Amazon Reportedly Pressures Staff to Use More AI
Amazon employees are reportedly “tokenmaxxing” by overusing AI tools to meet internal usage targets, raising questions about how Big Tech measures AI productivity.
The AI race inside Big Tech is no longer just about building better models. It is also becoming a workplace performance metric.
A recent report highlighted by Futurism claims that Amazon is increasing pressure on employees to actively use artificial intelligence tools as part of daily work. According to the report, the company wants more than 80% of its developers to engage with AI tools every week, while also tracking usage through internal metrics tied to “token consumption”.
In AI systems, tokens are small chunks of text processed by a model. Every prompt, command, or generated response consumes tokens, making them a common way to measure AI activity and computing usage.
The rise of “tokenmaxxing”
Sources claim some employees have started using Amazon’s internal AI agent, MeshClaw, for trivial or unnecessary tasks simply to increase their usage numbers. Workers reportedly refer to the behaviour as “tokenmaxxing”, a term describing the practice of inflating AI activity metrics rather than focusing on meaningful productivity gains.
The idea reflects a wider challenge emerging across technology companies as businesses rush to integrate generative AI into workflows. While organisations want employees to adopt AI quickly, measuring usage through raw numbers can create unintended consequences.
If workers are judged by how often they use AI instead of what they accomplish with it, the focus can shift from productivity to performance signalling.
Amazon’s broader AI strategy
Amazon reportedly told the Financial Times that thousands of employees already use AI every day to automate repetitive work. The company also reiterated its commitment to building and deploying generative AI responsibly and securely.
The push aligns with a much larger industry-wide trend. Nearly every major technology company is racing to prove that AI tools are improving efficiency, reducing repetitive work, and accelerating software development.
Inside engineering teams, AI coding assistants are increasingly being positioned as standard workplace tools rather than optional experiments. Companies are investing billions into these systems and are under growing pressure to demonstrate returns on those investments.
When metrics start shaping behaviour
The issue raised by the news is less about AI itself and more about how companies measure success. Targets and dashboards can accelerate adoption during the early stages of technological change.
However, workplace experts have long warned that metrics can become distorted if employees begin optimising for numbers instead of outcomes. In this case, measuring token usage may encourage employees to generate more prompts, automate unnecessary tasks, or rely on AI even when simpler solutions would work better.
That can increase operational costs while making it harder for companies to understand whether AI tools are genuinely improving productivity. It may also create additional stress for workers who feel pressured to constantly demonstrate visible AI usage.
The bigger workplace question
This trend places Amazon’s situation within a broader corporate AI trend where enthusiasm from leadership sometimes collides with employee concerns over workload, monitoring, and expectations.
As companies continue pouring money into generative AI, executives increasingly want measurable proof that employees are adopting the tools. But forcing adoption too aggressively can risk turning AI from a productivity enhancer into another workplace target.
The coming months may reveal whether companies like Amazon refine these systems to focus more on measurable outcomes instead of raw activity. Better training, clearer guidance on high-value use cases, and smarter evaluation methods could help prevent employees from gaming the system.
For now, it highlights a growing tension across the tech industry: how to encourage meaningful AI adoption without reducing work to a numbers game.


