How Trends works
Trending doesn't mean true. Every story carries two independent scores: Trend measures attention, and Confidence measures how much of the story the evidence supports. A viral rumor can score high on one and low on the other.
What we track
- XLive
- GitHubLive
- Hacker NewsLive
- Official SourcesLive
- RedditNot connected
- LinkedInNot connected
- YouTubeNot connected
- Hugging FaceNot connected
From signal to story
- 01
Collect
We read posts, threads, repositories, and official announcements from the sources below. Low-engagement posts and reposts are dropped.
- 02
Group
Posts about the same event are grouped into one story. Unrelated news that happens to share a keyword stays separate.
- 03
Write
A model drafts the story and breaks it into individual factual claims. It interprets; it does not decide what is true.
- 04
Check
Each claim is tested against the evidence. A quote only counts if it appears word for word in the cited source. Unsupported claims are removed or reworded.
- 05
Publish
Only stories with real attention make the feed. A rumor needs to be genuinely viral to appear, and stories retire once the conversation goes quiet.
How to read Trends
- Trend Score
- Trend - how fast attention is growing: velocity, spread, engagement, source reach, freshness. It never encodes truth.
- Confidence Score
- Confidence - how much of the story the evidence supports, from claim-level JEV verification against primary sources.
Latest pipeline run
- No pipeline activity yet.
Apify finds the signals · Claude turns noise into intelligence · JEV verifies the claims.