Anthropic's median employee spent 20% more time in meetings over a year, company calendar data show, as the AI developer's expansion and changing job mix added to meeting time.
A separate analysis found that work involving Claude agents occupied a larger share of employees' interactive sessions. For the average employee, that share was 10% in fall 2025 and 22% in spring 2026. Anthropic derived the estimate through an automated assessment of 2,000 one-line chat summaries. The work included preparing agent instructions, checking one agent's output with another and monitoring workflows.
Meeting time rose 16% for technical individual contributors, or technical workers without management duties. Anthropic cited its expansion and a growing proportion of nontechnical jobs, where meetings are more common, as contributors to the overall increase. It did not quantify each factor's effect. The calendar data do not show that using agents caused more meetings.
Anthropic asked an open-ended question in an April 2026 survey; 20% of respondents described moving toward managing Claude agents rather than doing tasks themselves. The result applies to survey participants, not the entire workforce, and does not measure the hours employees spent overseeing AI.
The volume of Claude's output offers another view of the work involved. Anthropic assumed 0.75 words per token and a reading speed of 250 words per minute in estimating that nontechnical groups would need more than half a workweek to read every response from interactive sessions. That is a calculation based on text volume, not a record of how long employees actually spent reading.
Interactive sessions also included long pauses. Agents spent about 80% of elapsed session time waiting for a person to reply, Anthropic found. The figure measures delays within those exchanges; it does not mean employees devoted 80% of their working hours to Claude.
Anthropic's assessment drew on staffing, AI use, output, survey, communication, calendar and hiring data, along with interviews with 17 employees. Its findings describe the company's own workplace and do not establish whether U.S. workplaces generally are experiencing the same changes.
