Your Salary Could Become a Transaction: How AI Is Changing the Way Companies Pay for Work

Last month, a mid-sized insurance firm in Ohio let 50 service contracts expire and replaced them with a single API key. It was the moment the “seat” died. For nearly a century, the basic unit of white-collar work has been the physical or digital space occupied by a salaried employee working roughly 2,000 hours a year. To a CFO, that seat represented a fixed cost: a predictable bundle of salary, benefits, and overhead.

But as of April 2026, that fixed-cost model is giving way to a variable-cost reality. The “seat” is being replaced by the “run.” In this new economy, intelligence is no longer rented by the year; it is purchased by the task.

The economic shift is evident in the margin improvements seen across standard customer support operations. According to industry reports, AI agents now handle resolutions at a fraction of the cost of traditional staffing. More importantly, when demand fluctuates, the cost of these digital agents scales precisely with usage, eliminating the expense of idle capacity.

Source: MainStreet Research / Klarna Data, 2026

This is the “seat-to-run” shift, and it is reshaping the payrolls of the global knowledge economy. The transition reflects a broader move toward “Service-as-Software,” a concept highlighted by Sequoia Capital. Instead of selling tools that humans use to perform work, software companies are increasingly selling the work itself.

The Efficiency Arbitrage

The economic logic of this transition is most visible in industries where work is high-volume and routine. Customer support has served as a primary indicator of this trend. In February 2024, the Swedish fintech company Klarna disclosed that its AI assistant had performed the work of 700 full-time human agents in its first month of operation, saving an estimated $40 million annually.

The company reported that it was able to handle two-thirds of its customer service chats through AI without a corresponding increase in headcount. This has become a benchmark for enterprises seeking to decouple revenue growth from headcount growth.

The pressure is now extending to the Business Process Outsourcing (BPO) industry, a $250 billion global market. Back-office tasks like data entry and document processing are being automated as the cost of processing information collapses. While the cost of processing units of text was once a significant barrier, data from early 2026 indicates that the price for high-capability model outputs has reached a level that makes manual data processing economically uncompetitive in most developed markets.

Where the Headcount Is Moving

The shift is now climbing the corporate ladder. In legal services, automation is reshaping research and drafting roles. While high-level strategy remains human-led, the preparatory work once reserved for junior staff is increasingly automated.

In the world of high finance, buy-side firms are seeing substantial productivity gains. Entry-level analysts are now using agents to synthesize earnings calls and model financial scenarios in seconds. This has led to a noticeable cooling in entry-level hiring cycles for administrative and coordination roles, as highlighted in recent labor market reports from PwC.

White-collar job openings reached a 10-year low in late 2024 and have struggled to recover as companies transition from hiring junior staff to deploying AI task-runners. U.S. labor costs remain a primary driver for this adoption, as firms seek to reduce overhead and improve margins through technical integration.

The Resilience of Human Capital

Despite the rapid adoption of AI agents, large swaths of the economy remain resistant to the “run” model. These holdouts generally fall into three categories: physical presence, fiduciary accountability, and high-stakes relationship capital.

Roles requiring physical movement or sensory judgment—from plumbing to bedside nursing—cannot yet be reduced to a software run. More importantly, work where mistakes have non-linear downsides remains a human domain. In large-dollar negotiations or crisis response, the value of a human who can be held legally and morally accountable outweighs the cost savings of an agent.

There is also a regulatory moat. Compliance-heavy roles in healthcare and finance often require specific human audit trails. In these sectors, the law currently mandates a “human in the loop” for liability purposes. Until the legal framework for AI accountability evolves, these roles will likely remain salaried positions. Industry veterans in M&A and corporate law continue to emphasize that for a $10 billion closing, the “human guarantee” is a non-negotiable requirement that software cannot satisfy.

The Bimodal Wage Effect

For those who remain in the workforce, the “seat-to-run” shift is creating a K-shaped divergence in pay. The middle is hollowing out, but the top is thriving.

Data from WorldatWork indicates a widening pay gap between the highest and lowest earners in white-collar sectors. This is the “AI-native” premium: a phenomenon where workers who can supervise fleets of agents earn significantly more because they produce multiples of the output of their peers.

According to PwC’s Global AI Jobs Barometer, workers with AI skills commanded a 25% wage premium in 2025.

This dynamic suggests that while fewer people may be hired, those who are will be expected to act as “managers” of agents rather than “doers” of tasks. On a typical Tuesday, an agentic worker might spend their morning reviewing the outputs of fifty separate research “runs,” refining prompts, and auditing agent decisions for accuracy. The worker becomes a force multiplier for the firm, shipping significantly more volume than was possible under the traditional “seat” model.

The Global Race for Adoption

The United States continues to lead in private AI investment. According to the Stanford HAI 2024 report, U.S. private investment in AI reached $67.2 billion, significantly outpacing other nations. However, global integration varies by region.

In Japan, where a shrinking population has created a severe labor shortage, AI adoption in finance has received positive performance ratings from employers in OECD surveys. Japan’s experience suggests that AI agents may be a solution to demographic decline, whereas in the U.S., the focus remains largely on margin expansion and cost reduction. The EU, meanwhile, is facing hurdles in retraining its existing workforce to meet its goal of 20 million ICT specialists by 2030.

The Shrinking Firm

The most profound long-term effect of the shift from seats to runs is the structural shrinkage of the company. Revenue growth per employee is now significantly higher in AI-exposed industries than in those with low adoption, according to PwC data.

We are entering the era of the small-team conglomerate. Venture capital funding is increasingly directed at small, agile teams that are hitting high annual recurring revenue milestones with fewer than 20 employees. In a previous decade, such milestones would have required a staff of 100 or more.

Large corporations are responding with restructuring efforts that emphasize attrition-based shrinkage. Major tech firms have indicated they expect to maintain or reduce total corporate workforce levels as generative AI agents handle more routine workflows.

The New Middle Office

As old roles vanish, a new “middle office” is emerging to manage the infrastructure of the agent economy. These roles focus on the reliability and safety of the AI “runs” that companies now rely on.

Titles such as “Eval Engineer” and “Prompt Ops” are becoming common. Demand for “Trust and Safety” auditors—humans who review AI decisions for bias or error—is currently outstripping supply. These positions require a blend of technical literacy and traditional managerial oversight.

This shift is supported by the improvement in AI reliability. The Stanford HAI report notes that the performance of AI agents handling complex, multi-step tasks has seen qualitative improvements year-over-year. As agents become more capable, the human role moves from “doing” to “auditing.”

The Future of the Payroll

The transition from seats to runs represents a structural shift in employment. For the worker, it offers the potential for higher wages for the highly skilled. For the employer, it offers an elastic, efficient way to manage intelligence.

However, for the average white-collar worker, it introduces a new kind of precariousness. When value is measured in “runs” rather than “years,” the stability of the traditional salary may become a relic of the past.

The challenge for policy-makers in 2026 is ensuring the gains from this collapse in the cost of intelligence are shared by the humans who remain. In the modern office, the lights are still on, but the person in the next cubicle might not be a person at all—and their work is being accounted for by the click.