Anthropic's latest Economic Index report reveals a structural shift in how people use Claude. Between November 2025 and February 2026, task concentration on Claude.ai dropped from 24% to 19%, meaning users are spreading AI across a wider variety of everyday activities rather than clustering around a few high-value tasks.
Wage Slightly Down, Personal Queries Surge
The sample, taken Feb 5–12 (overlapping Opus 4.6's launch), shows coding remains the top use case but its share eroded. Academic queries fell from 19% to 12%, while personal tasks like sports scores, product comparisons, and home repair jumped from 35% to 42%. The average task wage, measured by the median hourly pay of the corresponding U.S. occupation, edged down from $49.30 to $47.90. This aligns with a classic adoption curve: early adopters concentrated on coding, while later entrants bring broader, lower-value use cases.
Automated Trading Workflows See Double Frequency
Two API workflows roughly doubled between sampling periods: business sales and outreach (auto-generated sales content, lead research, cold outreach) and automated trading and market operations (market monitoring, portfolio alerts, investment suggestions). These tasks require minimal human intervention, signaling a move from AI as helper to AI as executor.
Learning by Doing: Veteran Users Shine
For the first time, the report analyzes tenure effects. Users on Claude.ai for over six months are 7 percentage points more likely to use Claude for work and tackle tasks requiring higher education. Even after controlling for task type, model choice, language, and geography, experienced users complete conversations successfully 3 to 4 percentage points more often. The most seasoned users gravitate toward AI research, git operations, manuscript revision, and venture funding; newer users explore haiku composition, sports scores, and party food ideas.
Model Selection: Opus Reserved for High-Value Work
Among paid users, 55% of software coding tasks use the most capable Opus model, compared to 45% for educational tasks. On the API, a $10 increase in task wage correlates with a 2.8 percentage point rise in Opus usage — double the effect on Claude.ai. This suggests users are developing sophisticated cost-performance strategies.
U.S. Convergence Slows, Global Inequality Widens
Per-capita Claude usage across U.S. states continues to converge but at a slower pace. The top five states' combined share fell from 30% to 24%, and the estimated time to full parity extended from 2–5 years to 5–9 years. Internationally, concentration rose: the top 20 countries now account for 48% of per-capita adjusted usage, up from 45%.
The report warns that success disparities could worsen labor market inequality — early, high-skilled users capture more AI benefits, while later adopters face steeper learning curves.