Anthropic's new Economic Index report introduces five 'economic primitives' to measure how Claude was used in November 2025. These metrics cover user skills, task complexity, autonomy, success rates, and usage purpose. The data shows usage remains concentrated, with coding tasks dominating. Augmentation patterns edged back above half of conversations, reversing a previous shift toward automation.

Geographic adoption remains uneven, following GDP per capita trends. Within the U.S., states with more computer professionals show higher usage, though convergence is occurring. Claude succeeds on most tasks but struggles with complexity, mirroring human task horizons. The report suggests AI adoption may follow a curve, with education-focused use in developing markets diversifying into personal use in mature economies.

The findings reshape job exposure assessments by factoring in success rates. Occupations like data entry show high potential for automation, while others see task composition shift. This data provides a new window into AI's economic impact, helping researchers understand which jobs might be affected and how productivity could change. The full dataset is now available for public analysis.