Reid Hoffman Supports 'Tokenmaxxing' as AI Usage Metric

LinkedIn co-founder Reid Hoffman has expressed support for the practice of tracking employee AI token usage, or “tokenmaxxing,” as a way to measure how staff engage with AI tools, according to TechCrunch. His remarks came during an interview at Semafor’s World Economy summit.

An AI token represents a unit of data processed by a model when generating responses, and companies have begun using total token consumption as an informal gauge of employee AI adoption. The concept, which gained traction in Silicon Valley, faced scrutiny after Meta reportedly shut down its internal “tokenmaxxing” dashboard following a leak.

Hoffman said that while token usage isn’t a perfect measure of productivity, it can serve as a useful indicator of experimentation and engagement with AI systems. He recommended pairing token tracking with reviews of how employees apply AI tools in their work.

He also encouraged companies to embed AI across all departments and hold regular check-ins to share learnings. According to Hoffman, this approach helps teams identify effective uses of AI and improve overall productivity.