Recent studies have shown that, when supplied with a contextual reference at Inference, LLMs opt to make use of the contextual data as opposed to model training data.