GuideIntermediate
AI Automation Basics with n8n
Learn to build AI automations in n8n that classify, extract, summarize, and generate content: no prior AI experience or programming beyond simple expressions required. The guide sets up a 100% local, zero-cost AI workshop with the Self-hosted AI Starter Kit (n8n, Ollama, Qdrant, and PostgreSQL), connects cloud models (Claude, GPT, Gemini) and local ones, teaches prompt writing inside workflows, structured output, and building a RAG pipeline over your own documents. It closes with native AI agents that use tools and MCP, all integrated into a smart email assistant that gets debugged with real executions and is ready for a portfolio.
- lessons
- 64
- modules
- 8
- available in
- English · Spanish
- certificate
- Yes
- access
- Free
What you'll be able to do
- Tell what AI does and doesn't do inside an automation, and decide between cloud and self-hosted
- Install and use the Self-hosted AI Starter Kit with free local models via Ollama (Llama 3.2 and other open models)
- Connect cloud AI models (Claude, GPT, Gemini) and local ones, and choose a model by cost, latency, and privacy
- Write prompts inside workflows: system vs. user, injecting data with expressions, few-shot, and guardrails against hallucination
- Build AI chains to classify, extract, and summarize text with the Text Classifier and Information Extractor nodes
- Get validated structured output with the Structured Output Parser and use it to generate content
- Set up a 100% local RAG pipeline: embeddings with Ollama, a vector store (Qdrant), document chunking, and retrieval
- Use the native AI Agent node with tools, memory, and delegation to sub-agents, plus an introduction to MCP (client and server)
- Bring it all together in a smart email assistant with triage, RAG, agent-drafted replies, and human review before sending
What you need to bring
It's for you if...
- People who already automate with n8n and want to add AI to their workflows with no prior AI/ML experience
- Anyone who wants to evaluate 100% local, zero-cost AI before committing to paid APIs
- Automation builders who need to move from fixed rules to classification, extraction, and content generation with LLMs
- Anyone who wants to take the first step toward RAG and AI agents without writing code from scratch
Requirements and materials
- n8n fundamentals: building workflows, adding nodes, triggers, mapping data between nodes, and basic expressions
- Basic Docker knowledge to run the Self-hosted AI Starter Kit with `docker compose`
- No prior AI/ML experience, math, or programming beyond simple expressions required
The syllabus, module by module
Open any of them to see its lessons.
- 1. Overview: what you're going to build in this guide
- 2. What AI does and doesn't do in an automation
- 3. n8n 2.x tour for AI workflows
- 4. Cloud vs self-hosted: which to use and why
- 5. Installing the Self-hosted AI Starter Kit
- 6. Local models with Ollama: Llama 3 and Mistral
- 7. Your first AI node (100% local)
- 8. Project: your zero-cost AI workshop
- 1. Module overview: the model is the engine
- 2. How chat model sub-nodes work
- 3. The Basic LLM Chain node
- 4. Connecting cloud models: Claude, GPT, and Gemini
- 5. Connecting local models with Ollama
- 6. The parameters that matter: temperature, tokens, and more
- 7. Choosing a model: cost, latency, and privacy
- 8. Project: swap a model without rebuilding the workflow
- 1. Module overview: the prompt as code
- 2. System prompt vs user prompt
- 3. Injecting node data with expressions
- 4. Few-shot, context, and constraints
- 5. Prompt templates and reuse
- 6. Guardrails against hallucination
- 7. Debugging prompts with execution replay and variable tracing
- 8. Project: a reliable prompt template
- 1. Module overview: the three tasks that drive 80%
- 2. Classifying text with the Text Classifier node
- 3. Extracting data with the Information Extractor node
- 4. Summarizing without losing what matters
- 5. Sentiment and intent
- 6. Chaining AI steps together
- 7. Processing multiple items and respecting limits
- 8. Project: an incoming text triage chain
- 1. Module overview: from loose text to data that flows
- 2. Why structured output matters
- 3. The Structured Output Parser
- 4. Defining a schema for AI output
- 5. Validating and repairing output
- 6. Generating content: drafts and replies
- 7. From AI output to tables and apps
- 8. Project: structured generator with validation
- 1. Module overview: giving AI memory and sources
- 2. What embeddings are
- 3. Local embeddings with Ollama
- 4. The vector store node: Qdrant and Milvus
- 5. Loading and splitting documents (local Document AI)
- 6. Building the retrieval step
- 7. Question answering over your documents
- 8. Project: a zero-cost local RAG
- 1. Final project overview: what you're going to deliver
- 2. Designing the assistant's architecture
- 3. Triage: classify and extract
- 4. Retrieving context with the local RAG
- 5. Drafting replies with an agent
- 6. Routing and human approval
- 7. Debugging and hardening the flow
- 8. Project: ship and defend your assistant
What people usually ask
No limit. It's a free guide: come in whenever you like, as often as you like.
No. Modules run from easier to harder, but you can jump to the one you need. Progress is saved per lesson.
Whatever is needed is listed under “What you need to bring”, above. If nothing is listed there, you can start from zero.
In the WhatsApp group, and every two weeks there's a live with an instructor where questions get worked through.
Yes. When you finish every lesson you get a statement of completion. Pass the guide's certification exam and it becomes a certificate, with a verifiable code you can share on LinkedIn.
Start whenever you like
What students say
These reviews are from enrolled students who completed at least 50% of the course. We moderate reviews only on content grounds (spam, offensive language, personal data), never for being critical or negative.
No approved reviews yet.
Be the first to share your experience!