Amazon Engineers Use Copilot as AI Halves Team Sizes and Saves 4,500 Developer-Years

Amazon has accelerated its adoption of artificial intelligence (AI) in software development, with managers increasingly pushing engineers to use tools such as Copilot to automate coding tasks, raise output goals, and meet tighter deadlines.

Some teams at Amazon are now expected to produce the same amount of code with roughly half the staff compared to the previous year, relying on AI to maintain productivity. AI coding assistants have contributed to a more than 25% rise in output, and the company reports saving the equivalent of 4,500 developer-years by automating software updates.

Engineers describe their work as becoming more routine and faster-paced, with much of their job shifting from writing code to reviewing and verifying AI-generated code. Many report less time for thoughtful work and reflection, and some express concern that automation may limit opportunities for junior staff to develop skills needed for advancement.

Amazon has encouraged coders to develop new AI productivity tools through internal hackathons, and engineers use AI not only for coding but also to draft memos and test software features. However, some engineers are reluctant to adopt new AI tools due to the need for extensive double-checking and a desire for greater control over their work.

Amazon Employees for Climate Justice, a worker group, has become a forum for discussing the stress associated with AI-driven changes. Amazon maintains that AI is intended to augment, not replace, engineers' expertise and says it regularly reviews staffing levels. Similar shifts toward routine use of AI are occurring at other technology firms, such as Shopify and Google, where AI adoption is now a baseline expectation and performance reviews increasingly focus on AI usage. At Google, a companywide hackathon offers a $10,000 prize for teams creating productivity-enhancing AI tools.