Interactive Demo

Experience the Scoring Engine

Watch 4 agents collaborate on a real task. Make choices that change which agents speak, who gets silenced, and what the team builds. Every score is computed from the same engine that runs in production.

Scoring decisions from Cohort's real engine. Agent responses generated live by qwen3.5:9b on local GPU -- real tokens, real timing, real scores.
View raw engine log (live run) · API scenario data · OAuth2 scenario data