n8n: visual workflow automation you can self-host

What is n8n?

n8n is a workflow automation tool that lets you wire together the services you already use without writing a full application from scratch. Instead of glue scripts and cron jobs scattered across servers, you build flows on a canvas: each step is a node, and data flows from one node to the next. It sits in the same broad space as classic no-code automation tools, but with two distinguishing traits: it is source-available and self-hostable, and it treats raw API access and custom code as first-class citizens rather than an afterthought.

The project is released under a fair-code model (the Sustainable Use License). In practice that means the Community Edition is free to download and self-host, including for internal business automation, while a managed cloud offering and certain enterprise features are commercial. This combination is why n8n is popular with developers and operations teams who want the convenience of a visual workflow builder but also need control over where their data lives.

The visual no-code / low-code workflow builder

The heart of n8n is its editor. You start with a trigger — a schedule, an incoming webhook, a new row in a database, or a message in a chat app — and then chain nodes that fetch, transform and send data. Because the canvas shows the real data passing between steps, you can inspect the exact JSON output of each node and map fields into the next one without guessing.

This makes it genuinely usable as no-code automation for straightforward tasks, while still being a low-code tool when you need precision. A marketing operations specialist can build a lead-routing flow by clicking through nodes; an engineer can extend that same flow with a code node, a custom API call, or a sub-workflow. You are not locked into prebuilt blocks the way you are with many closed app-automation platforms.

What a typical node graph contains

• Triggers — webhooks, schedules, app events, chat messages or manual runs that start the workflow.
• Action nodes — hundreds of integrations for SaaS apps, databases, message queues and email.
• Logic nodes — IF/Switch branching, merges, loops over items, and wait/delay steps.
• Data nodes — set, edit, filter and aggregate fields, or run JavaScript/Python on the items.
• HTTP Request — a universal node to call any REST or GraphQL endpoint that has no dedicated integration yet.

Self-hosted vs n8n Cloud

One of the most common questions is whether to self-host or use the managed service. Both run the same workflow engine; the difference is who operates the infrastructure. Self-hosted automation gives you maximum control and predictable cost at scale, at the price of doing your own updates, backups and monitoring. n8n Cloud trades some control for convenience.

If data residency, internal-network access, or running local AI models matters to you, self-hosting is usually the deciding factor. If you simply want workflows live this afternoon, the cloud option removes the setup work.

AI workflow automation and AI agents

n8n has grown well beyond basic app integrations into AI automation. Alongside its standard nodes, it ships purpose-built AI nodes: chat-model nodes for large language models, embeddings and vector-store nodes for retrieval, memory nodes to keep conversation context, and an AI Agent node that can reason over a task and call tools to complete it.

What makes this practical is that the agent's "tools" can be your existing workflows and integrations. An AI agent can look up a customer in your CRM, query a database, send an email, or trigger another flow — all as steps it chooses dynamically. Because n8n can self-host, you can also point these nodes at local models for local AI automation, keeping prompts and data inside your own network rather than sending everything to an external API.

Patterns people build with n8n AI

• Retrieval-augmented chatbots that answer from your own documents using a vector store.
• Triage and routing where a model classifies incoming tickets or emails before they reach a human.
• Content and data enrichment that summarizes, extracts fields, or translates records in bulk.
• Tool-using agents that combine reasoning with real actions across connected apps.

Common use cases

How to install and run n8n

At a high level you have three routes: the managed cloud, a Docker container, or a global npm install. For a local PC setup on Windows, macOS or Linux, Docker is the most reliable because it bundles everything and isolates the app. The steps below are the general shape of the process — always confirm the current commands in the official documentation before running them.

1. Pick how you want to run it

   Choose n8n Cloud for zero setup, or self-hosting via Docker or npm if you want it on your own machine or server.

2. Run it on your PC with Docker (Windows / macOS / Linux)

   Install Docker Desktop (on Windows this uses WSL2), then start a container that maps the editor port and keeps your data in a persistent volume.

   docker volume create n8n_data
   docker run -it --rm \
     --name n8n \
     -p 5678:5678 \
     -v n8n_data:/home/node/.n8n \
     docker.n8n.io/n8nio/n8n

3. Or install via npm

   If you have a recent Node.js LTS version, you can install the CLI globally and start it directly. This is handy for quick local experimentation.

   npm install n8n -g
   n8n

4. Open the editor

   Visit http://localhost:5678 in your browser, create your owner account, and you are in the visual builder ready to add your first node.

5. Secure and persist it for real use

   For anything beyond testing, put it behind HTTPS, set authentication, and back up the data volume. Review the official self-hosting guide for environment variables and production hardening.

There is no officially maintained standalone "n8n desktop" installer today, so when people search for an n8n download for Windows, the practical answer is Docker or npm rather than a single .exe. To get the real, current binaries and instructions, use the official project by name — see n8n.io and its documentation.

Frequently asked questions