One event, three frames, a divergence score of 87. Here's how each side reported it, side by side.
Divergence pulls fresh articles from 45 outlets across left, center, right, and international, every 30 minutes. Sentence embeddings cluster articles describing the same event into one record. Then we score how sharply the framing, sentiment, and fact-inclusion diverge across the group.
No editorial board. No opinion added. The number above the story you just read, that 87, is what came out of the math.
45 outlets, RSS + API, 30-minute cadence.
Embeddings group articles about the same event.
Each article labeled by framing and sentiment.
Spread + delta + omission, normalized 0–100.
More stories where outlets are still figuring out the same facts.
Max plugs into a separate Markets desk: a live feed of headlines flagged as market-moving, with ticker tags, bullish or bearish signals per outlet, an AI summary of the session, and a watchlist with mini-charts beside the news.
The same feed is available as plain JSON over REST, or via the MCP server, so Claude, Cursor, or your own agent can read it directly and reason over today's coverage splits or your portfolio's exposure.
Hot CPI print pushed chip namesNVDA higher and weighed on EV exposureTSLA. Treasuries sold off as the curve repriced a later first cut. Coverage split was sharpest on the labor-market read.
GET /api/v2/markets?ticker=NVDAsame data, plain JSONThe Morning Split: today's three highest-divergence stories, the headlines that prove it, and a 60-second methodology note. Free, no account.