42AI 以 6250 万美元种子轮启动，用行为世界模型预测交易者决策

Israeli artificial intelligence startup 42AI launched today with $62.5 million in funding to build what it calls a behavioral world model.

Its first customers are brokerages that want to know how individual traders will act when markets turn volatile. 42AI offers software that estimates the odds of each choice a person might make in a given situation. “Behavior is not deterministic; it’s constantly probabilistic,” co-founder and Chief Executive Amit Gal-Or told TechCrunch.

The company turns what it knows about a person into mathematical representations of personality traits, and a transformer-based world model works out the probabilities from there. A separate language model writes up which traits and context drove each prediction.

42AI says a brokerage can see what each client is likely to trade over the next 24 hours, including the instrument, the direction and the volume. If clients look set to pile into the same trade, a risk desk can hedge before the exposure lands on its book. Know-your-customer checks done at onboarding are one source of input, and the model also weighs how often a client calls and the sentiment in their voice.

Ten large brokerages are working with 42AI, which charges either a flat fee or a percentage of the value it helps generate. Revenue has reached seven figures, Gal-Or said, without giving a precise number.

By the company’s account, one brokerage gained about $30 million in additional value through dynamic hedging. The company also claims 98% accuracy on trades. That figure tracks whether trading flows land inside a predicted range, according to Gal-Or, and does not measure whether every individual trade was predicted correctly.

Three years of research went into the model. Thousands of people took psychological assessments in the first study, and by the time the team moved on to government and biobank datasets, the sample had passed 2 million. Customer data now spans about 30 million people.

Chief Technology Officer Ehud Ezra and three co-authors described one piece of the work in a paper posted in February. Their Large Behavioral Model adapts an 8 billion-parameter version of Meta Platforms Inc.’s Llama 3.1 to predict a person’s strategic choices from a detailed psychometric profile.

On privacy, Gal-Or said the system does not need personally identifiable information and that “the model never knows who the person is.” Profiles are still built for each individual from linked account activity.

Asked whether a brokerage could turn those predictions against its own clients, Ezra called it “a valid concern.” The founders pointed to regulation, monitoring and an internal ethics committee as checks on that kind of use.

Gaming, fraud prevention and cybersecurity are the next markets 42AI is exploring, with helping robots and AI agents understand people as a longer-term goal. Wherever it runs, the model loses accuracy when it misses outside influences, such as a trader acting on a colleague’s advice. “There’s a natural ceiling because you really don’t have access to everything that a person experiences,” Gal-Or said.

The $62.5 million, which 42AI describes as seed funding, came from the founders and Infinity Group over three years. Amir Gal-Or, the startup’s co-founder and chairman, founded Infinity Group. Dan Shechtman, who won the 2011 Nobel Prize in Chemistry, is among the advisers to the research team. 42AI plans to raise more money.

Image: 42AI