What is n8n
n8n is a workflow automation platform: a tool that connects applications and systems so data moves from one to another without anyone copying it by hand. Workflows are built on a visual canvas by linking blocks, and when you need to, you can write code inside the same workflow.
What sets it apart is that you can self-host it on a server or cloud account your company controls, or use the cloud version that n8n manages. Zapier and Make don't offer that choice in the same way, and it's the main reason many companies pick n8n when they handle sensitive data or high volume.
The name is pronounced "n-eight-n" and comes from "nodemation": nodes plus automation. That's a good hint about how it works.
How n8n works: nodes, triggers, and workflows
An n8n workflow is a chain of steps. Each step is a node, and there are three kinds worth telling apart:
- Triggers. The starting point: the event that kicks off the workflow. It can be a submitted form, an incoming message, a new record in the CRM, a schedule (every Monday at 8 a.m.), or an external call to a workflow URL, known as a *webhook*.
- Action nodes. They do something in another system: create a contact in the CRM, post an order to the ERP, send an email, add a row to a spreadsheet. n8n ships with hundreds of ready-made integrations and, for everything else, a generic node that calls any API.
- Logic and data nodes. They decide and transform: filter records, branch on a condition, merge data from two sources, or reshape it. This is also where the code node lives, letting you write JavaScript or Python when the logic doesn't fit in a visual block.
Every time a workflow runs end to end, n8n records an execution. That log shows what data went into and out of each node, and it's the starting point for debugging. It's also the unit n8n uses to price its cloud plans, as covered below.
What you can do with n8n: business examples
The useful question isn't what n8n can do in the abstract, but which of your processes depend on someone copying data today. These are the cases we see most often:
- Orders coming in through a form or WhatsApp. The order arrives, the workflow validates it, creates the order in your business system, and notifies sales or the warehouse. If a field is missing, it asks for it instead of loading an incomplete record.
- CRM and ERP sync. When a rep closes a deal in the CRM, the customer and the sale show up in the ERP with no retyping. And the other way around: invoicing status flows back to the CRM so sales can see it.
- Customer and vendor onboarding. One data entry that reaches every system that needs it, validated once.
- Automated reports. The weekly report someone assembles from spreadsheets is generated and sent on its own, with data from each system.
- Workflows with AI models. One step passes an email, message, or document to a language model to classify it, extract data, or draft a reply, and the rest of the workflow follows fixed rules.
- Alerts. A payment that didn't clear, stock below the minimum, or a delayed order triggers a notice to the person who needs to act.
What they have in common: they happen often, follow rules you can write down, and cost hours or errors today. Those are the best candidates for automation, with n8n or any other tool. Our page on AI automation and system integration lists the questions we use to prioritize.
Is n8n free?
It depends on how you run it. There are two paths:
- Self-hosted Community Edition: free. The standard version of n8n is available from its public repository and runs on your own server. According to the official docs, it includes almost the complete feature set. There's also a Registered Community Edition, still free, that unlocks a few extra features once you register an email. What isn't free is the infrastructure (server, database, backups) or the time of whoever installs, updates, and monitors it.
- n8n Cloud: paid. n8n runs the instance for you and charges a monthly plan based on the number of executions. It offers a 14-day free trial.
One licensing detail worth knowing: n8n isn't open source in the strict sense. The project describes itself as fair-code and is distributed under the Sustainable Use License. In practice, you can use and modify it for free for your company's internal business purposes, but you can't resell it or offer it as a paid service to third parties without an agreement with n8n. Enterprise features (single sign-on, environments, shared projects) require a paid license, including on self-hosted instances.
n8n pricing: cloud plans
These are the plans n8n lists on its official pricing page, checked in September 2026. They are the vendor's prices, shown in euros and billed annually, and they can change:
| Plan | Listed price | Executions per month | Where it runs |
|---|---|---|---|
| Starter | €20 per month | 2,500 | n8n cloud |
| Pro | €50 per month | 10,000 | n8n cloud |
| Business | €667 per month | 40,000 | Self-hosted |
| Enterprise | Contact sales | Custom | n8n cloud or self-hosted |
| Community | Free | No plan | Self-hosted |
According to n8n, every paid plan includes unlimited users, workflows, and integrations, and you pay per full workflow execution regardless of how many steps it has. That's a real difference from tools that charge per step: a long workflow costs the same as a short one.
n8n vs Make vs Zapier
All three solve the same underlying problem, and all three work. The difference comes down to who will use them, where the data has to live, and how much volume you expect.
