RobotsHelp

n8n Tutorial for Beginners: Your First AI Workflow in an Hour

Written by

in

You have heard n8n can take whole chores off your plate. Then you open it, see an empty canvas full of plus signs, and close the tab. That is normal. Nobody learns n8n by staring at it.

The short answer: start on n8n Cloud’s free trial, build n8n’s own 15-minute “first workflow” to learn the three ideas that matter (trigger, nodes, data passing from one node to the next), then build one small workflow for your own work with an AI step in the middle. This guide walks you through exactly that, in about an hour, without writing code.

No password needed – your reset opens instantly.

No password. Just your email, and you’re in.

Want short, practical AI lessons like this one? Join Launchpad free, right above.

What n8n actually is, in plain words

n8n is a tool that connects your apps and runs steps for you. You draw the steps on a canvas, like boxes on a whiteboard, and n8n runs them in order. Each box is called a node, and each node does one job: read a form, ask an AI model a question, write a row in a sheet, send an email.

Every workflow starts with a trigger. That is the answer to one question: “When should this run?” Every morning at 8? When someone fills in a form? When a new email arrives? n8n’s own quickstart puts it the same way: the first thing you decide is how the workflow starts (n8n docs: Build your first workflow).

The third idea is the one that makes it click. The output of one node becomes the input of the next. You see that data on screen and drag fields from it into the next step. That drag-and-drop is how you use live data without code.

That is really it. Trigger, nodes, data flowing between them. Everything else is detail you pick up as you go.

Cloud or self-hosted: what a beginner should pick

You can use n8n two ways. n8n hosts it for you (n8n Cloud), or you run it on your own computer or server (self-hosted). n8n’s own guide is clear about who should do what: if you want to start right away and you do not have technical expertise, pick n8n Cloud, because there is nothing to install. If you want to run n8n for free, self-host the Community edition (n8n docs: Choose how to use n8n).

My honest advice: start on Cloud. Self-hosting is great later, but on day one it adds Docker, updates and backups to a learning curve you do not need yet.

Here is what I checked on n8n’s pricing page today (n8n pricing):

  • Free trial: no credit card needed. It comes with 1,000 executions, 5 at the same time, a 180-second run time limit and 100 AI credits. n8n’s docs describe it as a 14-day trial with Pro features.
  • Starter: 20 euro a month, billed annually. Unlimited workflows and users, AI nodes included, 5 executions at the same time.
  • Pro: 50 euro a month, billed annually, for people running automations every day.
  • Community edition: free, but you host it yourself (n8n docs: Host n8n).

One thing that makes n8n easy to budget: it charges per execution, and an execution is one full run of a workflow, no matter how many steps it has. A workflow that runs once a day uses about 30 executions a month. That is how n8n explains it on the same pricing page. If you want the full comparison with a per-task tool, read n8n vs Zapier for agency workflows.

Step 1: Build n8n’s own first workflow (15 minutes)

Do not skip this. n8n’s official quickstart is short, it needs no accounts or API keys, and it teaches exactly the moves you need for everything else (n8n docs: Build your first workflow). You build a small “Weather Log”:

  1. Schedule Trigger: run every day at 8am.
  2. HTTP Request: fetch the weather for Berlin from n8n’s sample-data service.
  3. Edit Fields (Set): turn the raw data into one readable sentence by dragging fields in.
  4. Data table: save each report as a new row, so you build a log over time.

When you drag a field into a box, n8n writes something like {{ $json.city }} for you. That is an expression: a placeholder n8n swaps for the real value when the workflow runs. You never have to type these by hand. Drag, look at the preview under the field, done.

At the end you press Publish. That is the switch between “I am testing this” and “run it for me from now on.” Remember that word, because it trips up almost every beginner later.

Step 2: Your first useful workflow with an AI step

Now the part that actually saves you time. We build a workflow many freelancers and small teams would love: someone fills in your contact form, AI reads the message, sorts it and writes a draft reply, and the draft waits in your Gmail for you to check.

The problem it solves: enquiries sit unanswered because writing that first reply feels like a task. The result: you open Gmail, find a ready draft, tweak one line and press send. You stay in charge of every word that leaves your inbox.

Four nodes. That is the whole build.

Node 1: n8n Form Trigger (the “when”)

Click Add first step and pick the n8n Form trigger. n8n builds the form page for you (n8n docs: n8n Form Trigger). Give it a title like “Ask me anything” and add three fields:

  • name (Text, required)
  • email (Email, required)
  • message (Textarea, required)

Press Execute step. n8n opens the test form. Fill it in as if you were a client, something like “Hi, do you do bookkeeping for small webshops? What would a monthly package cost?” Submit, and you will see your answers appear as data in the node.

Node 2: OpenAI (the AI step)

Click the plus after the form and add the OpenAI node, then choose a text action that sends a message to a model. On n8n Cloud you can pick Use Gateway credits in the credential field, so you can run this step without your own OpenAI account (n8n docs: OpenAI node). If you prefer Claude or Gemini, n8n has an Anthropic node and a Google Gemini node too. The idea stays the same.

In the prompt field, paste this and drag the form fields into the spots in brackets:

You help me answer enquiries for my small business.

