AI Automation with n8n is a free n8n AI automation course for people who want AI to do real work between their apps, not just answer questions in a chat window. It has 10 short modules in a set order: the three ideas behind every automation, your first n8n workflow, rules, APIs, an AI agent, two partner tools, error alerts and finally an assistant that answers from your own documents. You practice every click in a copy of the tool in your browser, then repeat it in your own account. You can start the course here, beginning with the free Triggers, Actions and Data Mapping lesson.
This page is for office managers, operations and marketing staff, and owners at accounting firms, agencies, law offices and other white-collar teams. It explains what each module covers, why the order matters and what you need in your own account. The examples use made-up businesses and fake client data only. Disclosure: Stepthrough has an affiliate relationship with n8n, so I may earn a commission, at no cost to you, if you sign up through an n8n link on Stepthrough. There are no affiliate links on this page.
Why learn AI automation in n8n?
Most teams use AI one chat at a time. Someone pastes an email into ChatGPT, copies the answer into Gmail, then updates a spreadsheet by hand. It helps, but the person is still the glue between every app. The real time savings come when AI sits inside the workflow: an email arrives, AI reads it, a rule routes it, a person approves the draft, and the reply is waiting.
n8n is the tool Stepthrough teaches for this, for three reasons. Its AI Agent node gives a chat model memory and tools, such as a Google Sheet or an API, in one place. It counts one execution per run of a whole workflow, however many steps it has, so adding an AI step does not multiply the bill. And it has built-in nodes for answering from your own documents, which is the job most businesses ask for once their first agent works.
If you are still choosing a tool, read n8n vs Make vs Zapier for AI automation. The same ideas are taught in AI Automation with Make and AI Automation with Zapier.
Who is this n8n course for?
- Office and operations staff who copy information between apps every day and want it to happen on its own.
- Marketing and sales teams who want AI to read, sort and draft, with a person approving what goes out.
- Owners and team leads who want to know what an AI agent can safely do before they let one near a client.
You do not need any coding background. The later modules use an API key, which you paste into an n8n credential, not code. If you have never built an automation at all, the first module starts from zero.
What are the 10 modules, in order?
The course goes beginner to advanced. Modules 1 to 4 build the foundation. Modules 5 to 7 add AI with a person in charge. Modules 8 to 10 make it dependable and give the AI your own knowledge.
| # | Module | The business problem it solves | Level |
|---|---|---|---|
| 1 | Triggers, actions and data mapping | Copying the same details from one app to another by hand | Beginner |
| 2 | What is n8n? Build your first workflow | Client inquiries that sit unseen in an inbox | Beginner |
| 3 | IF and Switch: route work by rules | Every request landing in the same pile | Beginner |
| 4 | Call any API with the HTTP Request node | An app you need has no ready-made n8n node | Intermediate |
| 5 | Your first n8n AI agent | Hours spent on first drafts and ideas | Intermediate |
| 6 | ProLevelMax: live Reddit data for your agent | A workflow that goes quiet when Reddit RSS shuts down | Advanced |
| 7 | Guardian: make your agent ask before it emails | An AI agent emailing a client with nobody checking | Intermediate |
| 8 | Error alerts and automatic retries | A silent failure that nobody notices for a week | Intermediate |
| 9 | Load your handbook into Pinecone | Company knowledge the AI cannot see | Advanced |
| 10 | A chatbot that answers from your own docs | Staff waiting on Operations for policy answers | Advanced |
Modules 1 to 4: the foundation
Module 1 is a short concept lesson on plain diagrams, with no app at all. You learn that a trigger starts every automation, actions do one job each, and field mapping carries the details from one step to the next. You also fix the most common beginner mistake: typing words into a box instead of mapping it, so every row says the same thing.
Module 2 opens n8n. You learn how n8n thinks, in nodes on a canvas, and build a client inquiry form that catches every request and writes a one-line summary. Module 3 adds decisions: an IF node splits big jobs from small ones, and a Switch node routes each request by service, with a fallback for the rest.
