We Changed the Tool Before We Changed the Job

Key findings

• 73% use AI for business operations
• 45% continuously review and improve work as AI evolves
• 9% name culture as a top factor in AI success

AI may be moving into the office faster than the office can move around AI.

A new survey released Tuesday, Sept 8 by Eagle Hill Consulting captures a workplace that has largely crossed the threshold from experimentation to use. Among the senior business decision-makers it surveyed, 73 percent said their organizations use AI for business operations, 72 percent for decision support and analytics, and 71 percent for employee productivity and knowledge work. Leaders also reported gains in productivity, efficiency, work quality and customer experience.

Those numbers make an easy headline: companies are using AI. But the more interesting finding is what has not moved nearly as fast. The software is changing. The job around it often is not.

Eagle Hill found that fewer than half of the leaders surveyed - 45 percent - said their organizations have an established management practice for continuously reviewing and improving work as AI capabilities and business needs evolve. Only 31 percent said employee development for AI is planned as part of broader business and workforce planning. Just 37 percent said leaders review and adjust how work is organized across the enterprise as AI changes the organization.

The adoption number needs context

That is a different problem from adoption. It is the difference between installing a capability and redesigning an institution.

The survey was conducted by Ipsos from Aug. 3 through Aug. 6 among 306 people at the director level or above. All respondents worked at U.S. companies with at least $100 million in annual revenue, and the sample was limited to organizations where AI adoption was already established.

That qualification matters. The findings should not be read to mean that roughly seven in 10 American companies are using AI across their core functions. They describe what AI looks like inside a deliberately selected group of larger organizations that have already moved beyond the starting line.

Inside that universe, however, the structural gap is hard to miss. Organizations can purchase enterprise software, activate licenses and deploy models on a procurement schedule. Changing who owns a decision, which tasks belong to which role, what managers are expected to supervise, how work is evaluated and what employees are trained to do is slower, messier and more political.

A workplace can therefore become technologically new while remaining organizationally old. That creates the "old job plus AI" problem. An employee gets a faster way to research, summarize, draft, analyze or automate a task, but the same approval chain remains in place. A team can produce more, but performance expectations are not redesigned. AI takes over pieces of a workflow, but no one revisits where accountability begins and ends. Training explains the tool without redefining the role.

Culture is not the soft part of the story

The result can look like transformation from the outside while feeling like acceleration inside the same machine. One of the sharpest contradictions in the Eagle Hill findings is how leaders rank the ingredients of AI success. Fifty-two percent named data quality and availability as a top factor, and 47 percent pointed to technology platforms and infrastructure. Only 18 percent named work redesign. Just 9 percent identified culture.

Then the survey asked about culture directly.

Eighty-four percent of leaders reported at least one cultural factor limiting their organization's success with AI. The most common barriers included employees hesitating to change established ways of working, cited by 30 percent; AI governance processes slowing experimentation and learning, cited by 27 percent; and short-term priorities crowding out experimentation, cited by 26 percent.

In other words, culture can be treated as a secondary concern right up until it becomes the reason the technology cannot move.

That does not mean governance should disappear or that every employee should be free to experiment without controls. It means the design of governance becomes part of the operating model. A policy that is clear, fast and proportionate can enable responsible use. A policy that is vague, slow or detached from actual work can turn into another queue.

The same is true of management. AI can shorten the time required to do a task without shortening the time required to get permission, coordinate across teams, resolve ownership or secure a decision. When that happens, the bottleneck simply moves.

Eagle Hill is not alone in identifying that mismatch. Microsoft's 2026 Work Trend Index described a "transformation paradox" in which employee AI capability can outpace an organization's readiness to support it; Microsoft reported 31 percent of AI users in a misaligned state. A March World Economic Forum report argued that realizing AI's full value requires redesigning how work is performed, decisions are made and operating models are structured. McKinsey made a similar point in July, writing that many companies are using AI to accelerate existing activities while leaving governance, teams and capabilities largely unchanged.

The job can get better - if someone redesigns it

The recurrence of that idea across separate 2026 studies is important. The difficult phase of enterprise AI may not be getting the technology into the building. It may be deciding what the building is for once the technology is there.

The Eagle Hill survey also contains a more optimistic signal. Leaders said AI is increasing employees' opportunities to build expertise, focus on higher-value work and perform creative or intellectually engaging work. Eighty-nine percent reported more opportunity to build expertise, 88 percent more opportunity to focus on high-value work and 87 percent more opportunity for creative or intellectually engaging work. Fifty-eight percent said employee burnout had decreased because of AI.

Those findings deserve a caveat of their own: they are the perceptions of senior decision-makers, not a direct survey of employees. Still, they suggest that the workplace argument around AI does not have to reduce to replacement versus preservation. Work can expand as well as contract.

But that outcome is not automatic. If AI saves an employee three hours and the organization simply fills those three hours with more of the same work, the job did not become more meaningful. It became denser. If AI allows one person to perform tasks that previously crossed several roles, the organization has to decide whether that is empowerment, scope creep, a new role or a new risk. If junior employees offload foundational tasks before they have learned how to judge the output, training may need to change even if productivity rises.

The real implementation question

These are questions about job architecture, not software settings. The survey suggests that many organizations now have enough AI to discover a harder truth: a tool can be adopted without the organization truly adapting to it.

That distinction will matter more as AI moves from optional assistance into routine workflow. The more capable the systems become, the less sustainable it is to treat them as another app added to an unchanged job description. Managers will have to decide what employees should stop doing, not only what they can now do faster. Leaders will have to determine where human judgment is required, where AI can act, who is accountable for the combined result and how performance should be measured when the unit of work itself has changed.

Installing AI can happen under a technology budget. Redesigning work means negotiating authority, responsibility, incentives, risk and expectations. One is implementation. The other is organizational change.

So the useful question for companies may no longer be, "Do we have AI?" It may be: "What did we change because we have it?"

If the answer is the same job, the same approvals, the same incentives and the same expectations - only faster - then the tool changed and the organization did not.

SOURCE NOTES

• Eagle Hill Consulting / PR Newswire, Sept. 8, 2026: New Eagle Hill Consulting Research Finds AI Is Reshaping How Organizations Work, But Leadership and Culture Lag Behind
• Microsoft WorkLab, May 5, 2026: 2026 Work Trend Index: Agents, human agency, and the opportunity for every organization
• World Economic Forum, March 16, 2026: Organizational Transformation in the Age of AI: How Organizations Maximize AI's Potential
• McKinsey & Company, July 7, 2026: The operating model advantage: Why AI winners are rewiring their organizations