Ten AI stories every morning at 08:00 KST, in Korean and English, from twelve feeds. Each entry has a summary and a short note on why it matters, written from the article rather than the headline. The pipeline is automated. Daily intake and publication figures are public at /data, and when the selection logic fails I publish the numbers and the code before fixing it. Recent posts cover a scoring ceiling that excluded an entire language, and a per-outlet cap that switched itself off.