AI-Powered Automated Tech Newsletter

An intelligent, AI-powered newsletter automation system built with n8n that collects, categorizes and distributes technology news from multiple RSS feeds.

Overview

This n8n workflow automatically:

• Fetches articles from technology RSS feeds
• Filters and deduplicates content
• Uses AI to categorize articles into 5 tech domains
• Ranks articles by relevance using keyword frequency
• Generates AI-powered summaries
• Creates a beautifully formatted HTML newsletter
• Stores data in CockroachDB and Airtable
• Sends weekly newsletters via Gmail

Architecture

Workflow Components

1. Schedule Trigger

• Triggers every Saturday at 7:00 AM
• Initiates the entire newsletter generation pipeline

2. RSS Feeds

Aggregates content from curated sources:

• AI/ML: TechCrunch, MIT News, Ars Technica, Wired, TechXplore
• Cloud Computing: AWS Blog, Google Cloud Blog, Cloud Computing News, CloudTweaks, Thomas Maurer
• IoT: Ars Technica IoT, IoT Now, IoT Analytics, IoT Tech News
• Automation: n8n Blog

3. Split Out RSS Items

• Converts the RSS feed array into individual items
• Enables parallel processing of each feed

4. RSS Read

• Fetches article content from each RSS feed
• Extracts: title, link, publication date, content snippet
• Handles XML/RSS parsing

5. Remove Duplicates

• Eliminates duplicate articles by comparing titles
• Ensures unique content in the newsletter

6. Filter

• Date Filter: Articles published after the last 4 days (pubDate > now - 4 days)
• Keyword Filter (OR logic):
  • n8n
  • AI / Artificial Intelligence
  • ML / Machine Learning
  • Cloud Computing
  • IoT / Internet of Things

7. Normalize Fields

Standardizes article data structure:

{
  "title": "Article Title",
  "contentSnippet": "Article preview text",
  "pubDate": "2024-12-01T10:00:00Z",
  "link": "https://example.com/article"
}

8. Information Extractor

AI-Powered Category Assignment

Uses Google Gemini LLM to intelligently categorize articles:

• Input: Article title + content snippet
• Output: Category from predefined enum
• Categories:
  • Artificial Intelligence
  • Machine Learning
  • Cloud Computing
  • Internet of Things
  • n8n / Automation

Synonym Mapping:

'ML' → 'Machine Learning'
'AI' → 'Artificial Intelligence'
'IoT' → 'Internet of Things'
'AWS', 'Azure', 'GCP' → 'Cloud Computing'

9. Merge Articles & Category

• Combines normalized articles with AI-assigned categories
• Input 1: Articles from Normalize Fields
• Input 2: Categories from Information Extractor
• Merge mode: Combine by position
• Ensures each article has complete metadata

10. Balance Categories

Smart Relevance-Based Selection

Selects the top 2 articles per category using a three-tier ranking system:

1. Primary: Keyword count in title (highest first)
2. Secondary: Keyword count in content (tie-breaker)
3. Tertiary: Publication date (newest first, final tie-breaker)

Logic - sort priority:

titleCountB - titleCountA → contentCountB - contentCountA → dateB - dateA

11. Aggregate All Articles

• Combines balanced articles into a single JSON object
• Prepares payload for LLM processing

12. Check if it's Saturday

• Condition: new Date().getDay().toString() equals "6"
• Gates the newsletter distribution
• Only proceeds on Saturday

13. Basic LLM Chain

AI-Powered Summary Generation

Uses Google Gemini Model to generate professional summaries for each article:

• Creates 3-sentence summaries in English
• Preserves original titles, links and category assignments
• Outputs structured JSON array with: title, summary, link, category_name
• Uses Structured Output Parser for JSON schema validation

14. Markdown

Generates the complete HTML newsletter from LLM output.

Features:

• Category-based color coding
• Emoji icons per category
• Responsive card layout
• Professional typography
• Adds header: Technology Newsletter
• Converts Markdown to HTML

Output: Fully styled HTML newsletter ready for email

15. Split Out LLM Output

• Splits the output field from LLM response
• Converts JSON array into individual article records
• Enables simultaneous storage for:
  • CockroachDB upsert
  • Airtable upsert

16. Add HTML Styles

Wraps the newsletter content in styled HTML container.

Each category has custom:

• Colors
• Icons
• Background style

Output: HTML rendered inside Gmail.

17. CockroachDB: Upsert Newsletter Table

Stores articles in PostgreSQL-compatible database.

18. Airtable: Upsert Record

Syncs articles to Airtable for easy management.

19. Send a message (Gmail)

Sends the newsletter via Gmail.

Configuration:

• Subject: "Technology Newsletter - [Current Date]"
  • Format: "December 2, 2025"
• Body: HTML content from "Add HTML Styles"

Trigger: Only executes if "Check if it's Saturday" = true

20. Gemini Model (LLM Provider)

• Provides Google Gemini AI capabilities
• Connected to both:
  • Information Extractor (for category assignment)
  • Basic LLM Chain (for summary generation)
• Model: gemini-2.5-flash

Newsletter Format

The generated newsletter includes:

• Header: " Technology Newsletter"
• Subtitle: "Your weekly tech digest"
• Categorized Sections: Articles grouped by topic with:
  • Category icon and name
  • Article title (clickable link)
  • 3-sentence AI-generated summary
  • "Read article" call-to-action
• Color-coded Design: Each category has unique colors

Key Features

• AI-Powered: Automatic categorization and summarization
• Smart Ranking: Relevance-based article selection
• Automated: Runs & sends weekly
• Professional Design: Beautiful HTML formatting
• Multi-Storage: CockroachDB + Airtable backup
• Deduplication: Ensures unique content
• Structured Data: JSON schema validation