Indeed closely monitors how its employees use AI, but is “far removed” from any kind of “Tokenmaxxing” leaderboard.
“We track token usage, but we use it in the background,” Anthony Moisant, Indeed’s chief information officer, told Business Insider in an interview. “We’re definitely not going to use leaderboards.”
Moisant said there is “no inherent problem” with the types of leaderboards that Meta and other Big Tech companies are reported to be using to incentivize employees to spend as many tokens as possible. He just wants to track metrics that are close to results.
“I think any time you have metrics and measures like this as part of an incentive system, it creates that perverse incentive and people start doing things,” Moisant said. “Even if it’s not malicious, they’re essentially just following incentives. They’re following the carrot they’re chasing, and that often leads to bad outcomes.”
Certainly CIO Anthony Moisant Courtesy of Indeed
Across the technology industry, efforts are underway to enable employees to fully utilize AI in their work. Nvidia CEO Jensen Huang recently said that he expects an engineer making $500,000 to spend $250,000 worth of AI tokens.
Tokens are a way for large-scale language models to break down words into numeric inputs and numeric outputs, and are essentially the building blocks of AI chatbots like OpenAI’s ChatGPT.
Mozant said Indeed is not focused on the use of tokens, but rather on the speed at which products are shipped and how customers are responding to those changes.
“The activity has no value to our business and it has no value to the job market,” he said. “The result is better matching, faster matching. That’s what we’re aiming for.”
Moisant said that data-hungry organizations like his tend to focus too much on things that are easy to measure. That’s a mistake Indeed made in the past when it tracked the percentage of code written, a metric that became popular when generative AI coding tools like Anthropic’s Claude Code and Cursor became popular, he said.
“We used that metric for about three or four months and found that it was certainly a good proxy metric, but it wasn’t about results,” he said. “There really wasn’t any change. Productivity wasn’t really rebuilt.”
How Indeed thinks about AI costs
“Tokenmaxxing” is also consistent with the reality that many companies are spending far more on AI than they did a year ago, let alone six months ago.
Moisant said Indeed’s AI-related bills are expected to quadruple compared to 2025, with most of that coming from its use of research and development. He said the board recently discussed how to balance staying within projected budgets while not missing out on the potential advances made possible by AI tools.
“We had a rich discussion, but is that really what we want? Because the minute we start setting the tone for the budget, it also slows down the productivity that we’re looking for, the results that we’re looking for,” he said.
Ultimately, Moisant said, it’s important to be cautious with your spending, even in high-ROI areas where you essentially have a “green light” to “spend as much as you need.” One of those areas, he said, is “time-to-value,” or the time it takes Indeed to go from an idea to shipping a product to a customer.
“I think the transparency and understanding of how we’re spending our money and who’s spending our money in the background is looking at token usage. But what we’re looking at from that perspective is, ‘Hey, this is actually very interesting. Team 1 is kind of surging. Why are they surging? Are we seeing value?'” he said.