Customer foundation model platform that trains a private model on a company's own event data inside its cloud for multi-task prediction.
Lyon trains a foundation model from scratch on a company's proprietary transaction, payment, click, and customer interaction data, running entirely inside the company's VPC on AWS, GCP, or Azure. The resulting model produces rich customer embeddings supporting multiple downstream predictions—churn, credit risk, fraud, collections, income, cross-sell, lifetime value—without separate feature engineering projects per task. Companies own the weights and data never leaves their cloud. Built by a team from Stanford and NVIDIA. Lyon is a product of Lyon.