Amazon is rigorously monitoring the adoption and output impact of artificial intelligence across its retail engineering division. Internal documents obtained by Business Insider reveal that the company is tracking how frequently software engineers utilize AI tools and measuring whether this integration leads to meaningful productivity gains. This data-driven approach aims to navigate internal resistance while pushing for significant efficiency improvements within the organization. The initiative targets over 2,100 engineering teams in the retail arm, known as Stores, with a mandate to triple software code release velocity through AI-native practices. A smaller group of at least 25 teams has even stricter goals, aiming to boost output tenfold this year. Progress is closely supervised by Amazon's senior leadership, the S-Team. While generative AI tools like Anthropic's Claude and OpenAI's Codex have already revolutionized software production globally, Amazon is aggressively embedding similar technologies into its own engineering culture. CEO Andy Jassy has previously urged employees to treat AI as a critical automation investment, warning that failure to adopt it could jeopardize job security. To ensure success without falling victim to Goodhart's Law—where a metric becomes useless once it becomes a target—Amazon designed its tracking system to measure deployment rates and engagement while guarding against superficial compliance. The company emphasizes measuring both access to tools and actual usage to understand true adoption. As of February, about 60% of retail engineering teams had adopted AI-native practices, with a target of 80%. Several proprietary tools have seen rapid expansion. AI Teammate, a Slack-integrated agent, now serves over 700 teams, while Pippin, which assists in creating technical designs, has gained traction even within the AWS cloud division. Tools like Kiro, an AI coding assistant, are also experiencing increased engagement. Despite these aggressive goals, the top-down approach has encountered friction. Internal feedback highlights concerns regarding centralized mandates, overlapping efforts, and the burden of self-reported metrics. Some engineers noted that onboarding processes for certain tools were overly complex, creating barriers to adoption. In response, Amazon plans to shift its guidance toward collaborative practices rather than mandating specific tools. The company is also moving to replace manual reporting with automated metrics and is developing a centralized learning platform to streamline best practices. Amazon spokesperson Montana MacLachlan stated that the company does not centrally mandate tool usage but instead encourages teams to select what works best for their specific needs. The internal document, titled Amazon Confidential, outlines six AI-Native Engineering Tenets to guide this transition. These principles prioritize delivering working solutions over cost optimization, ensuring AI tools are used where they add value rather than forcing them into every problem, and maintaining transparency so that all deployed systems remain auditable and understandable. The overarching strategy encourages engineers to experiment with different tools and identify manual tasks that AI can accelerate. Managers are instructed to set clear guidelines and ensure easy access to these technologies, aiming to make AI part of the daily workflow rather than an occasional resource. While the company acknowledges the practical challenges and internal pushback, it remains committed to embedding AI deeply into its operations to drive long-term innovation and speed.