| n8n | Make | Zapier | |
|---|---|---|---|
| Self-hosting | Yes, on your infrastructure | No, cloud only | No, cloud only |
| Learning curve | Medium: more powerful, needs more technical judgment | Medium: visual and flexible | Low: easiest to start |
| Code inside workflows | Yes, JavaScript and Python | Limited | Limited |
| How it charges | Per full workflow execution | Per usage of each step | Per task (each step that runs) |
| Best for | Sensitive data, high volume, custom logic, AI | Elaborate workflows without coding | Non-technical teams and simple workflows |
Our take: Zapier is the easiest to use, Make handles more elaborate workflows without code, and n8n wins when it matters where your data runs, when volume grows, or when the workflow needs custom logic. For truly complex processes, none of the three replaces a well-built integration.
n8n for AI and agents
n8n includes nodes for working with language models from different providers, including an AI Agent node: a step that receives a goal, decides which tools to use (query a system, look up a record, call another workflow), and responds. The official docs also cover requiring human approval before the agent runs certain tools, and a fallback to a person when the AI can't help.
That opens up useful cases, like an assistant that answers questions with data from your systems or a workflow that reads a document and routes it to the right team. But it helps to separate two things. A workflow with an AI step (classify an email, extract data from a PDF) is something n8n handles very well. An agent that serves customers and writes to your systems is another level: it needs written rules, clear limits on what it can and can't do, handoff to a person, and testing on real cases before launch.
An example of the second kind is M.I.C.A, the agent we built with Integrando Salud: it serves patients over WhatsApp and writes directly to each clinic's management system, registering patients, holding appointment slots, and recording verified payments, always handing off to a person when a case calls for it. What makes an agent like that work in production isn't the tool, it's the design of the rules and exceptions. We cover it in detail in AI agent development.
When n8n is enough and when custom code makes more sense
n8n is enough, and often the best option, when:
- The workflow chains steps across systems that have an API or a ready-made integration.
- The rules are clear and don't change much.
- It helps that your team can see the workflow, understand it, and make small changes.
- You want to get to production quickly and measure before investing more.
Consider custom code when any of these signals show up:
- Complex rules. Many exceptions, chained calculations, or decisions that depend on a lot of context. A visual workflow with dozens of branches becomes hard to read, test, and maintain.
- High volume. When a workflow runs thousands of times a day, performance, queues, retries, and cost per execution all matter. Measure before you decide.
- Sensitive data. Health, financial, or personal information calls for access controls, audit logs, and an architecture designed for it. Self-hosted n8n helps, but it doesn't replace that design.
- The process is core to the business. If the workflow is what sets you apart, it probably deserves to be real software, with automated tests and version control.
It isn't all or nothing. Often the best architecture combines both: n8n orchestrates the simple steps and a custom service handles the part that needs more control. For example, the automated receipt reader we built for a retail company ingests purchase and sales invoices, extracts their data, and posts it to the ERP and CRM according to the company's accounting rules. It's a process with its own business logic, and each piece was chosen based on what that step required.
Security and maintenance for workflows in production
A workflow that works in a demo isn't the same as a workflow in production. These are the points we define before launch:
- Properly stored credentials. Access keys for each system live in the credentials manager, with least-privilege permissions, never hardcoded in a node.
- Error handling. n8n lets you define an error workflow that fires when another one fails. Use it to retry, alert, and park the case in a review queue: nothing gets lost silently.
- Validation at every step. A value that doesn't match the target system isn't loaded; it's flagged for review.
- Execution data. The history stores what passed through each node. If it includes personal data, decide how long it's kept and who can see it.
- Updates. When self-hosting, updating n8n, the server, and the database is on you. Do it with a staging environment and backups.
- Versioning and documentation. Every workflow documented, with an owner, so someone else can maintain it if the person who built it isn't around.
When to hire an n8n automation agency or consultant
Building your first n8n workflow is simple. The hard part comes after: choosing what to automate, connecting systems that were never designed to talk to each other, handling exceptions, and keeping everything running when a vendor changes its API.
It makes sense to bring in an n8n consultant or agency when:
- The workflow touches core systems like your ERP, CRM, or invoicing, and a mistake is expensive.
- You need to choose between n8n, another tool, or custom code, and you want that call made by someone who doesn't sell a single tool.
- You want to self-host n8n with security, backups, and monitoring from day one.
- The process includes an AI step and needs controls and handoff to a person.
- Your team doesn't have time to maintain the workflows, and someone has to respond when something breaks.
At Axlan, we're an n8n expert team and an automation team in general. We use n8n when it's the right choice for the process and custom code when it isn't, and everything runs in your account and is documented, so you don't depend on us. If you have a process in mind, book a free assessment: we'll tell you whether it's worth automating, with which tool, and which systems it touches.

