The specifics:
Employees at Anthropic claim that they now utilize Claude for 60% of their work and that their productivity has increased by 50%, which is about twice as much as it was a year ago.
Dashboards, cleanup, and trials that weren't worth the manual labor accounted for more than one-fourth of AI jobs.
Up from 10 activities six months ago, Claude Code now strings together about 20 processes before requiring human input, enabling engineers to transfer more complicated procedures.
A number of respondents expressed discomfort despite the improvements, with one stating that it "feels like I'm coming to work every day to put myself out of a job."
Dashboards, cleanup, and trials that weren't worth the manual labor accounted for more than one-fourth of AI jobs.
Up from 10 activities six months ago, Claude Code now strings together about 20 processes before requiring human input, enabling engineers to transfer more complicated procedures.
A number of respondents expressed discomfort despite the improvements, with one stating that it "feels like I'm coming to work every day to put myself out of a job."
According to this survey, Anthropic's whole 4.5 family has since released and probably increased productivity even further. It's intriguing to see how a frontier lab is utilizing its own AI technologies, but employee worries indicate that even they are vulnerable to some of the most significant work-related risks facing the sector.