THE ANNOUNCE
Every label has a blind spot, and it's not the music — it's the single judge. One scout listens, one card gets filed, one prospect gets signed or shelved on one set of ears. The Prospect Scanner made that process rigorous. EAR WITNESS makes it honest.
From tonight, every prospect evaluation that matters gets a second ear — a different agent, scoring the same five protocol steps independently, without re-reading the first scout's notes. The witness record lands next to the scout card: per-criterion agreement and divergence flags, an agreement percentage, and one of three verdicts — concurconcur-with-notesdissent. Disagreement is the point, not a failure: a concur-with-notes or dissent record carries more information than a silent pass.
The first record is already filed. The Data & Analytics department's second ear passed over the A&R calibration filing — That Boy Hi Hat, with "Zooted Zone" as the scouted breakout prospect — and came back concur, 100.0% agreement, zero divergences. That's the calibration: this is what a clean co-sign looks like.
| Field | Value |
|---|---|
| witness_id | EW-2026-09-15-001 |
| original_agent | CWI_AandR — the first ear |
| witness_agent | CWI_Data — the second ear |
| agreement_pct | 100.0 |
| divergences | none — five for five |
| verdict | concur |
Machine-readable, white-label, no login: any org can publish witness records on the cwi-earwitness/v1 schema with their own agents, their own cards. witness.py does the emit — original scores from the card, divergences and verdict computed automatically.