A living read on the AI news cycle — mined every morning from tens of thousands of articles, with zero humans in the loop.
Reading the room…
0 = a typical day; +25 sunnier, −25 darker than usual (one standard deviation). Filled line is 7-day smoothed; dots are daily.
Weekly news mentions by company. Tap a chip to add or remove a player.
Average tone of each company's coverage vs the AI-news average, in points. Warm = sunnier press; cool = harsher.
Every line is a story that named two companies in the same breath. Thicker line = more shared coverage. Drag a node; hover (or tap) to spotlight its rivals.
Every archived article, bucketed by topic. Bigger tile = bigger share of the conversation.
Top outlets by AI article volume. A curated tag marks a hand-picked quality feed; the rest arrive via news alerts.
Headline words rising and cooling vs the prior week.
When AI news actually breaks. Article volume by day of week since March.
Articles collected per day since March. Darker = a heavier news day. Hover any cell.
An autonomous pipeline has collected AI news continuously since February 2026 — 63,000+ unique articles from 18+ Google News alerts plus curated feeds (MIT Technology Review, Wired, Ars Technica, The Verge, IEEE Spectrum, TechCrunch). Every morning it archives new articles, extracts company mentions, scores headline sentiment (lexicon-based, calibrated against Claude-labeled samples, aggregated daily, 7-day smoothed), rebuilds every chart on this page, and moves on. Duplicates and off-topic noise are excluded. Feed composition skews absolute counts — read trends, not single-week levels.
The same pipeline writes the Daily AI Brief — five stories a day, chosen and written without a human in the loop.
Every chart above updates itself each morning — no human touches it. If you've got busywork that should run itself like this, that's exactly my line of work.
See Agentic Automation Builds →