The metric measures activity, not outcome.
Across corporate offices in Brazil and beyond, a new kind of performance review has quietly taken hold — one that counts not what workers accomplish, but how often they invoke a machine. The practice, known as 'token maxxing,' reflects a deeper anxiety at the heart of the AI era: organizations under pressure to prove their technological relevance have begun mistaking measurable activity for meaningful progress. In ranking employees by their volume of AI interactions, companies may be optimizing for the appearance of transformation rather than its substance.
- Corporations are installing dashboards that rank employees by how frequently they use AI tools — turning a technology adoption question into a competitive performance metric.
- Workers flagged as low token users risk being seen as resistant to digital transformation, while heavy users are held up as models of innovation regardless of actual output.
- The ranking system breeds perverse behavior: employees begin gaming the leaderboard, feeding trivial tasks to AI simply to raise their numbers and avoid falling behind.
- A surveillance undercurrent runs through the trend — knowing that every AI interaction is tracked and compared changes how workers choose to engage with the tools at all.
- The critical gap no dashboard captures is the difference between genuine problem-solving and performative usage, leaving companies measuring motion while mistaking it for direction.
Inside offices across Brazil and beyond, a quiet new metric has begun to shape careers: how many times a day you use artificial intelligence. Dubbed 'token maxxing' in tech circles, the practice has companies installing dashboards that track employee interactions with tools like ChatGPT or proprietary language models — then ranking individuals, teams, and departments by the numbers.
The logic behind it reflects a particular corporate anxiety. Executives who have invested heavily in generative AI face pressure to demonstrate adoption. More usage, the thinking goes, must mean more productivity and more competitive edge. So the rankings circulate in meetings, influence how managers evaluate their staff, and in some cases factor into compensation and promotion decisions.
But the metric measures activity, not outcome. There is no built-in way to distinguish the engineer who uses AI to crack a critical problem from the one who routes every minor task through a chatbot simply to move up the leaderboard. The system meant to encourage genuine adoption instead rewards theatrical engagement — doing things the AI way not because it's better, but because it's visible and countable.
The surveillance dimension adds another layer of discomfort. Knowing that every interaction is tracked and compared to peers creates a panopticon effect, distorting behavior in both directions: some workers avoid AI to resist the appearance of dependency, while others use it excessively to avoid being seen as laggards.
The deeper question remains unanswered: do companies that rank employees by AI usage actually deliver better products, serve customers better, or grow more sustainably? The evidence is thin. What the metric offers is something easier — a number that looks like progress and communicates well to shareholders. The real reckoning will arrive when organizations discover that their highest-ranked token users haven't moved the needle on anything that actually matters.
Inside corporate offices across Brazil and beyond, a new metric has begun to matter: how many times you use artificial intelligence. Companies are now measuring their workers not by the quality of their output or the problems they solve, but by the sheer volume of AI interactions they log each day. This practice, dubbed "token maxxing" in tech circles, treats employee engagement with generative AI tools as a scoreboard—and workers are being ranked accordingly.
The trend reflects a particular kind of corporate anxiety. As artificial intelligence has moved from laboratory curiosity to workplace staple, executives face pressure to show that their organizations are actually using these expensive new systems. The logic seems straightforward: more AI usage equals more productivity, more innovation, more competitive advantage. So companies have begun installing dashboards that track how often employees invoke ChatGPT, Claude, or proprietary language models. They rank departments. They rank teams. They rank individuals. The numbers become visible, comparable, competitive.
What started as a way to measure adoption has evolved into something closer to a performance management system. An employee who generates fewer "tokens"—the technical term for chunks of text processed by an AI model—can find themselves flagged as a laggard, someone not fully embracing the company's digital transformation. Conversely, high token users become exemplars, proof that the organization is truly AI-forward. The rankings circulate in meetings. They influence how managers think about their staff. In some cases, they factor into compensation and promotion decisions.
The problem is that this metric measures activity, not outcome. A worker who uses AI ten times a day to draft emails, brainstorm ideas, or debug code might be genuinely productive. Or they might be someone who has simply learned to log every minor task through an AI interface. There is no built-in way to distinguish between meaningful usage and performative usage—between the engineer who uses AI to solve a critical problem and the one who uses it to generate boilerplate text they could have written themselves in half the time.
Moreover, the ranking system creates perverse incentives. Employees begin to optimize for the metric rather than the work. They find ways to increase their token count. They use AI for tasks that don't require it, simply to move up the leaderboard. The system that was meant to encourage genuine AI adoption instead encourages a kind of theatrical engagement with the technology—doing things the AI way not because it's better, but because it's measurable and visible.
There is also a surveillance dimension that makes many workers uncomfortable. Being ranked by AI usage means your interaction with these tools is being tracked, quantified, and compared to your peers. It creates a panopticon effect: you know you're being watched, so you adjust your behavior. Some employees may avoid using AI for fear of being seen as dependent on it. Others may use it excessively to avoid being seen as resistant to it. Either way, the ranking system has changed the nature of the choice.
The deeper question is whether any of this actually works. Do companies that rank employees by AI usage see better business outcomes? Do their products improve? Do their customers notice? The evidence is sparse. What we know is that the metric is easy to measure and easy to communicate to shareholders. It looks like progress. It feels like innovation. But it may be measuring the wrong thing entirely—confusing motion with direction, activity with accomplishment. As this trend spreads, the real test will come when companies realize that their highest-ranked token users haven't actually delivered anything of value.
The Hearth Conversation Another angle on the story
So these companies are literally counting how many times someone uses an AI tool and then ranking them by that number?
Exactly. It's become a visible metric—something you can put on a dashboard and show to leadership. The assumption is that more AI usage means more productivity, more innovation.
But that seems backwards. Couldn't someone use AI constantly and still produce nothing useful?
Yes. And that's the core problem. The metric measures activity, not outcome. You can't tell the difference between someone solving a real problem with AI and someone just running tasks through it to boost their numbers.
What happens to the people who don't rank high? Are they penalized?
Not always formally, but the rankings create pressure. Managers see them. They influence how people are perceived. In some places, they affect raises and promotions. So people start using AI not because it helps their work, but because they're being measured on it.
That sounds like it would make people resentful.
It does. There's also a surveillance element that bothers people. You know your AI usage is being tracked and compared. It changes how you work, whether you want it to or not.
Has anyone actually proven that this ranking system improves business results?
Not really. That's the thing—it looks like progress, it's easy to measure, but whether it actually makes companies better at anything is still an open question.