Here is a new message from [name] ([email]):
[message]

Do two things:
1. On the first line, write ONE label: QUOTE, QUESTION, SUPPORT or OTHER.
2. Below that, write a short, friendly reply in my voice (max 120 words).
   Thank them by first name, answer what you can, and suggest a 15-minute call
   if they asked about prices. Never invent prices or promises.
   Sign off with "Best, [your first name]".

Change the sign-off and the labels to fit your work. Press Execute step and read what comes back. If the tone is off, fix the prompt, not the output. Two or three rounds is normal.

Tip from experience: the line “Never invent prices or promises” is not decoration. Without it, a model will sometimes make up a number that sounds right. If you want to get better at prompts like this, my Claude tutorial for beginners covers the same three-part prompt idea.

Node 3: Edit Fields (tidy up)

Add an Edit Fields (Set) node, the same one you used in the weather workflow. Create two fields:

  • subject: type “Re: your question” and drag the label from the AI output after it, so you can see at a glance what kind of message it is.
  • reply: drag in the AI’s text.

This step is optional, but it keeps the next node clean and teaches you the single most useful habit in n8n: shape the data before you send it anywhere.

Node 4: Gmail, create a draft (the “result”)

Add the Gmail node and choose Create a draft (n8n docs: Gmail node). Connect your Google account when n8n asks. Then drag in the fields: the visitor’s email as recipient, your subject, and your reply as the message.

Why a draft and not “Send a message”? Because your first AI workflow should never email a client without you seeing it. Run it for a few weeks. Once you trust the replies for one label, say QUESTION, you can switch that branch to sending. Not before.

Test it, then publish it

Press Execute workflow, fill in the test form again and check Gmail. You should find a draft with the client’s address, your subject and a reply that sounds roughly like you.

Then press Publish and switch the form to its Production URL. This is the step beginners miss. n8n’s docs explain that the Test URL only works while you are testing in the editor, and the Production URL only works once the workflow is published (n8n docs: n8n Form Trigger). Put the production link on your website or in your email signature.

After it runs for real, open the Executions tab of the workflow to see every run and what each node did (n8n docs: Executions). That tab is where you find out why something did not happen.

What the finished workflow looks like

Read it left to right, like a sentence:

Form submitted  ->  OpenAI sorts it and drafts a reply  ->  Edit Fields tidies it  ->  Gmail saves a draft

Picture a one-person design studio. Before: enquiries pile up until Sunday night. After: the first draft is already waiting, and answering takes a coffee, not an evening. That is the whole point. Not magic, just one boring step gone.

Five beginner mistakes that cost the most time

  1. Forgetting to publish. The workflow works when you test it, then “does nothing” all week. Check that it is published, and that you use the Production URL.
  2. Building the dream workflow first. Start with four nodes that solve one small problem. Add branches later, when the simple version has run for real.
  3. Typing data instead of dragging it. Dragging fields in builds the expression for you and shows a preview. Typing by hand is where typos break things.
  4. Letting AI send on its own too early. Draft first, send later. One wrong automatic email to a client costs more than all the time you saved.
  5. Never looking at Executions. When something goes wrong, the Executions tab shows the exact node and the exact data. Look there before you rebuild anything.

Where to go after your first two workflows

Once these two run, you know enough to copy almost any n8n template and understand it. Good next steps:

Want structured lessons from n8n itself? The free n8n Academy has beginner courses, starting with “Essentials: Your First Workflows”. And the n8n template library is a great place to steal ideas once you know how to read a workflow.

Got your first two workflows running? The next step is an agent that decides which tool to use. Our n8n AI agent workflow guide walks you through building a meeting-prep agent that reads your calendar and email and drafts a short prep note for you to check.

FAQ

Do I need to know how to code to use n8n?

No. n8n’s own first-workflow tutorial says you need no coding experience, and everything in this guide is drag and drop. n8n does let you add code steps later if you ever want to, but you can get real work done without them.

Is n8n free?

The self-hosted Community edition is free. n8n Cloud has a free trial without a credit card, then paid plans starting at 20 euro a month billed annually, as shown on the n8n pricing page today. Prices change, so check that page before you decide.

How long does it take to learn n8n?

The official first workflow takes about 15 minutes, according to n8n. Plan about an hour for the AI workflow in this guide, including connecting Gmail. In my experience, feeling at home with branches and error handling takes a few weeks of building real things, not a weekend of videos.

Is n8n better than Zapier for a beginner?

In my experience, Zapier is a little quicker for a simple two-step automation. n8n shines once your workflow has more steps or an AI part, because you pay per full run instead of per step. Our Zapier free plan examples show what the other side looks like.

Can n8n use Claude or Gemini instead of OpenAI?

Yes. n8n has its own nodes for Anthropic (Claude) and Google Gemini, next to OpenAI. Swap the OpenAI node for the one you prefer and keep the same prompt.

Build yours this week

Pick one message you answer over and over. Build the four-node workflow tonight, run it as a draft-only helper for two weeks, and see how much of your inbox quietly gets lighter. If you want more small, practical wins like this one, every week, it is free:

No password needed – your reset opens instantly.

No password. Just your email, and you’re in.

Sources

🇳🇱 Deze pagina in het Nederlands