Most AI programs begin with a technology question: What can the tools do?
The more important question is: How will work change if they do it?
As AI moves beyond isolated pilots, it will take on more routine analysis, coordination and decision support. People will remain part of the operating model, but their responsibilities will change. Employees who once performed a task may instead supervise it, review exceptions and intervene when an AI-generated outcome is incomplete or wrong.
Making this transition requires more than tool training. Business leaders need to redesign roles, decision rights, and performance expectations through a deliberate organizational change management program.
Why roles must be redesigned, not merely “AI-enabled”
Adding an AI tool to an existing role is not role redesign. Organizations need to determine which activities will remain human-led, AI-assisted or substantially automated, then reset decision rights, accountabilities, controls and performance expectations.
Some roles will contract. Others will expand. New responsibilities will emerge around data stewardship, output validation, risk monitoring, and exception management. This may not require new departments, but it will require employees to exercise different forms of judgment.
Consider a financial analyst who currently prepares a monthly variance report. In an AI-enabled process, the system may compile the data, identify anomalies, and draft the commentary. The analyst’s job shifts to testing the conclusions, investigating material exceptions, and determining what management should do next. The work becomes less transactional, but the accountability remains. AI may perform more of the work, but management must still assign a person who is accountable for the outcome.
Without that clarity, employees may over-rely on the system, duplicate its work or avoid using it altogether.
Why supervising AI is a learned capability
A generic session on prompting will not prepare people for supervisory work. Employees need to understand the process they are supervising, the business rules the AI is expected to follow, its limitations and the conditions that require human intervention.
Training should be role-specific and grounded in actual workflows. Employees need to practice reviewing outputs, detecting unreliable results, resolving exceptions and documenting overrides. They should know which data the system uses, how poor data affects its output and why a convincing answer may still be wrong.
Supervisors also need clear escalation criteria. When can an employee approve an AI recommendation? When must it go to a manager, subject-matter expert, compliance function or customer-facing employee? Those thresholds should be tested before the technology is scaled.
Competence must then be monitored in live operations. Useful measures include the frequency and quality of human overrides, missed exceptions, escalation accuracy, rework and business outcomes. High intervention rates may indicate poor system performance; unusually low rates may mean employees are accepting outputs without enough scrutiny.
Leaders should assess the system and its human supervisors separately. Otherwise, poor system design can be mistaken for employee resistance or weak supervision can be blamed on the technology. The objective is reliable performance, not maximum adoption.
Why organizational change management belongs in the AI roadmap from the start
The people plan should shape the AI roadmap, not sit beside it as a communications and training workstream. Workforce readiness will constrain both the pace of adoption and the organization’s ability to realize value from AI.
The five-year business strategy should define where the company will compete and how its operating model must evolve. The three-to-five-year AI plan should identify where AI can enable that strategy and which capabilities must be built. For each wave, leaders should identify the affected processes, role changes, skills, decision rights, controls and measures of readiness.
The sequence might look like this:
- Year one: Establish governance, assess workforce readiness, select priority use cases, and define the first supervisory roles.
- Years two and three: Prove value in live workflows, train employees against real exceptions, refine controls and scale where the organization can absorb the change.
- Years three to five: Extend AI across connected processes, redesign organizational structures where necessary, and embed continuous learning into normal operations.
This cannot be a fixed plan. AI capabilities, risks, and workforce requirements will change too quickly. The strategy should establish direction and investment priorities; the roadmap should be reviewed annually and adjusted based on operating evidence.
AI innovation will continue to outpace most companies’ ability to absorb it. Access to technology will not be the primary constraint. The constraint will be whether the organization can redesign work, prepare people to exercise judgment and maintain control as systems assume greater responsibility.
Companies that treat organizational change management as a late-stage adoption exercise may deploy AI faster at first, but they will struggle to scale it safely. Those that address roles, supervision and learning from the beginning will be better positioned to scale AI without surrendering judgment accountability or operational control.
Frequently Asked Questions
About the Author
Jonathan Gross, LL.B., MBA, is Pemeco’s Managing Director and head of its technology contracts practice. As a former litigator turned consultant and commercial lawyer, Jon’s clients benefit from his unique practice that includes technology law, technology strategy, enterprise software selection, and implementation management. By bridging the gap between legal and business, Jon’s clients benefit from his holistic approach to negotiating deals that drive commercial interests, manage risk, rebalance contractual equities, and promote successful implementations and long-term business partnerships. From high-growth start-ups to multi-national enterprises, Jon works with a cross-sector client base in private equity, manufacturing, distribution, property management, technology, professional services, and construction and engineering industries.
About Pemeco Consulting
Pemeco Consulting helps organizations succeed where most ERP projects fail. With a 100% success rate across 800+ projects, Pemeco guides clients through ERP strategy, selection, implementation, and transformation. Its globally recognized Milestone Deliverables methodology brings structure and clarity to complex programs. Independent and vendor-neutral, Pemeco serves private equity firms, manufacturers, and public sector clients. From strategy to execution, Pemeco delivers the insight, tools, and leadership needed to achieve ERP success—on time and in scope.
