AI is Creating Capacity. Will Leaders Cut Costs or Build a Growth Engine?

We are past the point of asking whether companies will adopt AI. The more immediate question is whether leaders can turn access to AI into organizational transformation.

I have seen plenty of companies give employees a $20-a-month AI budget and a directive to “go experiment,” then wait for transformation to materialize. That is not an AI strategy. It is a software allowance with optimism attached. Employees may discover ways to complete individual tasks faster, but isolated efficiencies do not necessarily change workflows, roles, or how the organization operates.

Recent research makes that distinction visible. A six-month randomized field experiment involving 7,137 knowledge workers at 66 companies found that employees with access to Microsoft 365 Copilot spent 1.3 fewer hours per week on email. Regular users saved 3.6 hours. Yet the researchers found little movement in work that required coordination, including meetings and responsibilities. Individual efficiency did not automatically become organizational transformation.

From Experimentation to Transformation

Transformation requires more than access to a tool. Leaders must point experimentation toward consequential business problems, identify which gains are worth scaling, and redesign work around what the technology makes possible. The point is not to eliminate bottom-up experimentation. It is to stop confusing a collection of clever use cases with a strategy. A hundred isolated efficiencies will not transform an organization if they never converge on a consequential workflow, role or business result.

Rashmi Badwe, EVP and COO of TIAA Wealth Management and Advice, described a more deliberate approach when she was a recent guest on The CEO Daily Brief. Rather than pursue a hundred disconnected use cases, TIAA has concentrated development on a small number of priority roles and examined the sources of toil across them. Badwe offered adviser meeting preparation and estate-planning analysis as examples of the opportunity. Rather than pursue a hundred disconnected use cases, TIAA has concentrated development on a small number of priority roles and examined the sources of toil across them. Badwe offered adviser meeting preparation and estate-planning analysis as examples of the opportunity.

In one test, AI transformed an estate-planning task from a two-hour job to a 30-minute one. But the efficiency was only the beginning of the leadership decision. TIAA could capture it as lower cost (read layoffs) or reinvest the capacity so the same talent could serve more clients. Leaders chose growth, including expanding services to customers who previously could not access them.  

As Badwe put it, efficiency is “not all just cost out.” You’ve got to redesign the work. When her team was designing roles for TIAA’s evolving wealth-management business, she required a proposal that did not use a single existing role name. Existing titles pulled people toward existing assumptions. As AI absorbs more routine analysis, she sees advisers focusing more on relationships, family dynamics, and the emotional biases behind financial decisions. Perhaps the advisers of the future will include more psychology majors. Or theater majors. It did me well!

The Real Question: What Is the Capacity For?

That is the sequence many organizations are missing. First, create enough structure for experimentation to produce organizational value. Once AI creates capacity at that level, a second question emerges: What is that capacity for?

SHRM’s 2026 survey of 116 CEOs captures the tension. Forty percent identified adopting AI as a top priority, while 31% selected revenue growth. Those are not separate priorities. AI adoption is not a business outcome. Its value comes from what leaders do with the capacity it creates.

Efficiency Can Create More Than Cost Savings

But cost reduction and growth are only two possible destinations for efficiency. Some AI applications create a different kind of value. When I spoke at the GM-United Auto Workers Safety Conference last year, I learned how GM uses AI and advanced software to minimize ergonomic stressors and improve workplace safety. That is not merely capacity being recovered. It is safer work becoming possible.

The boundary is not always clean. One of our clients, Idaho National Laboratory, uses AI and machine learning to accelerate reactor design and licensing, reduce project risks, and protect critical infrastructure. Its 2025 collaboration with Microsoft uses AI to draft reports required for nuclear licensing, with humans verifying the documentation. Drafting reports faster is an efficiency. Shortening a bottleneck in bringing new nuclear technology online is the mission outcome that efficiency can serve.

IKEA shows the leadership work required. From 2021 to 2023, its chatbot Billie resolved approximately 47% of the customer inquiries it received, while 8,500 call-center employees were reskilled in areas including remote interior design, digital sales, relationship-building, and complex customer service. When I interviewed IKEA CEO Jesper Brodin, he told me IKEA was educating 500 senior leaders about AI’s opportunities, risks and leadership implications. Capacity does not redirect itself. Leaders must understand the technology well enough to connect automation to new work and a defined result.

Define What "More With Less" Actually Means

This is why the familiar promise to “do more with less” is dangerously imprecise. Cost reduction is a legitimate use of AI, and sometimes workforce reductions will be part of the equation. But leaders should define both words before making that promise. Less labor cost, time, error, risk, or harm? More transactions, customers, access, revenue, innovation, or safety? 

In our work with Suncoast Credit Union, I have seen that choice expressed simply: serve more members with the same number of employees (rather than the same number of members with fewer employees). Clutch, its AI vendor, reports that Suncoast’s collections agent, Emma, makes more than 17,000 calls a day, allowing employees to focus on members and problems requiring human judgment. Call volume shows the scale of automation. Serving more members is the result.

Efficiency is not the outcome. It is a resource. The leadership question is: What are we going to spend it on?