A financial services company wants its gen AI assistant to consistently adopt a very specific writing style and reasoning pattern used by its senior analysts. The company has 5,000 high-quality example question-answer pairs written in that exact style. The underlying facts change rarely. Which technique is the BEST fit for teaching the model to reliably produce output in this specialized style?