Amazon has issued a clear directive to its software engineers: use artificial intelligence to address genuine customer needs rather than simply increase usage statistics. In an internal memo that was subsequently circulated, the technology giant introduced updated guidelines for its Software Development Engineers (SDEs), encouraging them to prioritise lasting results over chasing AI trends.
The company wants its teams to develop useful products instead of incorporating AI into tasks where it offers little or no benefit.
The Rise of the Renaissance Developer
Amazon has traditionally avoided defining its engineering guidelines around specific technologies. Even major developments over the past two decades, including cloud computing, microservices and serverless technology, did not fundamentally alter its definition of an engineer's role. However, the company believes AI is different because it is transforming software engineering more rapidly and fundamentally than previous technological shifts.
According to Amazon, the industry has entered the era of the "Renaissance Developer", in which engineers must combine AI tools with human judgment and extensive knowledge of their fields. Although this transformation is still in its early stages, the memo cautions that employees need to learn how to incorporate AI into their daily work or risk falling behind.
The company emphasised that a developer's responsibilities and performance should not be judged solely by how much AI they use. Instead, Amazon is reinforcing its longstanding engineering principles, maintaining that the primary objective remains delivering high-quality outcomes for customers.
The Problem with 'Tokenmaxxing'
As Amazon expands its adoption of AI, it is warning employees against "tokenmaxxing". The term describes the practice of assigning unnecessary tasks to AI agents simply to increase token consumption and create the impression of higher productivity.
In May 2026, Amazon even discontinued an internal AI leaderboard after discovering that employees were manipulating the system to improve their rankings.
The memo describes metrics such as AI token consumption and the number of AI-assisted code changes as "deeply flawed". Depending on these figures can encourage counterproductive behaviour, drive up costs and conflict with Amazon's fundamental principle of frugality.
Measuring Real Success
Rather than measuring employees by their AI usage, Amazon wants to evaluate engineers based on sustained results and genuine productivity improvements.
Marc Brooker, a vice president at Amazon Web Services (AWS), told CNBC that the company is concentrating on "feature velocity", a measure of how many successful updates and new features actually reach customers.
The memo also recommends that leaders examine both ends of the AI adoption spectrum. They should assess the 10th percentile of teams that are struggling to adopt AI effectively, as well as the 90th percentile of heavy AI users, to determine whether extensive usage is genuinely helping those teams deliver work faster.
Leaders Must Get Hands-On
The memo makes clear that leaders should not depend on second-hand accounts when evaluating AI. Technical and organisational managers are expected to develop a thorough understanding of these tools by using them directly in their production work.
Leaders should be able to explain precisely where an AI tool has proved useful and where it has fallen short. Amazon observed that the leaders making the greatest impact are those embracing its "learn and be curious" principle and setting aside time for themselves and their teams to experiment with AI.
While encouraging employees to develop their AI skills, Amazon is also emphasising the importance of critical thinking. The company said forcing AI into everyday tasks is counterproductive when the technology does not improve the outcome.
The memo explicitly states that if a particular task does not benefit from current AI tools, identifying that limitation should be viewed as a finding rather than a failure. Employees are encouraged to document why a tool did not work and communicate their observations to their managers.
Reconnecting with the Customer
Amazon has also noted that engineers and managers who maintain close relationships with their products and businesses tend to make the most effective use of AI. As the company has expanded, some developers have become increasingly disconnected from the customers they serve.
The company now wants to reverse that trend by encouraging engineers at every level to develop a deeper understanding of their products and their strategic direction.
According to Amazon, an engineer who understands customer needs and recognises when AI should not be used demonstrates precisely the kind of sound judgment the company values.
