How do you prevent an LLM from hallucinating when queried about highly specific hardware specifications in a vector search application?

Interview question asked to Data Scientists interviewing at Babylon Health, Centrica, Scribd and other companies. Original question asked: How do you prevent an LLM from hallucinating when queried about highly specific hardware specifications in a vector search application?.