The disruptions caused by AI are now beginning to show up in jobs data.
A new dataset from Pave Data Lab, shared by venture capital firm Andreessen Horowitz (a16z), shows that the share of new hires going into customer support roles has collapsed over the past two years — from 8.3% in Q4 2023 to just 2.88% in Q3 2025. That is a decline of more than 65% in under eight quarters, a trajectory that closely tracks the rapid deployment of AI-powered support automation across the technology industry.
The data, drawn from 386,500 new hires across companies participating in Pave’s real-time compensation benchmarking platform, reflects a structural shift rather than a seasonal dip. The trendline held broadly steady through the first four quarters of the dataset — oscillating between 7.1% and 8.3% — before beginning a steep, uninterrupted descent from Q4 2024 onward. The sharpest drop came in the most recent quarter recorded, suggesting the pace of displacement is accelerating, not plateauing. The data appears to reflects a structural shift, and not a seasonal dip. The sharpest drop came in the most recent quarter, suggesting the pace of displacement is accelerating.
The trend aligns with a string of high-profile corporate disclosures about AI replacing human support workers. Salesforce CEO Marc Benioff revealed that the company cut roughly 4,000 customer support roles after deploying AI agents across its service operations. Swedish fintech Klarna made similarly notable headlines when it disclosed that its AI assistant was handling the workload of 700 employees. Indian e-commerce platform Meesho reported reducing support costs by 75% after replacing human agents with AI bots.
What makes the a16z/Pave chart particularly significant is that it captures these effects not as one-off announcements, but as an aggregate, economy-wide signal. When the share of new hires going into customer support fell from 7.41% in Q4 2024 to 6.2% in Q1 2025, then to 5.81% in Q2, and finally to 2.88% in Q3 — that is dozens or hundreds of companies independently making the same calculation: AI can now do this work at a fraction of the cost.
Customer support has long been one of the largest entry-level white-collar job categories in the technology sector. It has also, historically, been a pipeline into higher-skilled roles in operations, product, and account management. A contraction of this magnitude — if it holds — would have downstream effects not only on employment levels but on career mobility for workers who use support roles as a foot in the door of the technology industry.
The debate over AI and jobs has often been framed in hypothetical terms — speculation about what automation might eventually do to the labor market. The Pave dataset suggests that for at least one large occupational category, the future has already arrived. The remaining question is whether the companies redirecting headcount away from support are creating comparable roles elsewhere, or whether the net effect is simply fewer jobs.