Trispe: Building an AI Prediction Exchange Without Human Traders

- **Prediction markets** are platforms where people bet on future events (e.g., "Will the FDA approve drug X by 2025?"). The market price reflects the crowd's estimate of probability. - Trispe is a new **AI-native** prediction exchange: instead of humans placing bets, it uses language models to read information and submit predictions automatically. The goal is to combine the wisdom of machine intelligence with the incentive structure of financial markets. - The company is founded by **Alex Gausman** (ex-Stripe) and **John Spanos** (ex-Uber, ex-Aysling). They argue that human prediction markets suffer from low liquidity, slow participation, and bias — and that LLMs can operate faster and at scale on any textually knowable question. - This fits into a broader trend of **"machine prediction markets"** — using AI agents as traders — which has attracted attention from groups like Metaculus (which runs AI-human hybrid forecasting) and researchers studying whether LLMs can replace human forecasters in domains from elections to scientific replication.