Morgan Stanley, a leading investment bank, has implemented an in-house AI tool called DevGen.AI, which has been processing 9 million lines of legacy code this year. This initiative has significantly reduced the workload for the bank’s developers, saving them an estimated 280,000 hours of work. The tool, which is based on OpenAI’s GPT models, has been launched since January and is designed to convert outdated programming languages into plain English specifications, making it easier for developers to rewrite the code using modern languages.
The initiative is addressing a significant challenge within enterprise software, which is the modernization of decades-old code. This challenge weakens the security of the code and hampers the company’s ability to adopt new technologies. The bank’s global head of technology and operations, Mike Pizzi, highlighted this as one of the biggest pain points in enterprise software. While commercial AI coding tools are adept at writing new code, they lack expertise in older or company-specific programming languages such as Cobol. To overcome this, Morgan Stanley has trained its own system on its proprietary codebase, enabling the AI to understand and process the specific languages used within the company.
The primary strength of the DevGen.AI tool lies in its ability to create English specifications that map the functions of legacy code, allowing any of the company’s 15,000 developers worldwide to rewrite the code in modern programming languages. This approach reduces the reliance on a dwindling pool of specialists familiar with antiquated coding systems. By enabling every developer to contribute to modernizing the codebase, the tool fosters a more inclusive and efficient development process. Overall, the implementation of this AI tool is seen as a significant step forward for Morgan Stanley, demonstrating the bank’s commitment to leveraging technology to enhance efficiency and security in its operations.