n8n

What it is

n8n is a workflow automation platform in the same family as the tools described in connecting your apps without a developer: when something happens in one app, n8n carries the data into your other apps automatically. It has around 1,900 pre-built integrations, and workflows are built on a visual canvas.

Two things make it different from the mainstream options. First, you can drop real code (JavaScript or Python) into any workflow, so you're never stuck when the visual blocks run out. Second, you can self-host it: run n8n on your own server, keep every byte of data in-house, and pay nothing for the software. That combination has also made it a popular way to build AI agent workflows, where a language model makes decisions inside an automation you can fully inspect.

Who it's for (and who it isn't)

n8n fits teams with at least one technically comfortable person, businesses with data that shouldn't pass through a third party (self-hosting keeps automations on your own infrastructure), and anyone whose automations have outgrown simple two-step recipes. Its cloud pricing charges per workflow execution rather than per step, so a complex workflow costs the same to run as a trivial one. Heavy, multi-step automations are exactly where that model wins.

It's a weaker fit for a non-technical solo user automating simple things: the canvas assumes you're comfortable thinking in data fields, and the easier tools cover simple recipes with far less learning curve. If that's you, start with Zapier and move to n8n when you hit its limits or its bill.

Pricing

n8n is freemium with an unusual shape: the self-hosted community edition is free, and the hosted cloud starts from EUR 20 per month, billed annually. Cloud plans include unlimited users and workflows, with the price tied to how many executions you run. If you have a server and some patience, the free self-hosted route is the cheapest serious automation setup available.

Set it up

Official guide: Build your first workflow

1. Create an account on n8n Cloud, which is n8n running on n8n's own servers (cloud from EUR 20/mo, billed annually). You install nothing. The free community edition means installing n8n on your own server, so these steps do not use it. After you sign in, you see "Personal" in the left menu.
2. Select "Personal" in the left menu, then "Create workflow". A workflow is a chain of small steps that n8n runs for you. Rename it "Weather Log". You now have an empty workflow with that name.
3. Select "Add first step", then "On a schedule". This is the trigger, the step that starts the workflow. Set "Trigger at Hour" to 8am, then close the settings panel with the "x" at the top right. You now see the schedule step on your canvas.
4. Select "+" after the schedule step, search for "http", and select "HTTP Request". This step asks a website for data. Paste https://mockingbird.n8n.io/v1/weather/berlin into the "URL" field and select "Execute step". You now see the weather for Berlin in the output.
5. Select "+" after the HTTP Request step, then "Data transformation", then "Edit Fields (Set)". Select "Add Field", name it "report", and set its value to: The temperature in {{ $json.city }} is {{ $json.temp_c }}°C. It's currently {{ $json.condition }}. Select "Execute step". You now see a "report" field holding one sentence on the weather in Berlin.
6. Select "Execute workflow" to run all the steps once. Then select "Publish" in the header, which switches the workflow on. It now runs by itself every day at 8am.

When this works, you have a workflow that runs by itself every day at 8am, fetches the weather for Berlin and writes a one-sentence report.

Strengths and limitations

n8n's strengths are control and cost at scale: self-hosting for free, code when you need it, execution-based pricing that doesn't punish complexity, and AI agent workflows where you can see every decision the model made.

The limitation is the learning curve. The power assumes technical comfort, and self-hosting means you own updates, backups, and uptime. A non-technical team that just wants two apps talking will ship the same result faster on a simpler platform.