Gradient Descent is a free daily briefing at the intersection of AI, finance, and sports science — automated from source to inbox every morning.
Why it exists
I was spending 45 minutes every morning reading five different newsletters. Each one covered one slice of what I actually care about — AI tooling here, quant finance there, sports performance somewhere else. None of them talked to each other.
So I automated the whole thing.
The pipeline runs at 3:55 AM UTC every day. It fetches 80+ curated RSS feeds, scores each article against topic keyword lists, passes the best results to a single Groq API call, and distributes the briefing across email, this archive, and Telegram. Total runtime: ~15 seconds. Monthly cost: $0.
Who's behind it
I'm Pierluigi De Rogatis. I work as a business analyst building AI tools and automations. My background started in economics and finance — I still think in those terms — and sport has always run in parallel: I now build apps and AI systems for performance analysis.
Gradient Descent sits at that intersection. It's the briefing I'd want to read every morning: sharp enough for someone who works in data, useful enough for someone who thinks in finance, specific enough for someone who trains.
How it's built
The entire pipeline is open source. Every line — the RSS fetcher, the scoring logic, the Groq prompt, the GitHub Actions workflow — is public on GitHub.