The use of Generative AI (GenAI) is rapidly increasing within the financial services industry. Recently 48% of survey respondents in this sector reported regular use of GenAI technologies [19]. Responsible AI frameworks emphasise the importance of transparency and disclosure to ensure individuals understand how they are being impacted by AI systems. Despite these regulatory expectations, explainability for GenAI models remains a significant challenge. Some advances have been made on the model side, such as efforts to map the reasoning behind AI language models, but further progress is necessary to ensure that GenAI applications align with the rigorous transparency standards required in financial services. An example application for GenAI is to generate insights based on a business question by a banker. We outline an approach to ensure such insight generation is explainable and maintainable through the use of controlled language and Ripple-Down Rules.