About CLSTR
CLSTR groups articles from many news sources covering the same event into a single cluster: one record instead of fifteen near-identical takes, with every source linked.
When an event keeps developing across days or weeks, related clusters are linked into a situation: one page that carries the whole arc: what's verified, what's new versus recycled, and how the story got here.
That memory is the point. X tells you something happened. Google News resets every morning. CLSTR tells you what's happening, and can answer "what has actually happened here over the last six weeks" in ninety seconds.
What CLSTR is not
CLSTR is not a breaking-news product, on purpose. The pipeline ingests, clusters, verifies, and summarizes on a cycle measured in minutes to an hour. If you need second-by-second wires, use the wires. If you need the verified shape of a developing situation, and to never lose the thread, that's what CLSTR is for.
Who makes it
CLSTR is built and operated by one person. It exists because following a developing story meant reading the same article fifteen times across ten sites, and still losing the plot after a week away.
The pipeline clusters roughly 700,000 articles a week. Every piece of feedback gets read personally: if something's broken or missing, say so on the contact page and it gets fixed.
Corrections & publishers
CLSTR quotes short excerpts and links prominently to every original source. Publishers: for corrections, attribution issues, or removal requests, use the contact page. Requests are handled personally and quickly.