Anthropic’s internal employment data show that junior technical hiring has slowed as AI agent use has expanded. By August, interactive agents logged six times as many work hours as their human handlers. A report by Nathan Wilmers, Regina Catipon, Maxim Massenkoff and Peter McCrory draws on employment, AI-use, output, survey, communications, calendar and hiring records, plus 17 employee interviews.

The share of junior employees in technical departments fell from 21% to 14% since early 2025, while junior technical hires dropped from 27% to 9%. Technical departments’ share of total employment declined from 61% to 56% over the last year, although technical hiring continued growing in absolute numbers as sales, finance and recruiting expanded faster. In April, a majority of surveyed managers said AI had changed the skills they sought; around a third of machine-learning managers and a fifth of software-engineering managers said they were hiring less because of AI.

Coding agents also spread into non-technical work: sessions by non-technical employees rose from 2 per month in August 2025 to 90 in May 2026. Meanwhile, the share of technical-department employees who made no code commits in the prior 30 days fell by almost half through April 2026. The engineer-to-product-manager ratio reached around 17 to 1, and managers’ code commits doubled. Employees who had been at Anthropic for at least a year moved between departments 50% more often per month in 2026 than in 2025.

Agent hours surpassed human work hours in March 2026; by August, employee-directed agents logged six times the hours of their handlers. The report says engineers’ code commits reached 5x their level six months earlier, while output also rose among lawyers, accounts-payable staff and salespeople. A comparison followed 192 engineers who adopted coding agents against 74 who did not. Adopters added more code after adoption, but the researchers caution that tool adoption was not random and more code does not necessarily mean better code or business impact. They found no difference in the share of code lines later removed or reverted when comparing agent-written and employee-written code.

Sales representatives who adopted agents closed deals with more clients. One account executive said automations replaced around 20% of prior tasks; another said they could speak with 30% more customers. A lawyer used Claude to scan risks, triage legal questions and prepare copy for review, while a recruiter used it for candidate sourcing and application tracking. The report says human attention has become a bottleneck: reading all agent output would take non-technical employees half their work week, prompting staff to build monitoring and visibility tools. Hiring managers increasingly prioritize autonomy and judgment over specific technical expertise. Anthropic has established an early-warning system for jobs at risk of disruption and prioritized moving employees into other roles as jobs change.

Anthropic’s internal employment data show that junior technical hiring has slowed as AI agent use has expanded. By August, interactive agents logged six times as many work hours as their human handlers. A report by Nathan Wilmers, Regina Catipon, Maxim Massenkoff and Peter McCrory draws on employment, AI-use, output, survey, communications, calendar and hiring records, plus 17 employee interviews.

The share of junior employees in technical departments fell from 21% to 14% since early 2025, while junior technical hires dropped from 27% to 9%. Technical departments’ share of total employment declined from 61% to 56% over the last year, although technical hiring continued growing in absolute numbers as sales, finance and recruiting expanded faster. In April, a majority of surveyed managers said AI had changed the skills they sought; around a third of machine-learning managers and a fifth of software-engineering managers said they were hiring less because of AI.

Coding agents also spread into non-technical work: sessions by non-technical employees rose from 2 per month in August 2025 to 90 in May 2026. Meanwhile, the share of technical-department employees who made no code commits in the prior 30 days fell by almost half through April 2026. The engineer-to-product-manager ratio reached around 17 to 1, and managers’ code commits doubled. Employees who had been at Anthropic for at least a year moved between departments 50% more often per month in 2026 than in 2025.

Agent hours surpassed human work hours in March 2026; by August, employee-directed agents logged six times the hours of their handlers. The report says engineers’ code commits reached 5x their level six months earlier, while output also rose among lawyers, accounts-payable staff and salespeople. A comparison followed 192 engineers who adopted coding agents against 74 who did not. Adopters added more code after adoption, but the researchers caution that tool adoption was not random and more code does not necessarily mean better code or business impact. They found no difference in the share of code lines later removed or reverted when comparing agent-written and employee-written code.

Sales representatives who adopted agents closed deals with more clients. One account executive said automations replaced around 20% of prior tasks; another said they could speak with 30% more customers. A lawyer used Claude to scan risks, triage legal questions and prepare copy for review, while a recruiter used it for candidate sourcing and application tracking. The report says human attention has become a bottleneck: reading all agent output would take non-technical employees half their work week, prompting staff to build monitoring and visibility tools. Hiring managers increasingly prioritize autonomy and judgment over specific technical expertise. Anthropic has established an early-warning system for jobs at risk of disruption and prioritized moving employees into other roles as jobs change.