Woofun AI reports that Scott Wu, co-founder and CEO of Cognition, argued on the Founders podcast that assessing programmer performance based on AI token consumption is misguided. He emphasized that actual completed tasks and outputs should serve as the primary metrics for evaluation.

This practice, termed "tokenmaxxing", involves employees excessively using tools like Claude, Codex, or Cursor to inflate internal utilization dashboards. Jacob Lauritzen, CTO of Legora, criticized this approach in a recent podcast, describing it as a foolish method where staff artificially boost usage for appearances. Andrew Feldman, CEO of Cerebras Systems, echoed these sentiments at a Bloomberg conference, stating that unlimited token limits are unwise. He recommended prioritizing cost-effectiveness by utilizing low-cost open-source models for routine tasks rather than overusing proprietary ones.