AI Cronicle
An automated newsroom in a vintage five-column newspaper layout, filled from RSS and Atom sources, where every article is run through an AI pipeline.
The challenge
Turn a stream of RSS and Atom sources into something readable: every article needs a summary, a category and a relevance score.
What we built
A React 19 and Vite front end with an Express API, PostgreSQL through Prisma 6, and BullMQ workers that run each article through an AI pipeline (Gemini, OpenAI or Anthropic, with a local analytical fallback).
Key capabilities
- Automated ingestion from RSS and Atom sources
- Per-article summary, category and relevance score
- A "why it matters" note for every article
- Multiple AI providers with a local fallback
- Vintage five-column newspaper layout
Technology
React 19
Vite 6
Node.js
Express 4
PostgreSQL
Prisma ORM 6
BullMQ