A developer known as dudeson55 shares a detailed breakdown of an AI-powered newsletter system built entirely with n8n workflows that has grown to 10,000 subscribers. The system autonomously scrapes nearly 100 AI news stories daily from sources including Twitter, Reddit, HackerNews, and various AI blogs using RSS.app feeds and Firecrawl for content extraction. Stories are saved as markdown files in an S3 data lake, then processed through a sophisticated multi-stage workflow that selects top stories, writes newsletter sections, and formats the final output.

The newsletter generation workflow features several notable design decisions: an "AI Editor" prompt selects the top 3-4 stories, a human-in-the-loop approval step via Slack allows the author to override selections in plain English, and stories are written individually in a loop rather than in a single prompt to improve output quality. The system also attempts to scrape primary source materials (e.g., an OpenAI blog post referenced in a TechCrunch article) to build richer context for writing. After five months of prompt iteration, the system handles 95% of the newsletter creation process, achieving an open rate above 20%. The author provides the complete n8n workflow JSON files on GitHub and a YouTube walkthrough for others to replicate the approach.