Data Insight ยท Anthropic Economic Index
Anthropic publishes an open dataset on how people actually use Claude, built from anonymized, aggregated conversations matched to occupational tasks. We charted the highlights. Figures are a snapshot from May 2026 and describe observed usage, not jobs or employment.
51%
Augmentation
Conversations where the person stays actively involved in the task
43%
Work usage
Share of classified conversations that look like work tasks
121
Countries covered
Countries with a published Anthropic Usage Index
5 hrs to 40 min
Estimated task time
Tasks estimated at ~5 hours alone ran ~40 minutes with Claude
Share of conversations by collaboration style. Augmentation means the person stays in the loop; automation means Claude is directed to complete the task.
Of classified conversations worldwide, the share that looks like work, personal life, or coursework.
Share of global usage by job category. These describe conversations matched to tasks commonly done in each occupation group, not who is using AI. Top 10 of 22 categories shown.
Share of sampled conversations by request topic, worldwide. Top 10 of 20 published topics shown.
The Anthropic Usage Index compares a country's share of Claude usage to its share of working-age population. A score of 1.0 means usage is exactly proportional to population; Australia's 6.4 means over six times what population alone would predict. Top 15 of 121 covered countries shown.
Source: the Anthropic Economic Index, published by Anthropic under CC BY 4.0. Figures are derived from anonymized, aggregated conversation content from a May 2026 snapshot and cannot be tied to individual users. They describe observed Claude usage matched to occupational tasks and are not evidence about employment, job displacement, or automation of jobs.
From high-growth startups to enterprise brands, we help companies of all sizes launch digital products that drive user engagement. We'd love to discuss how we can help you.