n8n Guide: Automate Your Business with AI, No Code

Why n8n for AI?

n8n is not just a traditional automation tool. Recent versions added native AI nodes that integrate directly with language models — whether external APIs or local models on your server. This means you build intelligent automation chains without writing a single line of code.

The difference between traditional automation and AI automation:

• Traditional: If an email arrives from a specific address, send a predetermined reply
• AI: If an email arrives, read its content, understand the intent, classify it, and write an appropriate response

AI Nodes in n8n

AI Agent

The core node. Receives inputs and routes them to a language model with instructions (System Prompt):

Setting	Description
Model	The model used (Ollama, OpenAI, etc.)
System Message	Fixed instructions for the model
Tools	Additional tools the model can use
Memory	Conversation context preservation

Text Classifier

Automatic text classification:

• Support tickets (technical/billing/general)
• Email messages (important/normal/promotional)
• Customer reviews (positive/negative/neutral)

Summarization Chain

Summarizing long content:

• Reports and documents
• Meeting minutes
• Articles and posts

Information Extractor

Extracting structured data from unstructured text:

• Names, numbers, dates from contracts
• Product information from free-form descriptions
• Contact details from messages

Setting Up n8n with Ollama (Local Models)

Installation

mkdir -p /opt/n8n-ai && cd /opt/n8n-ai

cat > docker-compose.yml << 'EOF'
version: '3.8'
services:
 n8n:
 image: n8nio/n8n:latest
 restart: always
 ports:
 - "5678:5678"
 environment:
 - N8N_HOST=n8n.yourdomain.com
 - N8N_PROTOCOL=https
 - WEBHOOK_URL=https://n8n.yourdomain.com/
 - N8N_ENCRYPTION_KEY=your-secret-key
 volumes:
 - n8n_data:/home/node/.n8n
 extra_hosts:
 - "host.docker.internal:host-gateway"

volumes:
 n8n_data:
EOF

docker compose up -d

Connecting Ollama

1. In n8n settings, go to Credentials
2. Add a new Ollama API credential
3. URL: http://host.docker.internal:11434
4. Test the connection

Choosing the Right Model

Model	Size	Best Use	Arabic Performance
Llama 3.1 8B	4.7 GB (Q4)	General purpose	Good
Mistral 7B	4.1 GB (Q4)	Classification and extraction	Medium
Qwen 2.5 7B	4.4 GB (Q4)	Arabic content	Excellent
Gemma 2 9B	5.4 GB (Q4)	Summaries and reports	Good

5 Ready-to-Deploy AI Automation Chains

1. Intelligent Support Assistant

Goal: Every support ticket gets classified and receives an automatic initial response.

Node Chain:

1. Webhook — receives ticket from support system
2. AI Agent (Classification):

• System Prompt: "Classify this ticket: technical, billing, account, or general. Return JSON."
• Receives ticket subject and body

3. Switch — routes based on classification
4. AI Agent (Response):

• Different System Prompt per category
• Generates professional response with solution steps

5. HTTP Request — sends response to support system as draft
6. Slack — notification to the relevant team

Result: Tickets classified in under 3 seconds with a suggested response. Team reviews and sends.

2. Email Analysis and Task Extraction

Goal: Every important email converts to a task in project management.

Node Chain:

1. Email Trigger — monitors inbox
2. AI Agent (Analysis):

• "Analyze this email. Extract: main topic, priority (urgent/normal/low), required tasks, deadline if any."

3. IF — filter: only urgent or normal priority messages
4. HTTP Request — creates task in Notion or Todoist
5. Slack — notification about new task

3. Automatic Content Generator

Goal: Create draft social media content from a single topic.

Node Chain:

1. Webhook / Form — receives topic and target audience
2. AI Agent (Research): searches for key points about the topic
3. AI Agent (Writing):

• System Prompt: "Write 3 posts: one for Twitter (280 chars), one for LinkedIn (2 paragraphs), and one for blog (500 words). Professional and direct style. No empty marketing phrases."

4. Google Sheets — saves drafts
5. Email — sends drafts for review

4. Intelligent Price Monitor

Goal: Monitor competitor prices and analyze changes.

Node Chain:

1. Cron — daily at 8 AM
2. HTTP Request — fetches competitor pricing pages
3. AI Agent (Extraction):

• "Extract all prices from this page. Return JSON with product, price, and currency."

4. Function — compares with yesterday's prices
5. AI Agent (Analysis):

• "Analyze these changes. Is there a pattern? Do we need to adjust our prices?"

6. Slack — daily report with recommendations

5. Recruitment Assistant

Goal: Automatically filter and evaluate resumes.

Node Chain:

1. Email Trigger — monitors recruitment inbox
2. IF — filters messages with attachments
3. Read Binary File — reads the resume (PDF)
4. AI Agent (Evaluation):

• System Prompt: "Evaluate this resume for the [Job Title] position. Rate 1-10 with justification. Extract: name, years of experience, matching skills, missing skills."

5. IF — score 7+ continues
6. Google Sheets — logs all applicants with scores
7. Email — sends top candidates to the team

Connecting n8n to External API (OpenAI / Claude)

If you need more powerful models than local ones:

# In Credentials settings
# OpenAI:
API Key: sk-xxx
Base URL: https://api.openai.com/v1

# Or via OpenRouter (200+ models):
API Key: sk-or-xxx
Base URL: https://openrouter.ai/api/v1

Note: When using an external API, data leaves the server. Use local Ollama for sensitive data.

Advanced Design Patterns

Chain Pattern

One model answers, then another model reviews and improves the answer.

Fan-Out Pattern

Same input sent to multiple models in parallel, then results are combined.

Fallback Pattern

If the local model fails or produces weak results, it automatically routes to an external API.

Performance Tips

• Allocate at least 4 GB RAM to Ollama — separate from n8n
• Use one model per workflow — switching models slows performance
• Set max_tokens — do not let the model generate without limits
• Enable caching — same question does not need reprocessing

Getting Started

n8n with AI on a Saudi server from CLOUD (cloud.net.sa) gives you intelligent automation power with no operation limits and no monthly subscriptions for tools like Zapier. The Enterprise plan (8 vCPU, 16 GB) runs n8n and Ollama together. The n8n template is ready for one-click deployment.

---

Get Started

Ready to run AI on your own server? Explore AI hosting solutions or configure your server with the specs you need.