How to Build an AI News Pipeline with n8n and OpenRouter: Step-by-Step

An automated AI news pipeline does what a manual editorial process can’t: it monitors dozens of sources simultaneously, processes new content as it appears, and publishes structured posts to WordPress continuously — without human intervention for routine items.

What You’ll Build

A pipeline monitoring 5–20 RSS feeds simultaneously
Deduplication preventing the same story appearing multiple times
AI processing applying the Signal→Meaning→Use Cases→Opportunities framework
Automatic category and tag assignment
Impact level classification (High / Medium / Watch List)
Automated WordPress post creation in Draft status

Prerequisites

n8n instance (self-hosted or cloud)
OpenRouter API key — free at openrouter.ai
WordPress site with REST API enabled
Google Sheets for deduplication tracking

Stage 1: RSS Collection

Use a Google Sheets Trigger node reading a list of RSS URLs from a sheet. Loop through each URL with a RSS Feed Read node. This lets you add or remove sources without touching the workflow — just update the spreadsheet. Extract per item: title, URL, description, publication date, and source name.

Stage 2: Deduplication

Before processing any item, check whether its URL already exists in your Google Sheets tracking log. If yes: skip. If no: add to log and continue. This prevents the same story — appearing in multiple RSS feeds — from being processed and published multiple times.

Stage 3: AI Processing via OpenRouter

Send each deduplicated item to an HTTP Request node pointing to the OpenRouter API. The system prompt instructs the model to return structured JSON:

{
  "signal": "What happened — one clear sentence",
  "meaning": "Why this matters — 2-3 sentences",
  "use_cases": ["use case 1", "use case 2"],
  "opportunities": ["opportunity 1", "opportunity 2"],
  "category": "ai-news|ai-tools|ai-automation|ai-opportunities",
  "subcategory": "models|big-tech|open-source|writing|...",
  "tags": ["tag1", "tag2"],
  "impact_level": "high|medium|watch-list",
  "seo_title": "SEO title under 60 characters",
  "meta_description": "Meta description under 160 characters"
}

Stage 4: WordPress Publishing

Use a Code node (JavaScript) to construct the full Gutenberg-formatted post body, then the WordPress node to create the post with: title, content, categories, tags, meta description, and status set to draft for human review before publishing.