AI company Anthropic has published a report saying its workforce and the nature of its work have changed significantly over the past year. The share of junior employees in its technical department fell from 21% in early 2025 to 14%, while meeting time per employee rose 20% from the previous year. The more AI speeds up work, the more time people spend face-to-face. It’s an unexpected combination.
Why the share of junior technical employees has fallen
First, let’s look at the numbers. According to the report, the share of junior employees in the technical department fell from 21% in early 2025 to 14%. The drop was even steeper in hiring: junior employees’ share of new hires in technical roles shrank from 27% to 9%.
The technical department’s share of the company as a whole also fell, from 61% to 56%. Anthropic attributes the shift to business functions naturally expanding as the company grew, and to advances in coding tools that have enabled nontechnical employees to build software themselves.
This figure calls for some caution in interpretation. It reflects just one company, and businesses elsewhere may not follow the same path. Still, it is unusual for a company at the forefront of AI development to disclose this much about how it works, making the report an intriguing source for understanding the direction of change.
It’s easy to imagine that if AI takes over some of the routine coding and preliminary research once handled by junior employees, the entry point for new hires will narrow. At the same time, the fact that nontechnical employees are starting to build software suggests that the boundaries between job functions themselves may be blurring.
It wasn’t just output that increased; so did the time spent supervising
The report also shows that productivity has risen with the use of AI. It says the amount of code produced per employee has increased, sales representatives are closing more deals, and the number of contracts processed has grown. Anthropic acknowledges that more code does not necessarily mean higher quality or greater business impact, but says it has found no signs that code quality is declining.
What’s interesting is how employees feel about it. In an employee survey conducted in April 2026, 27% said the company’s AI, Claude, had increased their productivity, while 20% said the focus of their work had shifted from “doing the work itself” to “managing multiple Claude instances.”
One engineer compared the situation to babysitting. When you have several children, each demands your attention just as intensely as the others. That’s what working with multiple Claudes feels like. Another employee reportedly said there was an “endless pressure” inside the company to move fast and have Claude do everything.
The report also notes that Claude produces such large volumes of text that the human attention and judgment needed to read and organize it are becoming a bottleneck to productivity. The faster the people producing work move, the more strain falls on those checking it. It’s much like a factory line where one stage speeds up and inventory piles up at the next.
Why meetings are increasing in the AI era
Another number to consider is meeting time. An analysis of calendar data found that meeting time per employee was up 20% from a year earlier, and had risen 16% even among individual contributors in technical roles.
Anthropic attributes this to the importance of trust and accountability between people. Employees, it says, repeatedly stress that in-person alignment is essential. Even if AI can instantly produce drafts or implementations, people are responsible for the results, and confirming who is accountable ultimately comes down to conversations between people.
When people talk about efficiency, they tend to picture fewer meetings and more time for individual work. But in workplaces that deeply integrate AI, the share of time devoted to judgment and coordination grows as hands-on work declines—and meetings are where that happens.
Junior hiring is tightening, meetings are increasing, and people are becoming managers of AI. Whether this combination spreads to other companies will depend on how central each makes AI to its operations. But for executives and managers considering AI adoption, Anthropic’s figures offer a useful prompt: think ahead about what will increase and what will decrease in your own organization.
Comments