Anthropic Employees Spend 20% More Time in Meetings Over One Year

The middle employee at Anthropic spent 20% more time in meetings than a year earlier. At the same time, the company used Claude agents more often, changing how some employees did their work.

Calendar records show that meeting time went up by 16% for technical employees who do not manage a team. Anthropic grew, and more employees took nontechnical jobs that usually involve more meetings. These changes helped increase meeting time across the company. However, the data do not prove that using agents caused more meetings.

In an April 2026 survey, people could answer a question in their own words. About 20% of those who responded said they were moving away from doing tasks themselves and toward managing Claude agents. This figure applies only to survey respondents. It does not show what share of all Anthropic employees made this change or how much time they spent overseeing agents.

Work involving the direction of agents also made up a larger share of employees' interactive Claude sessions. For the average employee, this share grew from 10% in fall 2025 to 22% in spring 2026. This type of work includes writing instructions, asking one agent to check another agent's work, and watching how groups of agents carry out tasks. Researchers estimated these figures by automatically reviewing 2,000 short, one-line summaries of chats.

Anthropic's interactive Claude sessions often included long waits. During about 80% of the total time in these sessions, agents were waiting for a person to reply. This figure measures waiting during conversations that go back and forth between people and agents. It does not mean employees spent 80% of their working hours watching Claude.

The amount of text produced created another challenge. For employees in nontechnical teams, reading every response from interactive sessions would take more than half a workweek. This estimate assumes people read at 250 words per minute and that each token equals 0.75 words. It estimates how much text there is to read, not how long employees actually spent reading it.

An engineer interviewed about these changes compared the experience to “an army of Claudes, with every one asking for my attention.”

The study brought together data on employees, AI use, output, surveys, communications, calendars and hiring, along with interviews with 17 employees. The findings show what happened at one company where AI tools are part of everyday work. They do not show whether workplaces across the United States are experiencing the same changes. The study also did not measure how much meeting time increased because the company grew and how much increased because the types of jobs changed.