Legora's tech chief says tokenmaxxing is a 'really stupid way' to encourage AI use

In a recent episode of the '20VC' podcast, Legora's Chief Technology Officer, Jacob Lauritzen, expressed strong criticism of the practice known as tokenmaxxing, suggesting that it is an ineffective way to promote the use of artificial intelligence within organizations. Lauritzen explained that many companies incentivize AI usage by implementing leaderboards that track token consumption during performance reviews. However, this approach often leads to employees burning through tokens merely to enhance their standing, which he described as a 'really stupid way to do anything.' Tokenmaxxing is characterized by the excessive use of various AI tools, such as Claude, Codex, and Cursor, in a bid to boost productivity metrics and improve rankings on internal AI usage dashboards. Lauritzen, who took on his role at the legal AI startup in 2024, emphasized that there are more effective strategies for integrating AI into the workplace. He advocated for initiatives like hack days or demonstrations where employees can showcase their projects and the efficiency improvements they've achieved. 'We should reward effectiveness and output rather than just AI usage,' he argued. Despite his criticism of tokenmaxxing, Lauritzen acknowledged the importance of AI for rapidly growing companies like Legora. He noted the high opportunity cost of not leveraging AI effectively, stating, 'Is it worth us spending a ton of tokens to learn if it maybe gives us 20% efficiency for us? Yes.' His remarks resonate with broader concerns in the tech industry as companies reassess their AI strategies. The conversation around tokenmaxxing has gained traction as firms transition towards token capping. There is growing unease among tech companies about the financial implications of their AI spending. Recently, Uber announced it would impose a limit of $1,500 per month on token expenditures for each AI tool, following a budget blowout earlier this year. Similarly, the Financial Times revealed that Amazon had dismantled an internal dashboard meant to track AI usage after employees manipulated the system to rise on the leaderboard. An Amazon spokesperson clarified that the dashboard was never intended to encourage AI use for its own sake. At a recent Bloomberg conference, Andrew Feldman, CEO of Cerebras Systems, echoed these sentiments, labeling the idea of providing employees with unlimited tokens as 'boneheaded from the get-go.' He advised companies to consider more cost-effective solutions, stating, 'You don't need a Ferrari to go to the grocery store; use a lower-cost open-source model.' As the tech industry navigates these challenges, the focus on smarter and more sustainable AI practices is becoming increasingly vital.