We ensure data isolation and zero data retention for our Generative AI API calls. Client data used for generation (e.g., a claims file) is processed temporarily and is never used to train the underlying public-facing LLM, maintaining strict confidentiality.