A Beginner's Guide to AI Automation with n8n

What is n8n?

n8n (pronounced "n-eight-n") is a fair-code licensed workflow automation platform. Think Zapier or Make, but self-hosted, extensible, and with native support for large language model integrations. As of 2026, n8n ships with built-in nodes for OpenAI, Anthropic, Ollama (for local models), and dozens of other AI services.

Self-Hosting n8n

The quickest path to a running n8n instance is Docker Compose:

services:
  n8n:
    image: n8nio/n8n:latest
    restart: unless-stopped
    ports:
      - "5678:5678"
    environment:
      N8N_BASIC_AUTH_ACTIVE: "true"
      N8N_BASIC_AUTH_USER: admin
      N8N_BASIC_AUTH_PASSWORD: changeme
      WEBHOOK_URL: https://n8n.yourdomain.com
    volumes:
      - n8n_data:/home/node/.n8n
volumes:   n8n_data: 

Point Caddy or Nginx at port 5678, provision a TLS certificate, and you have a production-ready automation server.

Core Concepts

Workflows are directed graphs. Each node is a step: trigger, action, or transformation. Data passes between nodes as arrays of JSON objects called items.

Triggers start workflows:

• Webhook (HTTP POST from any external service)
• Schedule (cron expression)
• File watch
• Manual ("Test workflow" button)

Nodes do work:

• HTTP Request — call any REST API
• Code — run JavaScript or Python
• AI Agent — orchestrate LLM reasoning loops
• Database — read/write Postgres, MySQL, SQLite

Your First Workflow: Summarize RSS Feed Items with an LLM

Goal: Every morning, fetch the last 24 hours of posts from an RSS feed, summarize each one with an LLM, and email you a digest.

Step 1 — Schedule Trigger

Add a "Schedule Trigger" node. Set it to run daily at 07:00.

Step 2 — HTTP Request to RSS Feed

Add an "HTTP Request" node:

• Method: GET
• URL: https://news.ycombinator.com/rss

Add a "RSS Feed Read" node (built-in) to parse the XML into structured items.

Step 3 — Filter by Date

Add a "Filter" node to keep only items where pubDate is within the last 24 hours:

// Filter node expression
{{ new Date($json.pubDate) > new Date(Date.now() - 86400000) }}

Step 4 — Summarize with OpenAI

Add an "OpenAI" node:

• Operation: Message a model
• Model: gpt-4o-mini (cheap, fast)
• Prompt: Summarize this article in 2 sentences for a developer audience: {{ $json.description }}

Step 5 — Aggregate and Send Email

Use the "Aggregate" node to collect all summaries, then an "Email Send" (SMTP) node to deliver the digest.

Total build time: about 20 minutes. No code beyond one filter expression.

AI Agent Workflows

n8n's AI Agent node lets you build ReAct-style reasoning loops. The agent has:

• A system prompt defining its role and constraints.
• Tools — other n8n nodes it can call (HTTP Request, database queries, calendar lookups).
• A memory node to retain context across calls.

Example: Automated Inbox Triage

The agent reasons through the email, decides on a classification, and calls the appropriate tool — all without a predefined decision tree.

Practical AI Automation Ideas for Developers

Connecting to Local Models with Ollama

For sensitive data you don't want leaving your network, swap OpenAI for Ollama:

# Add to your compose file
  ollama:
    image: ollama/ollama:latest
    restart: unless-stopped
    volumes:
      - ollama_data:/root/.ollama

In n8n, use the "Ollama Chat Model" node and point it at http://ollama:11434. Models like llama3.2, mistral, and phi4 run well on a machine with 16 GB RAM.

Conclusion

n8n makes AI automation accessible to solo developers without requiring a data engineering background. Self-hosting gives you full control over your data and eliminates per-task pricing. The WEDC member library includes a curated set of n8n workflow templates for common developer automation scenarios, updated monthly.