This article explores a novel approach that combines large language models, live risk telemetry, and orchestration pipelines to automatically generate and adapt security policies for vendor questionnaires, reducing manual effort while maintaining compliance fidelity.