Amazon is asking its software engineers to rethink how they use AI. The company has warned against a growing trend called tokenmaxxing, where workers use AI tools as much as possible. But more AI use does not always mean better work.
What is tokenmaxxing?
As reported by CNBC, AI tools process text in units called tokens. These are parts of words, whole words or bits of code. The term tokenmaxxing describes the habit of using as many tokens as possible. Amazon has updated its guidance for software engineers and managers. The memo warns staff not to treat token use, AI tool adoption or the amount of code changed as clear signs of good work.
These numbers may look useful, but they do not always show whether a team has built something valuable for real users. The memo was shared with staff in September and posted on Amazon’s website on October 27. It says the company is entering the era of the “Renaissance Developer”. This is an engineer who can use AI tools while also relying on their own skills, knowledge and judgement.
Why is Amazon worried?
The main concern is waste. AI tools use computing power, and heavy use can raise costs. If workers ask AI to do tasks that add little value, the company may spend more without getting better results. It can also make work harder to manage. Amazon’s memo calls token counts and lines of code “deeply flawed” ways to judge productivity. The company says such measures can lead to extra cost and needless complexity.
It wants staff to focus on the value of the work, not the amount of AI used to do it. Amazon has already seen how these measures can be misused. Some workers reportedly ran AI agents on meaningless tasks to raise their token counts. They wanted to climb an internal leaderboard called Kirorank. Amazon shut the board down in May after it found that staff were gaming the system.

What does Amazon want from engineers?
Amazon still sees AI as a major part of the future of software work. The memo says the best way to judge progress is through long-term results. Managers should look at whether AI helps teams launch useful features, make better changes and solve problems for customers. Marc Brooker, a vice president at Amazon Web Services, said that the business enterprise looks at “feature velocity”. This includes how many features teams launch and how many changes work well in real use.
The guidance asks engineers to use their own judgement instead of chasing AI usage numbers. It also asks managers to understand why some workers use AI less than others, and whether those who use it more are producing better results. This shift does not mean Amazon is stepping back from AI. It wants its staff to learn how to use it well. The company is trying to draw a line between useful use and use that only makes the numbers look good.
Author’s Opinion
Amazon’s warning raises an important point. AI should help people do better work, not become a target in itself. Companies must judge workers by their results and skills, rather than how often they use a tool.
READ MORE: Microsoft offers 60,000 employees an exit package as AI costs rise.
ALSO READ: Apple iPhone Duo buyers report cracked displays days after launch.


Comments
0Be the first to join the discussion.