42AI comes out of stealth with $62.5M and a world model that can ‘predict human behavior’

Israeli startup 42AI has emerged from stealth with $62.5 million in seed funding, backing what it calls a "behavioral world model" — a system designed to predict what an individual will do, including in situations they haven't encountered before. The company is initially running the predictive system at a handful of financial brokerages, where it says the model can forecast how individual traders will respond to volatile market conditions.

CEO and co-founder Amit Gal-Or grounds the claim in probabilistic rather than deterministic behavior: "Behavior is not deterministic; it's constantly probabilistic." The premise is that people carry underlying psychological tendencies that shape their decisions even as circumstances change, and that these tendencies can be measured and converted into data a model can consume. Co-founder and CTO Ehud Ezra framed it as a stable signal beneath surface change: a week later a person may differ, but "your tendencies will always play out in certain areas to a probability or distribution."

Rather than extrapolating from a person's past actions, 24AI says the model infers internal motivations by observing reactions to seemingly unrelated events. For a trader or investor, those events include KYC completion, onboarding, support calls, money transfers, and stock trades. Ezra described the inputs broadly: sentiment and voice during early support or sales interactions, call frequency, later buy and sell timing, and market conditions at that stage — "we're trying to look at how they're looking at the world. In a world model, every individual is the world."

Technically, those observations become mathematical representations of traits, which feed a transformer-based world model that outputs probabilities for possible decisions under different circumstances. A separate language model generates explanations for the predictions. Training rewards are structured to preserve a person's stable tendencies during simulations — if a usually risk-averse individual is simulated as a risk-taker, Ezra said, "the target is off."

The company spent three years developing the approach, combining psychological research with neuroscience and genetics, plus advice from experts in psychology and PTSD and from Dan Shechtman, who won the 2011 Nobel Prize in Chemistry.

Open questions surround the rollout: how accurately the model generalizes beyond brokerage settings, what the brokerages actually do with its predictions, and whether claims about forecasting behavior in genuinely novel situations will hold up as the system meets more users and market conditions.