The AI Restructuring Test: Are You Removing Work—or Just People?

Before the spreadsheet becomes the strategy

Imagine walking into a leadership meeting with a deceptively simple proposal: the company will remove 20% of its positions while investing more heavily in AI.

The spreadsheet may be persuasive. The strategy is not—not yet.

Before approving the plan, someone should ask the question that separates transformation from ordinary cost-cutting: Which work is disappearing?

That question matters after Monday.com announced in July that it would reduce its global workforce by about 20%, affecting roughly 630 employees. The work-management software company described the move as part of a leaner, more focused model aligned with its AI Work Platform. Its leaders have also insisted the decision is not simply about replacing people with machines or expanding margins. The company plans to keep hiring in selected strategic areas.

Monday.com may ultimately prove its thesis. But its decision exposes a much larger executive challenge. AI has become a compelling explanation for restructuring before many organizations can demonstrate exactly how the work itself has changed.

The arithmetic arrives before the operating model

The attraction is obvious. If AI enables smaller teams to produce more, companies should be able to redirect spending, simplify management, and move faster. Monday.com has reported encouraging internal results: during its May earnings call, the company said AI had increased output per developer by 32% and reduced time to market for products by 38% since 2025.

Those numbers sound like permission to resize the organization. They are better understood as the beginning of an investigation.

A faster development team does not automatically mean the company needs fewer people everywhere. It may mean engineers can spend more time on architecture, customer problems, security, or new products. It may expose bottlenecks in product approval, sales enablement, implementation, or customer support. Productivity in one activity does not tell leaders how value moves through the entire business.

The same caution applies beyond Monday.com. IBM recently told investors it is using AI and automation to drive productivity while also investing heavily in commercializing innovation. That combination—efficiency and reinvestment—is strategically different from treating labor removal as the primary measure of AI success.

Start with the work that will no longer exist

A credible AI restructuring should be explainable at the level of work, not slogans.

Suppose an organization plans to eliminate 40 positions in customer operations. Leaders should be able to identify the inquiries an AI system will resolve, the decisions it is authorized to make, the exceptions that remain with people, the expected volume, the accuracy threshold, and the escalation process when the system fails. They should know whether the eliminated administrative work is actually gone or merely transferred to managers, customers, or the employees who remain.

Without that evidence, the company has not redesigned work. It has reduced capacity, and assumed technology will absorb the difference.

This distinction is especially important because enterprise adoption is still uneven. A study released on August 12 analyzed more than 17 million ChatGPT Enterprise messages across more than 1,500 organizations. It found broad use across functions and seniority levels, but wide differences in the speed, breadth, and purpose of adoption. The authors concluded that organizations are still learning how to integrate AI into workflows.

That is not evidence that AI lacks value. It is evidence that the path from individual productivity to an operating-model change is neither automatic nor uniform.

Four questions before the headcount target

Executives considering an AI-era restructuring should require four forms of proof.

1. Task proof. What work is being automated, reduced, or eliminated? Name the tasks, their current volume, their owners, and the time they consume.

2. Workflow proof. What happens before and after the AI performs its part? A brilliant agent inside a broken process can simply produce errors faster. Map the handoffs, decision rights, data access, controls, and exceptions.

3. Value proof. Which business outcome improves? Lower cost may be one answer, but it should not be the only one. Look for faster customer resolution, greater throughput, better quality, reduced risk, or new revenue capacity.

4. Capacity proof. Who handles the work when the model is wrong, unavailable, or overwhelmed? Calculate the human capacity required for review, escalation, customer empathy, governance, and improvement. The remaining team cannot be treated as an unlimited shock absorber.

This is also where a small controlled release is more useful than a sweeping promise. Redesign one important workflow, establish a baseline, introduce AI with clear authority and controls, and compare the results. If cycle time, quality, customer experience, and employee workload improve together, leaders have evidence for expansion. If only labor costs improve, the organization may be borrowing performance from the future.

The real transformation may be a skills shift

The most responsible outcome will not always be “no layoffs.” Markets change, business models mature, and some work genuinely disappears. But AI transformation often changes the mix of capabilities before it determines the ideal workforce size.

Companies may need fewer people to perform repetitive coordination and more people to redesign processes, prepare data, supervise agents, investigate exceptions, manage customer relationships, and translate business objectives into reliable systems. Those capabilities do not simply appear when a budget line is removed.

The strongest AI strategy, therefore, connects workforce decisions to a visible capability plan: what the company must learn, what it must retain, what it must recruit, and what it must stop doing. That is a richer leadership exercise than asking each function to cut 15% and “use AI” to compensate.

Monday.com’s restructuring will be judged over time by product adoption, customer outcomes, execution speed, employee load, and durable growth—not by whether it completed the reduction. Other leaders should apply the same standard to themselves.

AI can justify a different organization. It cannot substitute for designing one.

Questions for executives

1. Can we identify the work that will disappear before naming the positions that will?
2. What customer, quality, risk, and employee-workload measures must remain healthy after the change?
3. Which capabilities must we build or retain for AI-enabled workflows to succeed at scale?