Anthropic says its expanding use of Claude agents has changed how work is organized inside the company, with fewer junior technical hires and more employees directing, reviewing and coordinating automated systems.
The share of junior employees in Anthropic’s technical departments fell from 21% in early 2025 to 14% later in the comparison period.
Anthropic linked the staffing shift to rising use of coding agents and higher measured output. Agent hours overtook human employee hours in March 2026, and by August, employee-directed agents worked about six times as many hours as their human handlers. Code commits per engineer rose roughly fivefold over six months, while the company also reported increases in contracts processed by legal workers and deals closed by sales staff. Among nontechnical employees without previous code commits, coding-agent sessions increased from two per month in August 2025 to 90 in May 2026.
The company said the gains have made human attention a constraint rather than removing the need for people.
Outside experts disputed whether the internal figures establish a wider labor-market trend. Morgan Frank said job markets exposed to AI had already deteriorated before ChatGPT’s release and that employers may sometimes use AI to justify layoffs caused by other factors. Julián Posada questioned the report’s lack of independent auditing and said its data do not show how employees experience or use the systems. Dylan Baker called the reduced junior hiring and expanded supervision of agents shortsighted, arguing that software development depends on social coordination, architecture and human judgment. Anthropic’s report suggests workers should have a financial stake in successful AI adoption, but its evidence does not establish that the company’s model will become the standard across other industries.