Module 4 is the skill that makes n8n work with almost anything: the HTTP Request node. You GET a client list from a practice API, filter it with a query parameter, then POST a note back using an expression. After this module, "does n8n connect to our CRM?" stops being a blocker.
Modules 5 to 7: AI, with a person in charge
Module 5 builds your first AI agent with n8n's AI Agent node, a chat model and memory. Module 6 is a partner lesson built in the real ProLevelMax console and n8n: Reddit is switching off its RSS feeds on November 13, so you replace an RSS Read node with one HTTP Request node that pulls the newest subreddit posts from ProLevelMax with a bearer token, then let an AI agent answer questions only from those posts.
Module 7 is the one to take before any agent talks to a client. Guardian is a partner tool that sits between an n8n agent and the actions it can take. You turn a plain policy, "team emails go out alone, client emails wait for a person, deletes never happen", into Guardian rules, test it, and see where the Guardian node and the approval inbox fit in the workflow. It is human in the loop that the agent cannot skip.
Modules 8 to 10: dependable, and grounded in your own knowledge
Module 8 makes failures find you. You build an error workflow that emails you when any workflow fails, make a flaky step retry on its own, and retry a failed run from the Executions list.
Modules 9 and 10 are RAG: retrieval-augmented generation, or letting the AI look things up before it answers. In module 9 you create a free Pinecone index and build a workflow that reads a handbook from Google Drive, cuts it into chunks, turns them into embeddings and stores them in a vector database. In module 10 you build the chatbot staff actually use: it searches the handbook before every answer, cites the page and says "I don't know" instead of guessing.
How does each module work?
- Why it matters. The business problem and what fixing it is worth.
- Practice. Every step in a copy of the tool in your browser. It checks your work, and nothing touches your real accounts.
- Do it for real. A short checklist to repeat it in your own n8n, with sample files where the module needs them.
- Quick check. Three questions, then Up next to the following module.
Most modules take 6 to 15 minutes. Your place is saved in your browser, so you can stop anywhere.
What do you need in your own account?
- n8n. n8n Cloud has a free trial, then the Starter plan is listed at €20 a month billed annually for 2,500 executions (n8n pricing page, October 2026). The self-hosted Community Edition is free software if you have someone to run the server.
- An OpenAI API key from module 5 on. It is billed per use by OpenAI, separately from any ChatGPT plan, and a few dollars of credit covers the practice.
- ProLevelMax for module 6: pay as you go, with free starter credit for new users, as listed on prolevelmax.com in October 2026. Guardian for module 7. Pinecone for modules 9 and 10, which has a free starter plan.
- No real client data. Use the sample data in each module until your settings and approvals are in place.
Key takeaways
- Learn AI automation in order: triggers, actions and mapping first, then rules and APIs, then AI with a person approving, then reliability and your own documents.
- n8n's AI Agent node gives a model memory and tools, and n8n counts one execution per run however many steps it has.
- Put an approval step, such as Guardian, between any AI agent and your clients.
- RAG lets an assistant answer from your own handbook and cite the page instead of guessing.
Frequently asked questions
- Is this n8n AI automation course free?
Yes. Every module and every practice is free and runs in your browser without an n8n account. You only pay for the tools when you repeat the modules in your own account, and n8n Cloud has a free trial.
- Do I need to know how to code to use n8n with AI?
No. Everything is point and click. You will paste an API key into a credential and drag fields into boxes to make expressions, but you never write a program.
- How long does the course take?
About two hours of practice in total. Do one module, use it on a real job, then take the next.
- What is the difference between an n8n workflow and an AI agent?
A workflow follows the same steps every time. An AI agent is one node in a workflow that takes a goal, decides which tools to use and writes the result. Most useful automations mix both: rules for the predictable parts, an agent for the reading and writing.
- Is it safe to let an AI agent email clients?
Only with an approval step. That is why module 7 puts a Guardian approval gate in front of client emails, and why every AI module starts with fake data.
Sources
- n8n pricing (checked October 2026)
- n8n documentation: AI Agent node
- OpenAI API pricing
- ProLevelMax pricing (checked October 19, 2026)
- Pinecone pricing