Analysis
42AI, a Tel Aviv-based AI startup, emerged on November 2, 2026 from three years of stealth development with a $62.5 million seed round led by Infinity Group, TechCrunch reported. The round was raised over three years and, notably, was not entirely funded by outside investors.
What is 42AI's behavioral world model
42AI's product is what the company calls a "behavioral world model" -- a system that analyzes psychological tendencies and underlying motivations to predict how individuals will behave in new situations. It examines interactions across multiple touchpoints -- support calls, transactions, onboarding flows -- and converts those observations into mathematical trait representations fed into a transformer-based model that outputs decision probabilities. TechCrunch reported 42AI claims 98% accuracy predicting trading-flow ranges and F1 scores of 0.8 to 0.85 for direction and volume predictions, and says it has trained the system on datasets covering more than 2 million people sourced from government records and biobanks, with customer-facing datasets now expanding to nearly 30 million people.
Who founded 42AI
42AI was co-founded by CEO Amit Gal-Or, CTO Ehud Ezra, and chairman Amir Gal-Or -- Amit's father. Amir Gal-Or also founded Infinity Group, the firm that led this seed round, a related-party structure the company disclosed rather than concealed. The company intends to raise external capital next, according to TechCrunch.
“42AI was co-founded by CEO Amit Gal-Or, CTO Ehud Ezra, and chairman Amir Gal-Or -- Amit's father.”
42AI's customers and revenue
42AI says it's already working with 10 large brokerages and has generated more than seven figures in annual revenue applying its behavioral model to financial trading. The company is also pitching the technology for gaming, fraud prevention and cybersecurity, and says it sees future applications in robotics and AI agents that need to predict human behavior rather than just respond to language prompts.
Behavioral AI versus the language-model mainstream
42AI's bet is explicitly different from the dominant large-language-model race that companies like OpenAI, Anthropic and Google pursue. Where foundation-model labs compete on language generation and reasoning, 42AI is building narrower, behavior-specific prediction -- closer in spirit to quantitative trading models or fraud-scoring systems than to a chatbot. That positioning could insulate 42AI from foundation-model commoditization, since its value depends on proprietary behavioral datasets rather than raw model scale, a data-moat argument similar to one Pulse has covered in other specialized AI startups competing against general-purpose labs, including Lumaril's $66 million Series A.
Europe's regulatory exposure
42AI's profiling-based approach will face direct regulatory exposure in Europe, where it's also trying to sell. According to i-scoop.eu, even pseudonymized behavioral profiles count as personal data under GDPR, Article 22 protections apply to automated decisions with significant effects, and the EU AI Act prohibits manipulative techniques while separately classifying credit-related high-risk uses. 42AI's own co-founders describe their model's core assumption as "behavior is not deterministic; it's constantly probabilistic" -- a framing that works for a product pitch but won't exempt the company from proving to European regulators and MiFID II-bound financial firms that its probability scores don't cross into prohibited manipulation.