Following the recent controversy surrounding Meta's internal dashboard that tracked employee usage of artificial intelligence tokens, Reid Hoffman has publicly endorsed the underlying concept of what Silicon Valley now calls "tokenmaxxing." As a LinkedIn co-founder and venture capitalist, Hoffman spoke at Semafor's World Economy summit this week, offering his perspective on how companies should measure and encourage AI adoption. An AI token represents a fundamental unit of data that models process to generate responses. This metric is also used to calculate service costs for organizations. Consequently, many tech firms have begun monitoring the volume of tokens used by individual employees as a proxy for their engagement with AI tools. The term "tokenmaxxing" combines this metric with Gen Z slang for optimization, similar to "looksmaxxing." While some engineers have criticized the practice as a flawed productivity measure—comparing it to ranking employees by spending rather than output—Hoffman argued for its value. During his interview, Hoffman stated that tracking token usage is a beneficial approach, provided it is understood within a broader context. He emphasized the importance of getting employees across various functions to experiment with AI. "You should be getting people at all different kinds of functions actually engaging and experimenting," Hoffman said. He acknowledged that token volume does not equate directly to productivity but suggested it serves as a useful dashboard indicator of who is actively using these tools. Hoffman noted that high token usage might sometimes reflect random or exploratory behavior. Therefore, he advised companies to pair usage data with qualitative analysis of what employees are actually achieving. He clarified that failed experiments are acceptable and even necessary, as long as there is a collective, simultaneous effort to explore the technology across the organization. Beyond mere measurement, Hoffman outlined a strategy for integrating AI into the corporate fabric. He urged leaders to embed AI tools throughout every department rather than isolating them in specific teams. Furthermore, he recommended establishing regular communication channels to share successes and learnings. He proposed weekly group check-ins focused on new AI applications tried during the week for both personal and organizational productivity. According to Hoffman, these discussions often reveal impressive results that can be replicated across the company. The debate over "tokenmaxxing" intensified after Meta discontinued its internal tracking tool following reports that the data had leaked to the press. Despite the backlash and questions regarding the metric's accuracy, Hoffman's endorsement highlights a growing consensus that measuring engagement is a critical step in the early stages of AI adoption. His comments suggest that while the metric is not perfect, it offers a tangible starting point for fostering a culture of experimentation and innovation.