In our industrial research laboratory, we focus on enhancing the reliability of generative AI by integrating foundational research with practical application. On the academic front, we advance scientific boundaries through rigorous investigation and publication. Simultaneously, we transform the latest breakthroughs into tangible business outcomes. This presentation highlights the interplay of these efforts, beginning with a financial services use case that underscores the critical importance of reliability and oversight. We then introduce a core technology that improves the reliability of large language models and empowers users with effective oversight tools.