, databases and services so that an event in one place automatically triggers actions in others. A process is built visually by connecting blocks on a screen, and where ready-made blocks are not enough, you can add a piece of code.

The project started in Germany and is developed under a model its creators call fair-code: the source code is publicly available, and the tool can be installed on your own server and used in your own company. The license is not a classic open source license, though – among other things, it restricts reselling n8n as a commercial service. For a company automating its own processes this difference usually has no practical impact, but it is worth knowing about.

How it works: nodes, triggers and workflows

• A workflow is the whole process – from the event that starts it to the last action.
• A trigger is the starting point: a new row in a spreadsheet, an email, a submitted form, a call to a webhook address or a schedule, e.g. every day at 7:00.
• Nodes are the successive steps: fetch data from the CRM, filter it, save it to a spreadsheet, send a message. n8n has ready-made nodes for several hundred popular services, and for the rest a universal node for HTTP requests that lets you connect to any system that exposes an API.
• Logic – conditions, branches, loops, error handling and retries. This is what sets n8n apart from simple "if this, do that" tools.

Self-hosting or the cloud version

n8n can be used in two ways. The cloud version is run by the tool's creators: you create an account and start building processes right away, without thinking about a server. The self-hosted version runs on your server or in your cloud infrastructure – then you decide where the data is physically located, but you are also responsible for updates, backups and security.

For many companies, the ability to install it in-house is the main reason to choose n8n. Customer data flowing through automations does not have to pass through an external platform.

What a small company can automate

• Handling inquiries. A website form goes into the CRM, the customer gets an automatic confirmation, and the sales rep gets a notification with the key details.
• Documents and invoices. Attachments from a given mailbox are saved to the right folder, and invoice data goes to a spreadsheet or accounting system for verification.
• Reports. A daily or weekly summary of sales, orders or tickets sent automatically by email or messenger.
• Data synchronization. Stock levels between the shop and the warehouse system, contacts between the CRM and the newsletter tool.
• Reminders and deadlines. A reminder about an appointment the day before, a notice about a contract that is ending, a nudge for an unpaid invoice.
• Onboarding a new client or employee. Creating accounts, sending documents, creating tasks in the project management system.

n8n and artificial intelligence

In recent years n8n has significantly expanded its AI features. It has nodes for connecting to popular language models, both cloud-based and run locally, and an AI agent node that can decide on its own which tools to use to complete a task. Add to that the elements needed to build a knowledge base: conversation memory, connections to vector databases and document processing.

In practice this means you can add an "intelligent" step to an ordinary automation: the model reads an incoming message, recognizes whether it is a complaint, a request for a quote or spam, extracts the data and routes the case onward. Or it prepares a draft reply that an employee only approves. How an AI agent differs from ordinary automation is explained in the article what is an AI agent.

n8n vs Make vs Zapier

All three tools serve the same purpose – connecting apps into automated processes – but they differ in philosophy. Specific prices and limits change often, so we compare the approach, not the numbers.

• Hosting. Zapier and Make run only as cloud services of the provider. n8n can be used in the cloud or installed in-house – this is the most important difference.
• Billing model. Zapier and Make generally bill by the number of steps or operations executed, so an elaborate process that runs often quickly raises the bill. n8n in the cloud counts executions of the whole workflow, and in the self-hosted version you mainly pay for the server and the time of the person who maintains it.
• Ease of getting started. Zapier is the simplest for a non-technical person and has the most ready-made integrations. Make gives more control over the data flow with a still approachable visual interface. n8n requires somewhat more technical knowledge.
• Flexibility. n8n lets you add your own code in JavaScript or Python, build complex logic and connect to almost any API. For unusual processes this often decides the choice.

In short: for a few simple connections between popular apps, Zapier or Make can be quicker to set up. When there are many processes, they are complex, they handle personal data or they need to run on your own server – n8n usually comes out ahead.

When n8n is enough and when you need more

n8n works great where a process can be described as a chain of steps: X arrived, check Y, save in Z. Even with AI steps along the way.

The limits appear in a few places:

• Real-time conversation. An AI voice agent answering phone calls or a sophisticated chatbot serving many customers at once needs a separate architecture. n8n can be their back end – saving the conversation, creating a ticket – but should not be the core.
• Very large data volumes. For mass processing, dedicated code is usually more efficient and easier to maintain.
• Complex business logic. When a workflow has dozens of nodes and dozens of exceptions, it becomes hard to understand and test. That is a sign that part of the logic should be moved to a separate service.
• An agent that is meant to handle a case on its own. A simple agent in n8n works well, but a system that solves a customer's problem over many steps, using several company systems, usually requires a dedicated solution. We write more about this on the page about AI agents for business.

Security and GDPR with self-hosting

Automations usually process personal data: names, email addresses, phone numbers, the content of inquiries. With cloud tools this data passes through the provider's infrastructure, often outside the EU, which requires a data processing agreement and a check of the transfer basis.

n8n installed on a server in the EU or in your company simplifies this picture, but does not remove your obligations. You need to take care of:

• regular updates and backups,
• secure storage of passwords and access keys to connected systems,
• restricted access to the panel and an encrypted connection,
• considered logging – execution history may contain personal data, so it is worth deciding how long it is kept,
• including automations in the company's GDPR documentation.

If a process includes an AI step and the data is sensitive, you can connect n8n to a local language model. Then neither the data nor its processing leaves your infrastructure.

How we use n8n at Axisway

For us, n8n is above all the "glue" that connects the parts of an implementation. A chatbot or voice agent talks to the customer, and n8n handles what happens next: saving to the CRM, a calendar entry, a notification for the team, an end-of-day report. Where a ready-made integration does not exist, we write our own code or a separate service.

In projects with sensitive data we install n8n on the client's infrastructure, together with local LLMs on Mac Mini computers or GPU servers. The first working version of an automation usually takes 7–14 days. We describe the details on the page about process automation with AI.

Summary

n8n is a flexible automation tool that can run in the cloud or on your own server, with extensive AI features. It works well for processes that can be described in steps and as a back end for chatbots and agents. For simple connections Zapier or Make can be quicker, and for real-time conversations or very complex logic you need a dedicated solution. The most important thing is to start with the process, not the tool – which is why we begin implementations with an AI audit that shows which tasks are worth automating first.