Demonstrates how fine-tuning LLMs with readily available consumer data and using retrieval-augmented generation (RAG) to provide contextual information can create “digital twins” that accurately simulate consumer choice and evaluations.