Anthropic's junior technical ranks shrink as it adopts its own AI
Juniors down from 21 to 14 percent of technical staff since early 2025 with new junior hires at 9 percent, meeting time up a fifth at the company selling the automation
Images
techcrunch.com
Anthropic published a report on Friday on what its own AI has done to its own workforce. The share of junior employees in the company's technical departments has fallen from 21 percent to 14 percent since early 2025, and the share of technical hires at junior level has fallen from 27 percent to 9 percent.
Technical teams now make up 56 percent of the workforce, down from 61 percent. Anthropic attributes part of the decline to business functions expanding as the company scales, and part to its own coding tools, which increasingly let non-technical employees build software without engineers. In an April survey, 27 percent of employees said Claude had made them more productive, and 20 percent said their job had shifted "from actually doing work to managing Claudes." One engineer compared the arrangement to babysitting: every child brings concerns with an equal demand for your attention, and the various Claudes do the same.
The calendar data points the same way. Meeting time per employee is up 20 percent from a year ago, 16 percent among technical contributors, a change the company attributes to the growing importance of human trust and accountability; employees described face-to-face coordination as indispensable. The report argues the company is more productive — more code per person, more sales per representative, more contracts processed — while conceding that more code does not necessarily mean better code or business impact. It states it has found no indications that code quality is declining.
The obvious caveat is that a laboratory selling automation has every incentive to publish numbers showing that automation works, and no rival publishes comparable data about itself. But the direction matches what customers report, and Anthropic argues its own workplace is an early indicator of how AI will shift hiring and labor demand across the economy. If entry-level knowledge work disappears anywhere first, it should disappear at the firm that writes the models — and the report says that is where it is disappearing, with juniors going from more than a quarter of technical recruits to under a tenth since early 2025.
The report describes human attention and judgment as the bottleneck on productivity, with Claude producing a deluge of text that needs to be read and organized. The clearest evidence is the calendar: at the company that automated its code, the fastest-growing activity is meetings.