In manufacturing R&D departments, program assets in a variety of languages ​​accumulated over many years often face issues such as personalization and maintainability. This presentation will use the simulation development department of Sumitomo Rubber Industries as an example to demonstrate the use of generative AI to solve these issues. Specifically, we will explain a practical example of a development cycle using generative AI, including the conversion of legacy code such as Fortran and Perl to Python, a practical approach such as "summarize → generate new code," and the automatic generation of unit tests. We will also discuss a method for deploying the developed tool internally as a low-cost, secure web application using Google Cloud, offering hints for promoting digital transformation in the manufacturing industry.