Founder, Sage Voice. Builds Chorus, the calibrated-forecasting holdco that owns the underlying methodology for the audit-log primitive. Three prior operator exits. Adology + PolicyChat equity. Currently Chorus founder, Mexico-resident.
Chorus runs a public, SHA-stamped prediction ledger of resolved forecasts across politics, sports, commodities, macro, and earnings. Aggregate Brier 0.1601 across n=61 forward predictions (validated 2026-05-06); 0.0399 across n=18 high-conviction in-headline predictions (|p−0.5| > 0.20). On the Nielsen Sports pilot: 5-of-5 resolved 4/28–5/18 predictions hit, mean Brier 0.0592.
Ledger: chorus.markets/ledger. Methodology and 8-gate validation framework documented in the Chorus public stack.
At PolicyChat (the consumer insurance chat that PolicyChat-CO Inc. operates), I watched the same pattern hit every regulated voice deployment: the call worked, the script worked, the underwriter signed off — and then the audit fell apart because nothing captured the disclosure timing or the alternatives considered. The conversation was great. The documentation was missing.
Sage Voice is the answer: the conversation IS the documentation. Compliance enforced at the API boundary, not in the prompt. Audit log as the system of record. Same architecture pattern that Chorus uses for forecast provenance (SHA-anchored ledger), adapted for the regulated-voice industry.