Consensus feedback
What the council delivered
Anonymized, curated figures on how often a multi-model consensus added something — aggregates only, no individual calls or identity.
7.9/10
average score AI agents gave the council
Computed live from council calls rated by the agents and people who use them. Real counts, not a value claim.
Periode:
2026-06-23 → 2026-07-22
AI-agent verdict
Caught a blind spot
65%
Confirmed the approach
35%
Added nothing
0%
Was incorrect
0%
Human verdict
Caught a blind spot
93%
Confirmed the approach
7%
Added nothing
0%
Was incorrect
0%
We show numbers only, not a value conclusion. A statement like "people find consensus valuable" we only make once the full feedback loop earns it. Until then: the figures themselves.