This study delves into optimizing the performance of ChatGPT, in the realm of translation and summarization tasks, with a focus on the effective use of style guides. Conducted in November and December 2023, the research employed the latest version of ChatGPT to explore its capabilities and limitations in style guide adaptation. The study investigates how the design of prompts, incorporation of examples, complexity of instructions, and style guide format influence ChatGPT's ability to comprehend and accurately translate and summarize content. Results show that it is critical to align examples with instructions in style guide prompts for effective AI comprehension and performance. It is imperative that every element detailed in the instructions be vividly illustrated in the corresponding examples, ensuring cohesive functionality. Moreover, the research introduced a mixed approach to divide style guides by leveraging both attachments and prompts. It experimented with a customized spreadsheet designed for AI. The study employed confusion matrices as a quantitative tool to measure ChatGPT's performances with and without the approach. The results show that the mixed approach augmented the True Positives (TP) count from 44 to 95 out of 120 points that need to be edited according to style guides, and elevated the Recall rate from 36.67% to 79.10%, which underscores the effectiveness of the mixed approach in enhancing ChatGPT's adherence to prescribed style guides.