A transformer-based world model analyzes psychological tendencies from diverse customer interactions—onboarding, KYC, support, money transfers, and trading activity—to forecast decision probabilities.

42AI has raised a seed round totaling $62.5 million over three years from the founders and Infinity Group, with Infinity founded by Amir Gal‑Or and serving as chairman; the founders plan to pursue additional funding.

The company presents its approach as addressing variability in human behavior rather than merely generating fluent AI responses, noting that ongoing evaluation is needed to prove usefulness outside controlled studies.

Ethical and regulatory questions surround using predictive behavioral insights for financial advantage; an internal ethics committee and regulatory considerations are in place, but safeguards are not fully detailed.

Live deployments claim high accuracy on trading flows within predicted ranges and F1 scores around 0.8–0.85 for direction and volume, while noting accuracy drops if contextual data is missing.

Beyond finance, the company is exploring applications in gaming, fraud prevention, cybersecurity, and potentially teaching robots and AI agents to understand people, with plans to raise more capital for broader deployments.

42AI, an Israeli startup, exits stealth with $62.5 million seed funding to build a behavioral world model that predicts likely decisions in new situations, beginning with financial brokerages.

Development draws on psychology, neuroscience, and genetics, using large datasets from government and biobank sources; current customer data approaches 30 million individuals.

A February 2026 preprint discusses a Large Behavioral Model fine-tuned from Llama 3.1-8B-Instruct, illustrating a path from behavioral evidence to predictions.

A separate language model explains predictions, and behavior is treated as probabilistic, with dispositions expressed as probability distributions even if individuals change week to week.

In live use, about ten large brokerages utilize the platform for liquidity management, margin requirements, and hedging, generating seven-figure revenues and adding value through dynamic hedging.

The technology could influence AI assistants and enterprise decisions about customers or employees, requiring consent, use restrictions, and mechanisms to challenge predictions.