Amazon issues new AI rulebook for engineers, read full memo here

Amazon has sent a clear message to its software engineers: use Artificial Intelligence (AI) to solve real customer problems, not just to boost your usage stats. In an internal memo, which was circulated later, the tech giant outlined updated guidelines for its Software Development Engineers (SDEs), urging them to focus on long-term results rather than simply chasing AI trends.The company wants its teams to build meaningful products instead of forcing AI into tasks where it does not belong.The Rise of the Renaissance DeveloperHistorically, Amazon avoided making its engineering guidelines about specific technologies. Even major shifts over the last two decades, like the cloud, microservices, and serverless technology, did not change the core definition of the role. However, the company notes that AI is different. It is changing software engineering much faster and more fundamentally than past trends.Amazon believes the tech industry has entered the era of the "Renaissance Developer," where engineers must blend AI tools with human judgment and deep domain expertise. While it is still early, the memo warns that employees must take the time to learn and apply AI into their daily work, or they risk being left behind. The company stated that a developer's role and output are not defined solely by their use of AI. Amazon is focusing on what remains constant: that core engineering values remain exactly the same. The ultimate goal is still to deliver high-quality results for customers.The Problem with 'Tokenmaxxing'As Amazon rushes to integrate AI, it is strictly warning employees against "tokenmaxxing," a practice where developers give useless tasks to AI agents just to inflate their token consumption and look productive. In May 2026, the company even shut down an internal AI leaderboard after discovering that employees were gaming the system. The memo calls metrics like AI token consumption and the volume of AI-assisted code changes "deeply flawed." Relying on these numbers creates bad incentives, increases costs, and goes against Amazon's core principle of frugality.Measuring Real SuccessInstead of tracking how much AI an employee uses, Amazon wants engineers to be measured by long-term outcomes and actual productivity. Marc Brooker, a vice president at Amazon Web Services (AWS), told CNBC that the company is now focusing on "feature velocity," which tracks how many successful updates and new features actually reach users.The memo advises leadership to carefully monitor both extremes of AI adoption: the 10th percentile of teams who are not adopting AI effectively, and the 90th percentile of heavy AI users, to check if their high usage actually results in faster delivery.Leaders Must Get Hands-OnThe memo stresses that leaders cannot just rely on hearsay when it comes to AI. Technical and organizational managers are expected to understand these technologies deeply by using them hands-on in their production work.A leader must be able to describe exactly where an AI tool helped and where it did not. Amazon noted that the leaders having the most impact right now are those doubling down on the company's "learn and be curious" principle, making time for themselves and their teams to explore AI.While Amazon urges its employees to learn AI, it also respects critical thinking. Leadership made it clear that forcing AI into daily tasks is a bad idea if it does not actually help. The memo explicitly told staff that if their work does not benefit from current AI tooling, that is a finding, not a failing. Employees are encouraged to document why an AI tool failed and share that feedback with their managers.Reconnecting with the CustomerAmazon also observed that engineers and managers who are closest to their products and businesses take the most advantage of AI. As the company has grown, some developers have become disconnected from their customers.Amazon wants to reverse this trend, asking engineers of all levels to become deeply connected to their product direction. According to the company, an engineer who understands the customer and knows exactly when to avoid using AI is demonstrating the exact kind of judgment Amazon values most. - Ends