Israeli startup 42AI has emerged from stealth with $62.5 million in seed funding for a transformer-based behavioral model designed to predict human decision-making in volatile conditions. The company is piloting its system at financial brokerages to forecast how individual traders will respond to market swings.
Unlike traditional predictive models that extrapolate from historical patterns, 42AI's approach attempts to isolate and quantify underlying psychological tendencies—the consistent probability distributions that govern how a specific person makes choices, even as circumstances change.
"Behavior is not deterministic but probabilistic," said CEO and co-founder Amit Gal-Or. "People have core psychological tendencies that influence their decisions regardless of context. Our system converts those tendencies into data the model can process."
The company's methodology ingests behavioral signals from multiple touchpoints: KYC completion, onboarding flows, support interactions, money transfers, and trading execution. During early customer interactions, the model captures sentiment, voice patterns, and engagement frequency. Later, it layers in buy-sell behavior and concurrent market conditions.
42AI's engineers convert these observations into mathematical trait representations fed into the transformer model, which generates decision probabilities under different scenarios. A separate language model supplies explanations for each prediction. The training architecture penalizes predictions that contradict a person's baseline risk profile—if the model misclassifies a consistently risk-averse trader as risk-tolerant, the prediction is marked inaccurate.
CTO and co-founder Ehud Ezra noted that while individuals evolve over time, their behavioral tendencies manifest within stable probability distributions. "The model shows the likelihood a person makes a particular decision," he said.
The company spent three years developing this framework, integrating psychological research with neuroscience principles. The seed size reflects investor confidence in applying behavioral modeling to financial services—a sector where predicting trader response to stress can directly affect compliance, risk management, and customer retention.
