42AI launches from stealth with $62.5 million seed round and a behavioral world model for finance

42AI emerged from stealth this week after closing a $62.5 million seed round, announcing a “behavioral world model” that it says can forecast individual decisions in volatile market conditions. The company’s first customers are a group of large financial brokerages that are using the technology to anticipate how traders will react to price swings, liquidity pressures and margin calls.

The model differs from conventional predictive tools by focusing on internal motivations rather than simply extrapolating from past behavior. Co-founder and CTO Ehud Ezra explained that the system ingests a wide array of signals - from sentiment in support calls to the frequency of account interactions - and translates them into mathematical representations of traits. Those traits feed a transformer-based architecture that outputs probability distributions for possible actions under given scenarios. A separate language model then generates natural-language explanations for each prediction.

In live tests the startup reports a 98 percent accuracy rate for predicting whether trading flows will stay within a projected range, and an F1 score between 0.8 and 0.85 for direction and volume forecasts. The company says it has built behavioral profiles for roughly 30 million individuals using anonymized transaction and interaction data, while also drawing on public biobank datasets covering more than two million people. Although the system does not store personally identifiable information, it does create linked profiles that capture a user’s observed tendencies.

42AI’s revenue comes from flat fees or a share of the incremental value the platform generates for its clients. One brokerage disclosed that dynamic hedging informed by the model added about $30 million in value. The firm is currently working with ten major brokerages on liquidity management, margin requirements and hedging exposure, and it is generating seven-figure revenue, according to CEO Amit Gal-Or.

The seed round was led by the founders and Infinity Group, a vehicle founded by Gal-Or’s father, Amir Gal-Or, who also serves as chairman. While the round was largely internal, the company plans to raise additional capital to expand beyond finance. Future target markets include gaming, fraud detection and cybersecurity, and the founders envision a longer-term goal of enabling robots and autonomous agents to better understand human behavior. Ezra cautioned that each new domain will require bespoke training on relevant rules and context.

The technology raises ethical questions about the potential misuse of predictive insight. Ezra acknowledged the risk that a brokerage could exploit knowledge of a client’s likely actions, noting that the firm relies on regulatory oversight, internal monitoring and an ethics committee to mitigate abuse. Gal-Or expressed confidence that 42AI’s approach can be part of the solution to such concerns.

If the model can reliably capture probabilistic human tendencies, it could become a valuable tool for firms that need to manage risk in real time. However, its performance hinges on the availability of rich interaction data and the ability to incorporate situational factors that lie outside the system’s current view. Investors and founders will be watching closely to see whether 42AI can translate its ambitious scientific premise into sustainable commercial impact across multiple industries.