The clustering, trust, and claim-consensus layer
for live events.

We watch live streams across YouTube, Twitch, and Kick, cluster them by real-world event, score every source by tier and reliability, and surface claim consensus and contradictions — as a single REST API.

Get a free dev key Try the playground Read the docs →

One call, full enrichment

Every event comes back with cluster size, source tier breakdown, verified-source count, severity, regional context, and a confidence score. No second call to enrich; no joining tables in your dashboard.

curl https://api.livetube.ai/v1/events/flagship \
  -H "Authorization: Bearer YOUR_KEY"

What you get

Multi-source clustering

Ten chasers covering the same tornado outbreak come back as one event with ten sources, not ten redundant items.

Source-tier trust scoring

Three-tier source model (Major outlet / Verified / Default) plus per-channel reliability score updated by claim outcomes.

Claim consensus + contradictions

Claims are graded by how many independent sources confirm them. Contradictions are surfaced, not hidden.

Read the methodology for the full framework.