Lyon

Lyon builds private foundation models for banks, insurers, and fintechs, trained on a client's own proprietary data — transactions, payments, clicks, customer interactions — to predict credit, fraud, collections, income, churn, and lifetime value. Crucially, the models run inside each client's own cloud (AWS, GCP, or Azure VPC), so data and the intelligence derived from it stay under the client's control.

Financial firms are sitting on vast proprietary transaction data but are wary of handing it to third-party AI. Lyon's bet is that the winning model for financial services isn't a shared, general-purpose one, but a private one that never leaves the customer's infrastructure.

Founded in San Francisco in 2026 by Gabriel Noya, Lyon is tackling the gap between generic foundation models and the tightly governed reality of enterprise financial data.

Early proof points back the approach: the company says it trained a model on 28 billion transactions for a fintech serving tens of millions of active users, identifying premium-card converters four times more precisely than existing rules — a model now being deployed for credit. Lyon is also working with a major insurer and is part of Y Combinator's Fall 2026 batch.