Amazon has discontinued an internal ranking system that tracks employee AI token usage. Company executives recently told employees that AI should not be used for its own sake, but rather the focus should remain on solving customer and business problems and driving innovation.
Employee-created leaderboards have been suspended.
This internal dashboard, called KiroRank, was created by a group of employees to track AI Token usage. An Amazon spokesperson confirmed to Business Insider that the tool has been discontinued.
The company stated that the dashboard was originally intended as an informal tracking tool and not to encourage employees to simply pursue higher AI usage. Amazon says that each team can decide for itself how to use AI tools and how to record related usage.
The company opposes the "token-swiping" approach.
Amazon Senior Vice President Dave Treadwell told employees this week that AI should not be used for its own sake, but rather used to solve customer needs, business problems, and drive innovation.
The report mentions that "tokenmaxxing" is a term that has recently emerged in Silicon Valley, referring to using the consumption of AI tokens to measure the productivity of an individual or team. While Amazon tracks token usage to assess costs, it does not encourage this approach.
Token usage will increase in 2026.
Tokens are the basic units of measurement used by large language models when processing text, and they are also the foundation for chatbots and AI programming tools. With the rapid popularization of Agentic AI in 2026, the use of tokens has increased significantly, as these systems can run continuously for extended periods with minimal human intervention.
Business Insider previously reported in April that Amazon's retail business has been closely tracking how engineers use AI, including monthly user numbers, the frequency with which AI tools are embedded in workflows, and the results of these deployments.
The discontinuation of internal rankings shows that while large technology companies are promoting the adoption of AI, they are also beginning to more clearly distinguish between "effective use" and "use for the sake of making data look good."
