Israeli startup 42AI has emerged from stealth with $62.5 million in seed funding and a system it describes as a “behavioral world model.” The company says its technology can estimate the probability that an individual will make a particular decision, even in situations that differ from their previous experiences.
42AI is initially deploying the system at financial brokerages, where it is intended to help predict how traders may respond to volatile market conditions. The model analyzes customer records and interactions, including onboarding, KYC processes, support calls, money transfers and trading activity.
Rather than relying only on past actions, the company says it attempts to identify underlying psychological tendencies. It converts observations into mathematical representations of traits, which are then processed by a transformer-based model to produce probabilities for possible decisions in different circumstances. A separate language model generates explanations for the predictions.
According to co-founder and CTO Ehud Ezra, the system is trained to preserve a person’s underlying tendencies during simulations. For example, if a person is generally risk-averse, the model is designed to treat a sudden simulated shift toward risk-taking as an inaccurate result unless the available evidence supports it.
42AI says it spent three years developing the approach, drawing on psychological research, neuroscience and genetics, along with advice from psychology and PTSD specialists and Nobel Prize-winning chemist Dan Shechtman. CEO and co-founder Amit Gal-Or said the company initially studied thousands of people through psychological assessments and behavioral scenarios, later adding government and biobank datasets covering more than 2 million people. Customer datasets have since grown to almost 30 million people, he said.
The company says its system does not require personally identifiable information. However, its operation still involves creating individual behavioral profiles from linked customer activity, including transactions and support interactions.
In market-trading predictions, 42AI claims 98% accuracy, although Gal-Or clarified that this figure refers to whether trading flows fall within a predicted range rather than whether each individual trade is forecast correctly. He also said the model achieves an F1 score between 0.8 and 0.85 when predicting trading direction and volume.
The company acknowledges that the model has limitations because it cannot access every factor affecting a person’s decisions. Accuracy can decline when important context is missing, such as a trader changing strategy after receiving advice from a friend or colleague.
Gal-Or said 42AI is working with 10 large brokerages on liquidity management, margin requirements and hedging exposure. The company is onboarding additional customers and generates more than “seven figures” in revenue, although it did not provide a specific amount. Customers pay either a flat fee or a percentage of the value the platform helps create. Gal-Or said one brokerage generated approximately $30 million in additional value through dynamic hedging.
The $62.5 million seed financing was provided over three years by the founders and Infinity Group, rather than being raised entirely from external investors. Infinity was founded by Gal-Or’s father, Amir Gal-Or, who is also 42AI’s chairman and a co-founder. The startup plans to raise additional funding.
Beyond financial services, 42AI is exploring potential uses in gaming, fraud prevention and cybersecurity. It also hopes eventually to help robots and AI agents understand people, although the company says the model must be trained on the specific rules and context of each new field.
The technology also raises concerns about how customers could use predictions of people’s behavior to gain an advantage over those being profiled. Ezra described that concern as valid, while the founders pointed to regulations, monitoring and an internal ethics committee. They did not explain what specific measures would prevent customers from exploiting the system’s predictions.
Sources: TechCrunch Startups.