MU
← All Projects3.3
N8N Automated Marketing Pipeline
AI-PoweredAgentic Systems
A 15-node n8n workflow that takes a website URL and produces a full marketing analysis report delivered via email. Webhook trigger kicks off website scraping, content extraction, then parallel AI analysis branches (brand voice, audience segmentation, marketing plan) that merge into a cost-calculated, AI-written report sent through Mailjet. All 4 AI nodes use Code nodes calling OpenAI API directly for full prompt control.
Technology Stack
System Architecture
Loading diagram...
Specifications
- ▸15-Node Pipeline: Webhook → Scrape → Extract → AI Analysis → Merge → Cost Calc → Write Report → Email
- ▸Parallel AI Branches: Brand analysis, audience segmentation, and marketing plan run in parallel then merge
- ▸Code Nodes for AI: All AI calls use Code nodes with httpRequest — not langchain nodes — for full prompt control
- ▸Mailjet HTTP API: Email delivery via Mailjet REST API — SMTP times out from Docker/WSL2 containers
- ▸Webhook Trigger: POST a URL, get a complete marketing analysis report in your inbox
- ▸Cost Tracking: Calculates and reports OpenAI token usage and API spend per analysis
Design Decisions
| Challenge | Solution |
|---|---|
| Langchain nodes had expression resolution bugs | Switched all 4 AI nodes to Code nodes calling OpenAI API via httpRequest |
| SMTP times out from Docker/WSL2 | Mailjet HTTP API for reliable email delivery |
| Parallel branches need to reconverge | Merge node in append mode between parallel AI branches |
| Need full prompt control with dynamic data | JavaScript template literals in Code nodes for prompt construction |
| Debugging 15-node pipeline is painful | Test nodes individually via n8n API before running full pipeline |
Process Flow
Loading diagram...