TL;DR

Anthropic published research on 16 October examining how AI has changed work inside the company. The share of technical hires who were junior fell from 27% to 9%, and by August, AI agents directed by employees were logging six hours of work for every hour put in by the people directing them. Output per person rose across functions, and hiring managers now weigh judgement and autonomy more heavily than technical expertise.

Fewer juniors, more agents

The report draws on internal data on headcount, AI use, output, surveys, calendars and hiring decisions, plus 17 staff interviews. Junior staff fell from 21% to 14% of technical departments since early 2025, and job postings for juniors have dropped especially sharply since December 2025. Engineering and research teams had moved away from recruiting for lower-band technical roles by April.

Anthropic says it takes in two to four times the revenue per employee that Meta and Google did at a comparable stage of their growth. In March 2026, agent working hours, autonomous and interactive combined, overtook human hours.

Code commits per engineer are running at five times their rate of six months earlier. Following 192 engineers who adopted coding agents and 74 who did not, the authors found output rose sharply after adoption, with no difference in how often agent-written code was later removed or reverted. Lawyers, accounts payable staff and salespeople also processed more work.

Attention becomes the bottleneck

Agents mostly wait. Interactive Claude sessions spend around 80% of their time idle, waiting for a human. Even for non-technical staff, reading everything their agents produce would take half the working week. The share of sessions spent on “meta-work”, such as writing specs for multi-agent runs and monitoring pipelines, rose from 10% to 22% between autumn 2025 and spring 2026. Median time in meetings is up 20% on a year earlier.

Anthropic says it now runs an early-warning system that flags roles likely to be disrupted, and it prioritises moving affected staff into other jobs internally. In 2026, staff with at least a year at the company were half as likely again to change department in any given month as in 2025.

Looking forward

Anthropic presents itself as a possible leading indicator, and the authors acknowledge its unusual position, including heavy staff share ownership. Still, the junior hiring figures are the ones UK employers should note. In our view, if entry-level technical roles shrink first where AI use is deepest, UK graduate pipelines face pressure that the CFO headcount expectations we reported on 14 October only hint at. Redeployment schemes like Anthropic’s early warning system are one response other employers could copy.