Israeli startup 42AI came out of stealth on November 2 with $62.5 million in seed funding and a “behavioral world model” that predicts the probability of what individual people will do next — starting with traders at financial brokerages. Why it matters: most AI models predict markets, text or video. This one tries to predict people, and it is already running live at paying customers.
November 2, 2026 — AI News
What happened
42AI says its system estimates the probability that a specific person will make a particular decision, even in a situation that person has never faced before. Instead of extrapolating from someone’s past trades alone, the model studies how people react to everyday events — onboarding steps, support calls, money transfers, buying and selling — and converts those observations into mathematical representations of underlying traits. Those traits feed a transformer-based world model that lays out probabilities for possible decisions, while a separate language model generates explanations for the predictions.
The company spent three years developing the approach, combining psychological research with neuroscience and genetics, and taking advice from experts in psychology and PTSD as well as Dan Shechtman, who won the 2011 Nobel Prize in Chemistry. CEO and co-founder Amit Gal-Or told TechCrunch the team first studied thousands of people through psychological assessments and behavioral scenarios, then added government and biobank datasets covering more than 2 million people; customer datasets have since expanded to cover almost 30 million people, he said.
The details
Gal-Or said 42AI is working with 10 large brokerages, which use the model to manage liquidity, margin requirements and hedging exposure, and that more customers are onboarding. The company is generating more than seven figures in revenue, though he would not give a precise number. Customers pay either a flat fee or a percentage of the value the platform helps them gain; Gal-Or claimed one brokerage generated approximately $30 million in additional value through dynamic hedging.
On accuracy, the company claims 98% — but Gal-Or clarified that this measures whether trading flows fall within a predicted range, not whether each individual trade was predicted correctly. He also cited an F1 score of 0.8 to 0.85 for predicting trading direction and volume. Both figures are the company’s own and have not been independently verified. The founders were also candid about limits: the model cannot see everything a person experiences, and its accuracy drops when it misses situational context, such as a trader acting on a colleague’s advice.
One important caveat on the headline number: the $62.5 million was not a conventional outside seed round. Gal-Or said the money came from the founders and Infinity Group over three years and was not raised entirely from external investors. Infinity was founded by his father, Amir Gal-Or, who is also 42AI’s chairman and a co-founder. CTO Ehud Ezra is the third co-founder. The company now plans to raise additional money.
Why it matters
The obvious question is what stops a brokerage from using predictions about its own customers against them. Ezra acknowledged that is “a valid concern” and pointed to regulations, monitoring and an internal ethics committee, without explaining what would actually prevent misuse. Gal-Or also argues the system does not need personally identifiable information — “the model never knows who the person is” — yet it still builds individual behavioral profiles from linked customer activity, which is exactly the tension regulators tend to focus on.
Our take: treat the accuracy claims as marketing until an independent evaluation appears, and note the family-funded structure behind the $62.5 million. But a behavior-prediction model already deployed at 10 large brokerages, with paying customers and expansion plans into gaming, fraud prevention and cybersecurity, is a real business — and a preview of where “world models” go next: from simulating physics to simulating people. Everything AI, tested — and claims like these deserve testing most of all.
Frequently asked questions
What is 42AI’s behavioral world model?
It is a transformer-based model that converts observations of how a person behaves — transactions, support interactions, onboarding steps — into trait representations, then outputs probabilities for what that person is likely to decide under different circumstances.
How much funding has 42AI raised?
The company exited stealth with $62.5 million described as seed funding, but CEO Amit Gal-Or said the money came from the founders and Infinity Group over three years and was not raised entirely from external investors.
Who is using 42AI today?
According to the company, 10 large financial brokerages use it for liquidity management, margin requirements and hedging exposure. It is exploring gaming, fraud prevention and cybersecurity next, though each new domain requires additional training.
How accurate is 42AI’s model?
The company claims 98% accuracy, clarified as trading flows falling within a predicted range, and an F1 score of 0.8 to 0.85 for trading direction and volume. These are company-reported figures and have not been independently verified.
Sources: TechCrunch, Unite.AI