Anthropic employees are describing a notable shift in what their jobs actually look like day to day. Speaking to Fast Company, several staff members said their roles have moved away from hands-on execution and toward something closer to supervision, with one employee putting it bluntly: the job has gone from "actually doing work" to "babysitting Claudes." The accounts offer a candid window into how AI development is reshaping the people building the technology, not just the industries using it.
When the Tool Becomes the Worker
The pattern described by Anthropic staff tracks closely with what Anthropic's own data reveals about how AI is reshaping white-collar work more broadly. Employees say they now spend significant time reviewing outputs, correcting edge cases, and setting up context for Claude to operate within, rather than producing deliverables directly. For some, this feels like a productivity gain. For others, it raises questions about skill retention and job satisfaction when the hands-on parts of a role get handed off to a model.
Key Facts
- Anthropic employees described the shift in a Fast Company report covering the past 12 months of change at the company.
- Staff used the phrase "babysitting Claudes" to describe their new supervisory role over AI agents.
- The accounts cover roles across engineering, research, and operational functions.
- The trend mirrors what Anthropic has observed in external studies of AI's effect on knowledge work.
The experience is not entirely surprising given the pace at which Claude's model family has expanded its capabilities. As models have become more capable of handling extended, multi-step tasks, internal workflows at Anthropic have adapted to match. Employees are increasingly in the role of setting objectives and reviewing results rather than writing every line of code or drafting every document themselves.
"From actually doing work to babysitting Claudes."Anthropic employee, via Fast Company
A Company Watching Its Own Experiment
There is an unusual dynamic at play inside Anthropic. The company is both developing AI systems and serving as a live test case for what happens when those systems get integrated into daily professional life. Employees are, in effect, the earliest adopters of the very tools they help build. That puts them in a position to observe the friction points firsthand, from workflow disruption to the sometimes isolating nature of working alongside an AI agent rather than a colleague. Earlier reporting found that Claude Code left some Anthropic staff feeling isolated at work, a sentiment that now appears part of a wider cultural adjustment happening inside the company.
The broader implications stretch well beyond Anthropic's offices. CEO Dario Amodei has publicly addressed concerns about labor displacement, and his sweeping policy essay called for AI company taxes and universal basic income as potential responses to the disruptions AI will cause in labor markets. The candid accounts from his own employees suggest those disruptions are already arriving, at least in the sector closest to the technology itself.
What makes the Anthropic case worth watching is the speed of the shift. Employees are not describing a gradual transition over several years. They are describing changes that happened within a single year, compressing an adjustment that many industries may face over a longer arc into a much shorter window. Whether the supervision-heavy model of working with AI agents leads to new kinds of skilled roles or simply fewer roles overall remains an open question, one that Anthropic's own workforce is now helping to answer in real time.
