Operationalizing Large Language Models