Israeli Startup 42AI Launches with $62.5M for Behavioral World Model
Israeli artificial intelligence startup 42AI has emerged from stealth with $62.5 million in seed funding to build a behavioral world model for predicting trader actions.
Israeli artificial intelligence startup 42AI has officially launched, backed by $62.5 million in seed funding. The company is focused on building a behavioral world model designed primarily for brokerages that want to anticipate how individual traders will behave during periods of market volatility.
Co-founder and Chief Executive Amit Gal-Or explained that human actions cannot be reduced to deterministic outcomes. Instead, the firm's software estimates the probability of various decisions a person might make under specific conditions. To achieve this, individual background data is translated into mathematical personality trait representations. A transformer-based world model then calculates the probabilities, while a separate language model generates written explanations detailing the psychological traits and context driving each forecast.
Forecasting Trading Behavior and Market Risk
The software allows brokerages to foresee what specific instruments clients might trade over a 24-hour window, along with the expected direction and volume. Risk desks can then use these insights to hedge exposures before large market movements hit their balance sheets. Inputs for the system include data gathered during onboarding know-your-customer checks, communication frequency, and voice sentiment during phone calls.
Currently, 10 large brokerages are collaborating with the startup, generating a seven-figure revenue stream through flat fees or performance percentages. According to company figures, one brokerage achieved roughly $30 million in added value via dynamic hedging. 42AI reports a 98% accuracy rate on its forecasts, which tracks whether trading flows land inside a predicted range rather than guessing every single trade correctly.
The underlying research spans three years. Initial psychological assessments involved thousands of participants, and sample sizes eventually surpassed 2 million when incorporating government and biobank datasets. Customer data currently covers roughly 30 million individuals.
Chief Technology Officer Ehud Ezra and his co-authors detailed part of their work in a February paper, showing how their Large Behavioral Model adapts an 8 billion-parameter version of Meta Platforms Inc.’s Llama 3.1 to forecast strategic decisions from psychometric profiles. Regarding privacy, Gal-Or noted that the architecture does not require personally identifiable information, stating that the model never knows actual identities even though individual profiles are constructed from linked account activity. Addressing potential risks of brokerages turning predictions against clients, Ezra acknowledged the concern and pointed to regulations, monitoring, and an internal ethics committee as safeguards.
What it means for developers
For developers building complex predictive systems, 42AI’s approach highlights the growing intersection of large language models—such as fine-tuned versions of open-weight systems like Llama 3.1—with specialized psychometric and behavioral datasets. Handling massive behavioral datasets while maintaining user anonymity presents architectural challenges around data processing and privacy preservation. Developers exploring advanced machine learning architectures and transformer-based world models can experiment with leading foundation models by visiting https://apixoai.online, where they can access top AI models cheaply through one API. As 42AI looks toward future expansion into gaming, fraud prevention, cybersecurity, and robotics, developers will likely see more demand for systems capable of modeling human intent and probabilistic decision-making.
Funding and Future Horizons
The $62.5 million funding round was provided by the founders and Infinity Group over a three-year period. The startup was co-founded by Amir Gal-Or, who also established Infinity Group, and counts 2011 Nobel Prize in Chemistry laureate Dan Shechtman among its research advisers. 42AI intends to seek additional capital as it explores new markets.
Source: 42AI launches with $62.5M to predict how traders act when markets turn volatile — SiliconANGLE AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
