In an era of AI-assisted data engineering, it's crucial to recognize that code, especially when generated without contextual understanding, can be a liability. The text emphasizes the importance of providing AI with a well-defined environment, termed as "context engineering," to prevent the generation of erroneous or deprecated code. It highlights several best practices for using AI to write Airflow Dags, such as defining the universe with specific constraints, using a skeleton strategy to build code iteratively, and employing golden records to maintain governance and adherence to team standards. The document also underscores the importance of contract-driven prompting to define data schemas accurately, and the use of Markdown and pseudo-XML tags to structure prompts effectively. Tools like the Astro IDE are recommended for their ability to integrate these best practices by offering version-aware support and context-driven workflows. Ultimately, while AI can expedite code generation, the onus is on engineers to provide the context, review, and ensure the maintainability and correctness of the code, transitioning from merely writing to architecting and reviewing code with a focus on business value.