On Friday, October 16, 2026, Anthropic published an extensive research report on how AI agents inside the company are changing employment and the organization of work. The report, “How employment is changing at Anthropic,” was prepared by Nathan Wilmers, Regina Catipon, Maxim Massenkoff and Peter McCrory, based on the company's internal employment and workflow data. The report notes that AI agents are becoming an integral part of everyday work, and this process affects everything from hiring policy to employees' daily schedules.

How the Report Was Prepared

The report was written by four researchers — Nathan Wilmers, Regina Catipon, Maxim Massenkoff and Peter McCrory. The authors drew on data from the company's internal systems: hiring statistics, job postings, employees' calendar entries, change histories in code repositories, and telemetry on AI agents' working hours. The report covers the period from the start of 2025 to August 2026. Anthropic presents the document as an early indicator of how AI technologies may change the labor market in the future.

The Share of Entry-Level Employees Declined

One of the report's most notable findings concerns the declining need for entry-level specialists. Since the start of 2025, the share of entry-level employees in the company's technical departments has fallen from 21% to 14%. Among newly hired specialists, their share has fallen even more sharply — from 27% to 9%. The number of job postings for entry-level positions has dropped sharply since December 2025.

Over the same period, the share of technical departments in the company's total headcount fell from 61% to 56% within a year. The report attributes this to brisk hiring in sales, finance and recruiting outpacing the rate of technical hiring. In other words, while the company keeps growing, most of the growth is in non-technical areas.

Agents' Working Hours Are Six Times Human Hours

According to measurements cited in the report, by August 2026 the working hours of interactive AI agents inside Anthropic were six times those of employees. The agents' total working hours first surpassed human working hours in March 2026, and the gap has been widening ever since.

“Human attention and judgment are becoming the scarcest bottleneck for agent-produced output,” the Anthropic report says.

The authors estimate that employees in non-technical roles would have to spend nearly half of the work week just to read all the text produced by Claude agents. On this basis, the report's authors write that the role of humans is shifting from direct execution toward reviewing agents' work and taking responsibility for the results.

Coding Speed Up Fivefold, More Meetings

The growth is even more striking in engineering metrics: the number of code changes (commits) per engineer has grown fivefold compared with the pace six months earlier. The company compared the share of reverted changes and code deleted to fix bugs — the analysis found no difference in quality between agent-written and human-written code.

At the same time, TechCrunch reported on October 17, 2026, that according to calendar data, meeting time per employee grew 20% compared with the previous year. For technical specialists, the increase was 16%. Anthropic attributed the rise to the growing importance of “human trust and responsibility”: the more work agents do, the more coordination is required to discuss and approve their results.

Employee Survey and Early Warning System

According to an internal survey from April 2026 cited in the report, 27% of employees said their productivity had increased thanks to Claude. At the same time, 20% of respondents noted that their work had changed from “doing the work directly” to “managing Claudes”.

Anthropic also announced that it has introduced an early warning system that tracks jobs that could be at risk of disruption. As the content of jobs changes, the company is prioritizing the internal redeployment of employees. In conclusion, the company presents its experience as an early indicator of how AI technologies may change the organization of work and labor demand in the future.