To address the complex challenge of updating its extensive library of legacy code, coding, coding languages,Morgan Stanley developed an internal AI solution.

According to The Wall Street Journal, the AI tool, named DevGen.AI and powered by OpenAI’s GPT models, was launched in January. It specializes in translating code written in older languages-such as Perl, which dates back to 1987-into plain English. This translation then serves as a foundation for developers to rewrite the code in more modern languages like Python.

AI Efficiency Boost

Mike Pizzi,Morgan Stanley’s global head of technology and operations,told WSJ that DevGen.AI has processed nine million lines of code in the five months since its introduction.This has saved the company’s 15,000 developers approximately 280,000 hours of work.

“We found that building it ourselves gave us certain capabilities that we’re not really seeing in some of the commercial products,”

According to pizzi, morgan Stanley chose to create the tool internally because existing commercial solutions lacked the specific expertise needed to decipher the firm’s unique and older coding languages.

Human Oversight Remains Crucial

DevGen.AI was specifically trained on Morgan Stanley’s internal code base, including customized languages. While the tool can theoretically translate code from older to newer languages, it currently lacks the ability to write new code as efficiently or effectively as a human developer, according to Pizzi.

Consequently, Morgan Stanley is maintaining human developers in the code translation process. Pizzi clarified that the company does not plan to reduce its software engineering staff as a direct result of the AI tool, despite having laid off 2,000 employees from its 80,000-person workforce in March.

Expanding AI Applications

Morgan Stanley has launched multiple AI apps for its employees, including tools for summarizing video meetings and quickly retrieving information from the company’s research archives.

Morgan Stanley CEO Ted Pick told investors last year that these AI tools have the potential to save employees up to 15 hours per week, describing them as “potentially really game-changing,” according to Reuters.