At Anthropic, Junior Jobs Are Shrinking
Set Trending Topics as a preferred source on Google.
The A.I. company is living out what it sells to others: At Anthropic, the developer of the Claude A.I. model, A.I. agents now work many times the hours of their human colleagues, and entry-level hires have become rare. That is the finding of a report titled “How employment is changing at Anthropic,” which the company has just released. It draws on internal data on staffing, A.I. usage, output, calendars and hiring, along with interviews with 17 employees.
Junior Share Falls From 27 to 9 Percent
The shift is most visible in hiring. The share of junior staff among new technical hires fell from 27 percent to 9 percent. Among all employees in technical departments, their share has dropped from 21 percent to 14 percent since early 2025. Job postings for entry-level roles declined sharply starting in December 2025. In April, about a third of machine learning managers said they were hiring fewer people because of A.I., compared with a fifth in software engineering.
Growth has not come to a halt, however. Technical departments kept hiring in absolute terms, while sales, finance and recruiting grew faster. As a result, the technical share of the work force fell from 61 percent to 56 percent.
Agents Work Six Times as Much
In March 2026, the working hours of A.I. agents surpassed those of humans for the first time. By August, the agents directed by individual employees were working six times as many hours as the employees themselves. Engineers produced five times as many code commits as half a year earlier. Output rose outside engineering as well: Lawyers are handling more contracts, accounting more invoices, and sales is closing more deals.
The bottleneck is now human. Interactive Claude sessions spend about 80 percent of their time waiting for user input. Employees outside engineering would have to spend more than half a work week reading everything their agents produce. At the same time, staff spend 20 percent more time in meetings than a year earlier. “Employees increasingly see their work as managing teams of Claudes,” the researchers write.
The economist Peter McCrory, one of the authors, sees Anthropic about a year and a half ahead of the rest of the economy in A.I. adoption. Its experience could therefore serve as an early indicator, he wrote on X. Financially, things have been going well for Anthropic lately: In the third quarter, the company brought in $16.8 billion in revenue (about €15 billion).
The Other Side
Anthropic itself cautions that the report is merely a case study of a single company. The figures also rest on Anthropic’s own definitions of agent hours and productivity. The report leaves open how representative an A.I. lab that uses its own agents especially heavily is for other companies.
Research paints a mixed picture as well. Economists at Stanford University led by Erik Brynjolfsson found that companies hire significantly fewer young people in A.I.-exposed occupations such as software development and customer service. A Danish study, on the other hand, found no difference in entry-level jobs between companies that use A.I. and those that do not. Bharat Chandar of the Stanford Digital Economy Lab therefore concluded last fall that A.I.’s effects on overall employment are “likely small” for now.
