Four jobs you did today that nobody needed a person for

Think for a moment about how much of today's work a computer could really have done by itself. Four examples, and you will recognize at least one of them:

• In the morning you open Slack and paste today's meetings into the team channel one by one, so everyone knows where you are.
• A customer fills in a Google Form, you open the spreadsheet and copy the data across, and you write them a confirmation email while you are at it.
• An email from a VIP customer lands in your inbox and you forward it to the sales manager straight away, in case it gets missed.
• It is the end of the month, so you open the admin pages of three platforms, pull last month's sales figures, paste them into the report in Excel, and screenshot it for your boss.

Those four have one thing in common. The rules are fixed; the data is not. The steps are the same every time, and only the content inside them changes. That is precisely what an automation tool is best at: you state the rules once, and after that it runs itself every day. It does not skip anything, it does not forget, and it does not call in sick.

The tool the company picked for this is n8n, pronounced “n-eight-n”, short for “nodemation”. This first article is the outline. Everything after it walks you through building workflows that really run.

Hundreds of blocks that each do exactly one job

If n8n has to fit into one sentence, it is this: an automation tool built like Lego.

Every Lego brick has a fixed shape and a set of studs, and once you click them together you have a car or a house. n8n works the same way. It ships with hundreds of ready-made building blocks called nodes, and each node does a single job:

Your job is the other half: drag the blocks onto the screen and join them with connections. A connection is the direction the data travels in — the mail that Gmail reads runs along it to the next block.

So what would this look like without n8n? If you wrote the morning Slack job yourself, you would have to work out how to fetch mail with the Gmail API, which headers to send, and how to parse the JSON that comes back. n8n has already packed all of that into the Gmail block. What is left for you is the fields beside it: which account to use, how many recent messages to fetch, and whether to limit it to one sender.

Three blocks and two connections

Take the first of those four jobs — post the calendar to Slack every morning — and translate it into n8n blocks. There are three of them, and each one hands its result to the next.

flowchart LR
  T["Schedule Trigger
8:00 every morning"] --> C["Google Calendar
fetch every event
between 00:00 and 23:59 today"]
  C -->|"each event: name, time, location"| S["Slack
build one message,
post it to your channel"]

1. Block 1

   Schedule Trigger — a nudge at 8:00 every morning

   This is where the workflow starts. It has exactly one job: when the time comes, it sets everything downstream in motion. You can have it run every hour, at a fixed time each day, at 9:00 every Monday, on the 1st of each month, and so on.

2. Block 2

   Google Calendar — fetch every event for today

   Connect it with your Google account and tell it to fetch every event between 00:00 and 23:59 today. It sends the events — name, time, location — down the connection to the next block, one at a time.

3. Block 3

   Slack — send the list to the channel you chose

   It catches the list of events from the previous block, builds one message out of it, and posts it to a channel in your Slack, or sends it to you as a direct message.

Three building blocks and two connections, and “post the calendar automatically every morning” is done. This series walks you through building exactly this one, step by step, in the Your first workflow article. For now, take the scale from it: a workflow is usually three to seven building blocks, and rarely much longer.

n8n against Zapier, Make and Power Automate

n8n is not the only tool that does this kind of work, and it is not automatically the right one. You have probably heard of, or used, one of the others. Here are all four on the same axes.

Make was formerly called Integromat, in case you meet the old name in a forum post. The row that decides most of this in practice is the second one, so do the arithmetic rather than reading the word “free” and moving on.

The short version: Zapier for the occasional personal connection, n8n when a company runs things at volume. That is why this handbook is in front of you at work today.

Six reasons, and the one that mattered most

We evaluated Zapier, Make and Power Automate, and settled on n8n. Six reasons, in the order they weighed:

The first of those six is the hardest to reverse later, so it is worth seeing rather than reading.

flowchart LR
  subgraph W["Woow n8n, self-hosted Community edition"]
    direction LR
    W1["Your Gmail,
Sheets, Slack"] --> W2["The company server
runs the workflow"]
    W2 --> W3["Your execution history
and credentials stay there"]
  end
  subgraph V["A hosted tool: Zapier, Make, Power Automate"]
    direction LR
    V1["Your Gmail,
Sheets, Slack"] --> V2["The vendor's cloud
runs the workflow"]
    V2 --> V3["Your execution history
and credentials live there"]
  end

Nothing else crosses the line. The only data that leaves the company is what you explicitly tell a workflow to send out — a Slack message, a Gmail send. Your workflow definitions, your execution history and your credentials are shared with neither n8n itself nor any third party.

Credentials are stored encrypted in the database, encrypted with N8N_ENCRYPTION_KEY. That key belongs to the platform, not to a workflow — you never paste it, or any other secret, into a node parameter.

Six jobs it is made for, and five it is wrong for

n8n is particularly good at a recognizable set of work. If something similar is on your plate this week, find your row and note which blocks it involves — you will meet all of them later in the series.

To be straight with you, n8n is not the answer to everything. Force it into the five situations below and you will regret it.

Both tables collapse into a few questions you can run in your head before you open the Canvas.

flowchart TD
  A["A repetitive job
you want handed over"] --> B{"Do the rules change
from one run to the next?"}
  B -->|"yes"| N1["Not automation yet.
Settle the rules first"]
  B -->|"no"| C{"Does it need to run
hundreds of times a second?"}
  C -->|"yes"| N2["Write a dedicated service"]
  C -->|"no"| D{"Is it crunching millions
of rows, or heavy CPU work?"}
  D -->|"yes"| N3["Airflow, Dagster or dbt"]
  D -->|"no"| E{"Is it something customers
click through themselves?"}
  E -->|"yes"| N4["Retool, AppSmith,
or build it yourself"]
  E -->|"no"| F["This is n8n's shape.
Three to seven nodes, joining
tools you already run"]

The five things people ask before they start, and the map for the rest

Before you touch anything, some of these question marks may be in your head. Answering them here stops them distracting you halfway through.

That last one is the safety net under everything else in this series, so it is worth a picture.

stateDiagram-v2
  Inactive : Inactive
  Inactive : run it by hand from the Canvas, as often as you like
  Active : Active
  Active : the schedule fires on its own, and webhooks are answered
  [*] --> Inactive
  Inactive --> Active : you press Activate
  Active --> Inactive : you switch it off again

The habit that follows from it: the first time you take a workflow live, run it manually a few times to check, then switch it to Active.

How to read the rest of this

The guide behind this series is 22 chapters grouped into 7 parts. Read them in order, or pick out the ones you need.

Questions that come up again

Where to go from here