Anthropic’s junior hiring collapsed as AI agents took over

Entry-level technical hiring at Anthropic has cratered. The share of junior workers in the company’s technical departments dropped from 21% to 14% this year, while new junior technical hires fell even faster — plunging from 27% to 9%, according to a workforce study released by the artificial intelligence company. Open positions for those roles have plummeted since December.

The contraction tracks directly with an explosion in automated labor. By August, AI agent hours worked outnumbered human hours six to one at the company. Agents first surpassed total human working hours in March, the report found. Software engineers’ code commits increased fivefold over the past six months.

This data offers one of the first concrete looks at how aggressive internal AI adoption reshapes corporate headcount planning in real time. Anthropic, which employs more than 3,500 people and is nearing a potential $2 trillion IPO that could break records, operates its own flagship model, Claude, giving it access to tools most employers do not yet possess.

The vanishing junior tier arrives as younger workers already anticipate displacement. A survey from recruiter platform iCIMS published earlier this year found that 51% of Gen Z respondents considered AI the greatest threat to their job security. A Deutsche Bank Research survey from last year found that nearly one in five Gen Z workers worried AI would cost them their job within two years, compared to one in 10 Baby Boomers.

Hiring managers at Anthropic are responding by changing what they look for. The report said recruiters increasingly emphasize applicants’ autonomy and judgment relative to specific technical expertise. Technical proficiency alone no longer guarantees a seat when agents handle repetitive coding and email drafting.

Existing employees face constant reassignment. Workers who had been at the company for at least a year were 50% more likely to move across departments each month in 2026 than in 2025, the company claimed. A product design team that previously wrote website copy, including greetings displayed on Claude.ai, stopped doing so this spring. The team rotated into agent prompting and began writing model evaluations instead, the report read. The company said it has made an effort to identify at-risk jobs and reassign employees to other positions.

Supervising these systems now dominates the workday. Employees spend 80% of elapsed time in Claude sessions waiting on human input, meaning the agents finish tasks faster than people can review them. “This acceleration makes human attention the bottleneck for agent output,” the report read. One employee described the experience as managing “an army of Claudes, with every one asking for my attention.”

Sessions devoted to this coordination rose from 10% to 22% of interactive work. Reading everything the agents produce would consume roughly half a non-technical employee’s week. Private, unlogged work done in direct messages or offline conversations is now “invisible, confusing and discouraged in an AI-heavy workplace,” according to the report.

Median meeting time is up 20% from a year ago. Even among technical individual contributors — a group insulated from the company’s broader headcount growth and push to hire non-technical staff — meeting time climbed 16%. Humans coordinate. Agents execute.

The pressure compounds whenever a new model ships. Employees immediately begin tackling harder problems, driven by what one worker called the “eternal pressure at Anthropic to move as fast as possible and use Claude for everything.” Non-technical staff now rely on personal agents to monitor email, parse calendars, compile to-do lists from messages, and brief them before meetings.

Anthropic employees expect AI to complete three-quarters of their work by 2030, according to the study. That figure dwarfs the 28% predicted by other U.S. workers in a separate Anthropic survey. Whether the rest of the economy follows the trajectory of a well-funded AI developer remains an open question, but the internal mechanics are visible. As one employee put it, “that 20% that was strategic becomes 90% of the job.”