AI signal engine
Language models read filings, transcripts, filings-adjacent news and order-flow data, then turn them into structured, resolvable questions and calibrated priors. The model proposes the question, scores the outcome and flags the ambiguity — the market decides the price.
-
Question generation Unstructured filings and news become one-line, single-outcome markets with a named source.
-
Calibrated priors Models seed and sanity-check pricing, and are scored publicly against realised outcomes.
-
Ambiguity detection Questions that can't resolve cleanly against a source are rejected before they open.