Lyon

Founded 2024

Train foundation models directly on customer event data

Pricing model

Demo

Rating

★ People love it

Time to value

Requires Expertise

You can use Lyon to build and train custom foundation models directly from your raw customer event streams. The system ingests event sequences and profiles to create universal customer embeddings that run entirely inside your private cloud infrastructure. Instead of engineering separate features for every task, you can fine-tune task adapters with LoRA to forecast churn, detect fraud, evaluate credit risk, assess income, and predict cross-sell opportunities. All training and inference stay completely isolated within your VPC across AWS, GCP, or Azure.

What Lyon does

Ingest raw customer event histories and user profilesGenerate dense customer vector embeddings from behavioral dataFine-tune downstream task adapters using LoRADeploy inference models directly within a private VPCEvaluate credit risk and default probabilitiesDetect fraudulent transactions and suspicious activityForecast user churn and retention ratesTrains custom foundation models on customer event historiesOutputs unified customer embeddings for multiple downstream predictive tasksRuns all model training and inference inside your private VPCSupports LoRA task-specific fine-tuning on labeled tablesCombines existing engineered features alongside generated embeddingsEnsures data isolation compliant with SOC 2, GDPR, and LGPDDeploys across Amazon Web Services, Google Cloud, and Microsoft Azure