Project Showcase
Scheduled AI Newsletter Digest
A fully automated newsletter system built with n8n and Google Gemini 2.5 Flash that runs every Monday morning without any human trigger, fetching live tech articles from RSS feeds, summarizing each one using AI, and delivering a beautifully formatted HTML digest email directly to your inbox.
Overview
Project Summary
[{"type":"paragraph","content":"This workflow is triggered automatically by a schedule node set to fire every Monday at 8:00 AM. It fetches the TechCrunch RSS feed via an HTTP Request node, parses the raw XML using a JavaScript Code node to extract the top 5 articles, and passes each one individually through a Gemini-powered LLM chain. The AI rewrites each article into a newsletter-ready format containing a catchy headline, a one-sentence TL;DR, a why-it-matters insight for developers, and a relevant emoji. An Aggregate node collects all five processed articles into a single item, which a final Code node transforms into a fully styled HTML email. Gmail delivers the finished newsletter automatically with no human involvement at any stage."}]
Architecture
Technical Implementation
1. Schedule Trigger Layer: A cron-based schedule node replaces the webhook entirely, firing the workflow automatically every Monday at 8:00 AM. Demonstrates fully autonomous workflow execution with zero human trigger required.
2. HTTP Request Layer: A GET request fetches the live TechCrunch RSS feed, returning raw XML data. No API key or authentication required. This demonstrates how to consume public data sources programmatically inside n8n.
3. XML Parsing Layer: A JavaScript Code node parses the raw RSS XML using regex pattern matching to extract the top 5 articles, pulling title, link, description, and publish date from each item. Returns multiple items, one per article, for downstream parallel processing.
4. AI Summarization Layer: A Basic LLM Chain node runs in "Run for Each Item" mode, sending each article individually to Google Gemini 2.5 Flash. A structured prompt instructs the AI to return four fields per article: a rewritten headline, TL;DR, why-it-matters insight, and emoji. Demonstrates per-item LLM processing in a loop.
5. Aggregation Layer: An Aggregate node collects all five individually processed articles into a single unified data object, enabling downstream nodes to work with the full dataset at once rather than item by item.
6. HTML Generation Layer: Another JavaScript Code node iterates over the aggregated articles and builds a fully styled HTML email with a dark blue header, today's date, styled article cards with blue left borders, and a footer. Outputs a single newsletter_html string ready for sending.
7. Gmail Delivery Layer: A Gmail node sends the finished HTML newsletter to the recipient using a dynamic subject line containing today's date. The entire email body is a single expression referencing the pre-built HTML from the previous node.