Israeli startup 42AI has emerged from stealth with $62.5 million in seed funding and a system it calls a "behavioral world model." The technology aims to estimate how people may respond to different situations, initially helping financial firms anticipate trading activity.
Rather than relying only on past actions, the model analyzes patterns across customer interactions, such as transactions and support calls, to represent behavioral tendencies. It then uses a transformer-based model to estimate the probability of decisions under changing conditions. The company says its system does not require personally identifying information, although it builds behavioral profiles from linked customer activity.
42AI reports that its model is being used by 10 large brokerages for liquidity management, margin requirements and hedging. The company says it achieves a 0.8-0.85 F1 score when forecasting trading direction and volume. Its separate 98% accuracy figure refers to whether trading flows fall within a predicted range, not whether individual trades are called correctly.
Developed over three years, the system draws on psychological research, neuroscience and genetics. 42AI says its training data includes assessments and behavioral scenarios, as well as government and biobank datasets. It is also exploring potential uses in gaming, fraud prevention and cybersecurity, while noting that each field requires domain-specific training.
By turning behavioral tendencies into probability estimates, this approach could give organizations new ways to plan for changing human decisions. Its broader impact will depend on how effectively the technology adapts to context and is applied across different industries.
