A Generative AI Engineer is developing a GenAI assistant for HR departments that extracts key insights from employee satisfaction surveys. The surveys include open-ended comments and structured ratings. The engineer wants to generate actionable summaries per department and flag potentially harmful or discriminatory language automatically. Which combination of components is MOST appropriate for this solution?