Hello, World

I subscribe to too many tech newsletters. You probably do too.

Tech news is like a firehose.

Much of the content I never get around to reading. I wanted to see what kept coming up across sources without having to read all of them. So I built a feed that ingests newsletters and tech communities, dedupes the links, and surfaces what’s worth reading by mention frequency.

This is an experiment to test scraping, ingestion, tagging taxonomy, and AI integrations to inform future work and concurrent projects.

It also doubles as a portfolio. A product as a portfolio, yes, I know. Felt less embarrassing than a slide deck.

Stack: Elixir/Phoenix, Node scraper pipeline, Cloudflare Email Workers, Claude for story extraction, self-hosted via Dokploy on Hetzner.

Why Elixir?

Currenly drawn to functional programming languages.

Worked well for an earilier project and AIs seem to have a greater affinity for functional programming. Phoenix LiveView, Ecto, NimblePublisher fit together cleanly without fighting the framework. The BEAM’s fault tolerance is overkill for a personal site. The developer ergonomics aren’t.

What’s next

Newsletter ingestion is close: Cloudflare Email Worker relays to the server, Claude extracts the stories. Next is the weekly digest, surfacing whatever showed up in three or more newsletters that week.