AI Horizon Scanning

A fully automated tech watch pipeline, designed to never miss a relevant article about AI applied to development — without spending hours reading RSS feeds.

How it works

The pipeline runs on n8n (self-hosted via Docker) and executes automatically at regular intervals. It collects the latest articles from the Hacker News RSS feed, then submits each article to Claude (Anthropic) which performs three operations:

• ᚱEvaluates relevance to software development and AI
• ᚱGenerates a concise summary in French
• ᚱAssigns a score from 0 to 10

Only articles exceeding the score threshold are kept and exported to a local CSV file.

Why this project

I wanted to explore n8n as a workflow orchestrator and understand how to integrate an LLM API into a real data pipeline — not in a tutorial, but on an actual daily use case. The result: a tool I genuinely use for my own tech watch.

Planned improvements

• ᚱAdditional RSS sources (dev.to, arXiv, technical blogs)
• ᚱLightweight web interface to browse results
• ᚱDaily email notification with the filtered digest