n8n is a workflow automation platform that you can install on your own server or computer. Like Zapier and Make, it connects apps together, but two things set it apart: the self-hosted version is free with no limit on runs, and it has strong AI nodes for building agents, meaning agents that have memory, call tools and connect to a cloud model or a local one. That combination has made it popular with developers and businesses that want control over where their data and workflows run.

What is n8n good for?#

  • A Telegram support bot: it receives a customer message, drafts a reply using a language model and your product information, and escalates to a human operator when needed.
  • Connecting a store to accounting: it receives WooCommerce orders through a webhook and records them in a spreadsheet or internal software.
  • A daily news digest: every morning it reads the feeds of a few sites, summarizes them with AI and sends them to the team.
  • An agent with a local model: by connecting to Ollama, it processes internal company documents without sending them to an outside server.
  • Programming tasks: running JavaScript or Python code inside a workflow, calling APIs and scheduling periodic jobs.

Strengths and weaknesses#

Strengths:

  • A free self-hosted version with no run cap; your data stays on your own server.
  • Advanced AI and agent nodes, with MCP support and local models.
  • High flexibility for developers through code nodes and HTTP requests.
  • A large amount of tutorial content and an active community.

Weaknesses:

  • Installation, updates, backups and server security are your job.
  • Harder than Zapier for someone who doesn't program.
  • The license is not open source in the strict sense, and selling it to others as a service is restricted.

Free tier and pricing#

You install the Community edition on your own server for free. The n8n cloud version and enterprise features, such as single sign-on and team management, are paid. Check the official site for current pricing.

Language support#

The interface is in English, but text in other languages moves through workflows without problems, and the quality of the AI nodes' output in a given language depends on the model you connect. You can install the self-hosted version on a virtual server or even a personal computer. One thing to check: depending on where your server is hosted, it may be blocked from reaching some external services such as language-model APIs, and pulling Docker images sometimes needs a regional mirror. The cloud version requires an online payment method.

Getting started in three steps#

  1. Install n8n with Docker or npm on your server or machine.
  2. Start from a ready-made template, such as a Telegram bot or an email summary.
  3. Before exposing it to the internet, set a strong password, HTTPS and regular backups.

n8n alternatives#

  • Make: similar logic on a visual canvas, without the hassle of running a server.
  • Zapier: the simplest cloud option with the most ready-made integrations.
  • Lindy: a ready-made AI assistant for email and meetings, with no workflow to build.

For a better grasp of agents, read what an AI agent is and how it differs from a chatbot and what MCP is: connecting tools to an AI assistant.