42AI Exits Stealth With $62.5M to Predict Trader Behavior
42AI exits stealth with $62.5 million for a behavioral world model that predicts trader decisions, claiming 0.8 to 0.85 F1 scores at 10 large brokerages.
Summary
Israeli startup 42AI emerged from stealth on November 2, 2026, with $62.5 million in seed funding and a behavioral world model designed to calculate how individuals may act in unfamiliar situations. CEO Amit Gal-Or and CTO Ehud Ezra spent three years combining psychology, neuroscience and genetics with guidance from psychology and PTSD experts and 2011 chemistry Nobel laureate Dan Shechtman. The transformer-based model converts KYC, onboarding, support calls, voice, sentiment, transfers and trades into mathematical representations of personal tendencies, while a separate language model explains its predictions.
Training began with psychological assessments and behavioral scenarios involving thousands of people, followed by government and biobank datasets covering more than 2 million. Customer datasets now represent almost 30 million people. 42AI says it analyzes linked transactions and calls without personally identifiable information. It claims 98% accuracy for whether trading flows fall within predicted ranges, not individual trades, plus F1 scores of 0.8 to 0.85 for direction and volume. Ten large brokerages use the system for liquidity, margin requirements and hedging. Revenue exceeds seven figures, and Gal-Or says dynamic hedging created about $30 million for one customer. Pricing is either a flat fee or a share of generated value.
The founders and Infinity Group supplied the funding over three years, so it was not entirely external. Infinity founder Amir Gal-Or, Amit Gal-Or’s father, is 42AI’s chairman and co-founder. 42AI plans another raise and is exploring gaming, fraud prevention, cybersecurity, robots and AI agents, although each domain requires new rules and context. Missing information, such as advice from friends, limits accuracy. Regulations, monitoring and an internal ethics committee address potential misuse, but 42AI has not explained how brokerages are prevented from exploiting customer profiles.
Positives
- Ten large brokerages already use 42AI for liquidity management, margin requirements and hedging exposure.
- 0.8 to 0.85 F1 scores indicate consistent performance when predicting trading direction and volume.
- Almost 30 million people are represented across customer datasets, expanding the model beyond its initial research base.
- Approximately $30 million in additional value was generated for one brokerage through dynamic hedging, according to Amit Gal-Or.
- Seven-figure revenue and plans for another funding round show commercial traction beyond the $62.5 million seed financing.
Risks & concerns
- 98% accuracy measures whether trading flows remain within a predicted range, not whether individual trades are correctly forecast.
- Missing situational context, including advice from friends or colleagues, can materially reduce prediction accuracy.
- Linked customer activity still creates individual behavioral profiles despite 42AI saying personally identifiable information is unnecessary.
- 42AI has not explained what prevents brokerages from using behavioral forecasts against the customers being profiled.
- Gaming, fraud, cybersecurity, robotics and AI-agent applications each require additional domain-specific training, rules and context.
- $62.5 million came partly from the founders and Amir Gal-Or’s Infinity Group, rather than entirely from independent external investors.