42AI exits stealth with AI it says predicts brokerage customers’ trading decisions

42AI says 10 large brokerages are using its AI to anticipate customers’ trading activity and manage financial risk. The Israeli startup emerged from stealth on November 2, 2026, disclosing $62.5 million accumulated over three years—not a newly closed round—and pitching a model that estimates how individuals will act in different circumstances.

CEO and co-founder Amit Gal-Or told TechCrunch the money came from the founders and Infinity Group, rather than entirely from outside investors. Infinity was founded by his father, Amir Gal-Or, who is also 42AI’s chairman and a co-founder. The startup now intends to raise more money.

Customers pay either a flat fee or a share of the value the platform helps them gain. Gal-Or claimed one brokerage generated approximately $30 million in additional value through dynamic hedging—adjusting financial positions to manage market exposure.

From support calls to behavioral profiles

42AI describes its system as a behavioral world model. Rather than only extending patterns in past trades, it tries to identify psychological tendencies that shape decisions in unfamiliar situations. Its inputs include account onboarding, identity checks, money transfers, support calls and trading records. CTO and co-founder Ehud Ezra said it examines voice, sentiment and how often customers contact support.

The company turns those observations into mathematical representations of traits, then feeds them into a model that assigns probabilities to possible decisions. A separate language model generates explanations. Ezra said training rewards consistency with a person’s underlying tendencies: a simulated risk-averse customer should not suddenly become a risk-taker without the model treating that as an error.

Gal-Or said customer datasets now cover almost 30 million people; that figure does not establish how many are actively being profiled or what notice they received.

The narrow meaning of 98% accuracy

42AI claims 98% accuracy, but Gal-Or clarified that the figure measures whether trading flows fall within a predicted range. It does not mean the system correctly predicts 98% of individual trades. He also reported an F1 score of 0.8 to 0.85 for trading direction and volume. That metric balances correct predictions against false alarms and missed events.

The performance figures and claimed financial benefit lack independent verification. The brokerage customers remain unnamed, with no independent customer statements confirming their participation, Tech Times noted in its November 5 coverage.

Ezra said missing context, such as advice from a friend or colleague, can weaken predictions. Each new domain also requires training on its rules and context.

Whose advantage does the prediction serve?

Gal-Or argues the system does not need personally identifiable information. But it still links an individual’s transactions and interactions into a behavioral profile. Predicting customers’ actions could support legitimate risk management while also creating a conflict if a brokerage uses that insight for its own advantage.

Asked about that risk, Ezra acknowledged it as a valid concern. The founders cited regulation, monitoring and an internal ethics committee, but did not explain specific protections preventing customers from exploiting the people being profiled. There is no established evidence that 42AI’s brokerage customers have misused the technology.