Wealthy Knowledge Workers Use Claude to Augment Their Tasks At Home And Work

Anthropic published “The Anthropic Economic Index report: Uneven geographic and enterprise AI Adoption” on September 15, 2025. You can download it here. 

Here is what I found interesting along with my key takeaway’s:

1. Use of Claude for automation (49.1%)  exceeds that of augmentation (47.0%). Replacing tedious workflows with automation sounds like a good idea. Humans will benefit more from using the tools to give themselves superpowers and convenience. If I were an employer, I would be concerned about how much trust and autonomy less capable employees seem to be giving to these tools. 
   

2. Users in the US comprise most of the traffic, but users in Israel, Singapore, Australia, New Zealand, and Korea are more intense users. Authors’ hypotheses lie in these countries’ historical technology innovation, concentration of knowledge workers,  and per capita income - not unrelated. Historically, those populations who embrace technology innovations tend to do better economically. We could be headed towards an even bigger divide - among and within populations.
   

3. Just for fun, the authors reported on what use cases each country over-indexes on. In the US, we over-index on using Claude for comprehensive cooking, nutrition, and meal planning assistance, help with job applications, personal relationship advice, travel planning, and medical/healthcare guidance. Ironically, these are not fun or whimsical topics. They are very serious ones. Do folks lack access to other sources? Are other services too expensive or too generic? Feels like an opportunity.
   

4. “As Claude usage per capita increases, countries shift from automation-focused to augmentation-focused usage.” My hypothesis (and hope) would be that once consumers move beyond experimentation and the “wow factor” of the technology, that their critical thinking skills would kick in. Once that happens, they’ll realize that the outputs aren’t that magical. The tools help one get started and handle routine tasks, but they don’t yet replace humans or the jobs we would do. Keep in mind also, that affluent, well-educated knowledge workers dominate the core base.
   

5. Use of the APIs in the enterprise is fragmented. The top use case in this study (“resolve software development technical issues and workflow problems”) is just 6%. More creative tasks like writing, creating content, and design fall further down the list. Complexity of use cases is rising - makes sense given that the model capabilities are evolving along with user experience with the tools. 

Here are the author’s key findings:

• Use of Claude to “just do my work” (i.e., directive automation) is accelerating and jumped from 27% (of the sample) in 2024 to 39% in V3 of this report (August 2025). Use of Claude for automation exceeds that of augmentation. The authors have two hypotheses - 1) models are better so more dependable / need fewer iterations 2) better models extend what workflows can be automated = labor replacement. 

• Coding dominates (36%) / developer use cases lead in all countries

• Education (9.3% to 12.4%) and research (6.3% to 7.2%) use cases are growing

• Wealthy people are more likely to use Claude

• Low adoption countries over index on using Claude to code plus hand over more autonomy

• High adoption countries lean into Claude to augment what they do 

Notes on the research:

• Percentages are of total sample. Keep in mind that the total sample size is growing. 

• Anthropic AI Usage Index (AUI) - measure of over/under representation

• The study includes API usage - I didn’t include that b/c it leans mostly into automation and internal enterprise use cases like coding. 

• Sample is one million conversations in August 2025 timeframe.