Ask a scope. Watch it reason.
Paste a message, a policy, or a plain question. Pick the United States, one of 35 US metros, or any country over 2M people. A calibrated synthetic electorate reasons it through person by person: you get the split, who leans which way, real voices in character, and when the live model runs, how different newsrooms would spin the same result.
Census ACS + World Bank demographics · 35 US metros · economy questions pull live FRED/BLS when configured · Claude reasons per resident · measured against real elections → · data stack →
Includes 35 US metros (city-level marginals, not full ACS microdata).
~12–220 synthetic residents reason live with Claude, bounded by a daily budget. Real
public data; a simulation, not a poll.