Amazon’s retail business is facing what some employees describe as Artificial Intelligence sprawl, as teams rapidly build internal Artificial Intelligence-powered applications that automate workflows and organize information. An internal document said the company’s long-running problem with duplicate tools is worsening because Artificial Intelligence makes it much easier to create and maintain new software. Instead of reusing existing systems, engineers can now build alternatives quickly, leading to more overlapping tools and less pressure to retire them.
The document said Artificial Intelligence is accelerating duplication on both the software and data sides. Many systems ingest internal information and transform it into new outputs such as knowledge bases and summaries, which are then stored separately from the original source. That creates additional copies of the same information and introduces governance problems when permissions change or source data is deleted. In one internal example, Spec Studio continued showing software details that had been made private in Amazon’s internal code repository. The document warned that when systems transform data through Artificial Intelligence and store the output separately, derived artifacts can persist even after source permissions change or records are removed.
The issue reflects a broader pattern as generative Artificial Intelligence lowers the barrier to building tools across large organizations. Amazon has historically allowed small teams to move independently under its two-pizza team model, a structure that supports rapid experimentation but can also encourage parallel development. The internal team behind the document, which was produced in February and focused on Artificial Intelligence tools used across Amazon’s retail organization, argued that this culture can make it harder to identify overlaps early and consolidate systems before duplication becomes entrenched.
Amazon disputed the broader characterization. Spokesperson Montana MacLachlan said the document reflected the perspective of a single team and said it was inaccurate to use one group’s view to represent the wider workforce. Even so, the company is exploring a familiar response: using more Artificial Intelligence to manage the spread of Artificial Intelligence. According to the document, Amazon is considering tools that can identify duplicate systems, flag risks, and nudge teams to consolidate earlier. The central challenge is maintaining the speed and autonomy that have long defined Amazon’s engineering culture while restoring visibility and control over proliferating tools and data.
