About The Flow

The Flow reads 19 AI and machine-learning sources — arXiv categories, first-party lab and university research blogs, and a few independent writers — and turns them into one searchable feed with a machine-written summary on every story.

Who runs it

The Flow is built and maintained by an independent developer. It is one person and a small server, not a media company. Reach them through the contact page; operator details are on the provider information page.

How the summaries are made

Each story's summary and its “why it matters” line are generated by a language model from the publisher's headline and feed description, and nothing else. They are labelled as machine-generated wherever they appear, are not reviewed by a person, and can be wrong. The original is always one click away, and reading it is the point.

What it deliberately does not do

How it is built

A nightly crawler reads the feeds; a local embedding model puts every article into a vector index, which is what makes search by meaning and saved searches that keep working possible; a small open model writes the summaries. It runs on one server and costs a few euros a month to operate.