One year after examining the impact of artificial intelligence (AI) in various industries, attention is turning inward to how AI is transforming work within AI companies themselves. Anthropic, a prominent AI firm, recently released a detailed internal report analyzing employment patterns, AI usage, productivity metrics, and employee experiences. The findings provide insight into how AI is reshaping roles and workflows in the very sector developing these technologies.

Anthropic’s report draws from a combination of quantitative data — including hiring records, calendar usage, and internal communications — as well as qualitative input from 17 employee interviews. It reveals three key trends shaping the company’s workforce and operations.

First, AI appears to be influencing hiring practices, particularly at junior technical levels. Despite experiencing rapid growth, Anthropic reports hiring fewer employees per dollar of revenue compared to the first five years of earlier tech giants like Meta and Google. More notably, the proportion of junior technical staff has declined from 21% to 14% since early 2023, with new junior hires dropping from 27% to just 9%. Anthropic attributes this shift directly to the increasing capability of its AI models, which are now able to perform many tasks traditionally assigned to junior employees, thereby reducing the need for such hires. This trend aligns with broader economic data indicating a downturn in entry-level hiring, although other factors may also be contributing.

Second, employees at Anthropic are actively working to automate aspects of their own jobs by developing and relying on AI agents. For example, executive assistants are building AI tools to handle routine administrative tasks such as calendar and inbox management, enabling them to concentrate on more complex responsibilities. Similarly, teams that once created content directly are now guiding AI agents to generate written material, with human roles shifting toward overseeing AI outputs and conducting evaluations. This dynamic reflects an “eternal pressure” within the company to leverage its AI systems extensively to accelerate workflows.

Third, the nature of work has evolved from hands-on execution to managing a multitude of AI agents that demand constant human oversight. One engineer likened managing these AI collaborators to babysitting multiple children, each expressing urgent requests equally. This shift has increased the need for coordination among employees, leading to more frequent meetings and a workday punctuated by transitions between managing AI and interpersonal collaboration. The report acknowledges that this environment is intense and requires workers to adapt to continuously disrupted routines, often involving rapid interpretation of AI-generated outputs.

Despite these changes, Anthropic emphasizes the importance of stock ownership and shared financial incentives to maintain employee engagement amid the shifting landscape. However, the company also cautions against generalizing its experience as representative of all workplaces, noting its unique position as both an AI product creator and user, as well as its accelerated growth and highly specialized staff.

Commentary accompanying the report raises questions about the implications for hiring and education more broadly. While some roles previously filled by juniors may be supplanted by AI, the development of future managers and senior leaders might still require traditional pathways that involve hands-on junior-level experience. It remains uncertain whether the current reduction in entry-level hiring will persist or eventually rebound.

Furthermore, hiring managers increasingly seek evidence of creative AI tool use in applicants, suggesting a shift in skills required for employment. This development has sparked discussion about how educational institutions might assess student learning, balancing traditional offline evaluations with AI-assisted work to accurately measure both knowledge and AI literacy.

There is also ongoing debate about the relationship between domain expertise and transferable skills such as judgment and autonomy. Some experts argue that strong judgment depends on deep domain knowledge developed over time, often through junior-level work that AI may partially replace.

Anthropic’s experience provides a nuanced perspective on how AI is reshaping work inside AI companies, highlighting both opportunities for efficiency gains and new challenges in workforce management and skill development. Whether these patterns will extend to other sectors remains an open question as organizations continue to adapt to the evolving capabilities of AI.