42AI exits stealth with $62.5M for a behavioral world model

How the model works

To infer motivations, 42AI studies how people react to seemingly unrelated events, such as KYC checks, onboarding, support calls, money transfers and trades. It looks at sentiment, voice and call frequency, then converts the observations into mathematical representations of traits. These feed a transformer-based world model that lays out probabilities for possible decisions under different conditions. A separate language model writes the explanations for predictions.

The company says it spent three years on the approach, drawing on psychology, neuroscience and genetics. It started with thousands of people assessed in studies, added government and biobank data covering more than 2 million people, and says customer datasets now cover almost 30 million. Gal-Or says the system needs no personally identifiable information, though it still builds individual behavioral profiles from linked customer activity.

Claims and limits

42AI claims 98% accuracy in predicting market trades, but Gal-Or clarified that this measures whether trading flows land within a predicted range, not whether each trade was called correctly. He also cites an F1 score of 0.8 to 0.85 for trading direction and volume. Accuracy suffers when the model misses context, such as a trader acting on a friend's advice, and each new domain requires training on its own rules.

The company says it works with 10 large brokerages for liquidity, margin and hedging management, and earns more than seven figures in revenue. Customers pay a flat fee or a share of the value gained; one brokerage reportedly gained about $30 million through dynamic hedging. The seed money came from the founders and Infinity Group over three years, not entirely from outside investors. Infinity was founded by Amir Gal-Or, the chairman and a co-founder of 42AI and the CEO's father.

The startup plans to raise more and is exploring gaming, fraud prevention and cybersecurity. The obvious concern is that brokerages could use such predictions against the customers being profiled. Ezra called it a valid concern, and the founders point to regulation, monitoring and an internal ethics committee, without explaining what would stop customers from exploiting the insight.