New Delhi: Goldman Sachs has started piloting ‘Devin’, an autonomous AI software engineer developed by US-based startup Cognition. The bank is testing the system as part of a larger plan to improve developer efficiency and cut operational costs.
The pilot, announced by Goldman Sachs CIO Marco Argenti, involves integrating Devin into its existing team of over 12,000 human software developers. The move is part of a broader industry trend exploring the use of generative and agentic AI for handling complex workflows in software development. Unlike earlier AI tools that focused on limited tasks like summarizing documents or generating code snippets, Devin is designed to act autonomously. It can carry out full-stack development tasks, including application building and updating legacy systems. At Goldman Sachs, Devin is currently being tested to automate time-consuming work such as routine code updates, allowing human engineers to focus on architecture, design, and innovation.
Argenti said the bank plans to deploy “hundreds of Devins,” with the possibility of scaling up to thousands, depending on outcomes from the initial phase. According to internal assessments, early tests show that Devin could offer a three-to-fourfold increase in productivity compared to previously used AI tools. Goldman Sachs’ adoption of Devin signals a potential shift toward a hybrid model in tech teams, one where humans define the problem and supervise AI agents that execute tasks. This could reshape the nature of software development, emphasizing roles in prompt engineering, problem design, and quality assurance. However, the shift also raises concerns about workforce implications. Analysts note that such tools could lead to a reduction in demand for entry-level software engineers, particularly for tasks that are now automatable.
Also Read:
Discussion about this post