42AI emerges from stealth with $62.5 million in seed funding

42AI says people have underlying psychological tendencies that influence decisions even as circumstances change.[1] It says its approach aims to infer internal motivations rather than simply extrapolate from past actions.[1] The system turns observations into mathematical representations of traits, which feed a transformer-based world model that predicts possible decisions in different circumstances.[1] A separate language model generates explanations for those predictions.[1] The company says the model is trained to preserve a person’s underlying tendencies during simulations.[1] Possible events include completing KYC, onboarding, support calls, transferring money or trading stocks.[1]

42AI says it spent three years developing its approach, drawing on psychological research, neuroscience and genetics, as well as advice from psychology and PTSD experts and Nobel laureate Dan Shechtman.[1] The team first studied thousands of people using psychological assessments and behavioral scenarios, then added government and biobank datasets covering more than 2 million people.[1] Gal-Or said customer datasets had expanded to cover almost 30 million people.[1]

Gal-Or said 42AI is working with 10 large brokerages, which use the system to manage liquidity, margin requirements and hedging exposure.[1] He said the model consistently achieves an F1 score of 0.8 to 0.85 when predicting trading direction and volume.[1] That figure should not be read as a rate of individually correct trade predictions: the metric measures whether trading flows fall within a predicted range.[1] 42AI says its brokerage product can predict each client’s trades up to 24 hours ahead by instrument, direction and volume.[2] Gal-Or said one brokerage generated about $30 million in additional value through dynamic hedging.[1]

42AI’s model has an accuracy ceiling because it cannot access everything a person experiences, and accuracy can suffer when it misses relevant context, such as a trader changing a decision after advice from a colleague or friend.[1] Gal-Or said the model never knows who a person is, while TechCrunch reported that 42AI appears to create individual behavioral profiles from linked customer activity.[1] The founders pointed to regulations, monitoring and an internal ethics committee as safeguards; they did not explain what would stop customers from exploiting predictions about people they profile.[1] 42AI is also exploring gaming, fraud prevention and cybersecurity, but says each new domain requires training on its rules and context.[1]