Imagine an executive team reviewing a proposed AI investment. The business case looks compelling. The technology promises to automate repetitive work, accelerate analysis, improve response times, and increase productivity. Implementation costs have been estimated, vendors evaluated, risks discussed, and someone has calculated an expected return on investment.
Everything appears ready for approval.
But there is a question that may be buried somewhere underneath all those calculations: What exactly happens to the workforce if the productivity assumptions turn out to be right?
Suppose the technology really does reduce the time required for a particular category of work by 20 percent. What does leadership expect to happen next? Perhaps the organization can handle considerably more business with the same workforce. Maybe employees can redirect that capacity toward work that creates greater value. Jobs might need to be redesigned, future hiring needs could change, or the efficiency may ultimately allow the organization to operate with fewer positions.
Any of those outcomes could be legitimate. What concerns me more is the possibility that nobody has really decided which outcome the business case assumes.
Because somewhere inside almost every AI productivity projection is an assumption about human work, whether leadership has articulated it or not.
Productivity Is Not a Workforce Strategy
A business case might project that AI will improve productivity by 15 percent. That sounds wonderfully precise until someone asks what the 15 percent actually means.
If technology allows employees to produce the same output using fewer hours, the organization has created capacity. What happens to that capacity determines whether the productivity gain eventually becomes business value.
In a growing organization, the answer might be straightforward. Employees can handle additional customer demand without headcount increasing at the same rate. In another organization, recovered capacity could improve service, accelerate an important initiative, or allow employees to spend more time on complicated work that has been neglected because they were consumed with routine tasks.
In still another organization, the productivity improvement may eventually reduce the number of positions required.
These outcomes have very different workforce implications, even though all of them might begin with the same 15 percent productivity assumption.
That's why I think executives need to distinguish between a technology productivity projection and a workforce strategy. One tells us what the technology may enable. The other tells us what the organization intends to do with it.
The technology business case may be finished while the workforce business case has barely begun.
Saved Time Doesn't Automatically Become ROI
Time savings are particularly attractive because they're easy to understand.
If an employee spends ten hours each week performing work that AI reduces to five, we've recovered five hours. Multiply that across dozens or hundreds of employees and the potential productivity gain can become substantial.
But those five hours don't automatically turn into financial value.
The organization is still paying the employee. The value depends on what happens during the time that was recovered.
Perhaps the employee can serve more customers, reduce a backlog, perform work that otherwise would have required another hire, or devote time to an important strategic initiative. In each of those cases, leadership can begin connecting the productivity gain to an organizational result.
But suppose the recovered time simply disappears back into the workday. The employee attends more meetings, answers more email, picks up additional administrative tasks, or absorbs miscellaneous work that was already waiting.
The five hours were technically saved. The organization just may not have captured much additional value from them.
This is where workforce planning needs to enter the AI conversation much earlier. If productivity is part of the financial justification for the investment, leadership should have some idea how that productivity is expected to become a business result.
The Spreadsheet May Be Assuming Something Nobody Wants to Say
Sometimes the workforce assumption inside an AI business case is straightforward but uncomfortable: the projected financial return depends partly on reducing labor costs.
There is nothing inherently wrong with acknowledging that. Organizations have always used technology to perform work more efficiently, and there will be situations where automation genuinely reduces the amount of human labor required.
The problem comes when the financial model and the organizational message tell different stories.
Leadership may describe AI as technology that will free employees to concentrate on higher-value work, while the financial model quietly assumes headcount savings. Those aren't necessarily compatible strategies.
If reducing positions is part of the anticipated return, leaders need to understand that from the beginning. Workforce reductions affect more than salary expense. They can influence trust, retention, recruiting, workload, capability, and the willingness of employees to embrace the next technology initiative.
Likewise, if the organization has no intention of reducing headcount, the business case shouldn't depend on labor savings that are unlikely to materialize.
Clarity about the workforce assumption doesn't make the decision easier. It makes the strategy more honest.
“Higher-Value Work” Needs to Become Real Work
One of the phrases I hear frequently in conversations about AI is that the technology will free employees to focus on “higher-value work.”
I agree with the potential. But I'm increasingly interested in what happens after that sentence.
What exactly is the higher-value work?
Imagine AI removes several hours of administrative work from a manager's week. Perhaps the organization wants that manager spending more time coaching employees, solving customer problems, improving processes, developing talent, or working on strategic priorities that have consistently been pushed aside by day-to-day demands.
Any of those choices could make sense.
But unless the work is intentionally redirected, higher-value work can remain an attractive concept rather than an actual change in the job. The old tasks shrink, the old job remains, and the recovered time gradually fills with whatever else happens to be waiting.
Six months later, leadership may wonder why the technology improved efficiency without producing the transformation everyone expected.
The answer could be surprisingly simple.
Technology changed the task.
Nobody redesigned the work around it.
AI May Change Hiring Before It Changes Headcount
There is another workforce effect that may be easier to miss because it doesn't arrive as a dramatic restructuring announcement.
Imagine a department expects business volume to grow substantially next year. Historically, leadership might assume that workforce demand would grow at roughly the same rate. More customers, more transactions, or more work would eventually require more people.
AI can change that relationship.
If the existing workforce becomes capable of absorbing much of the additional volume, the organization may not need fewer employees. It may simply need fewer additional employees than it would otherwise have hired.
That's an important distinction.
Some of the earliest workforce impact from AI may appear through positions that aren't added, vacancies that aren't automatically replaced, contractor spending that changes, or business growth that occurs without proportional headcount growth.
This makes workforce forecasting more complicated because historical staffing ratios become less useful. The fact that it took 100 employees to produce a particular level of output three years ago doesn't necessarily tell us how many people the organization will need to produce that output three years from now.
Increasingly, workforce planning has to look at how the work itself is changing rather than simply projecting yesterday's staffing model forward.
Jobs Don't Lose Tasks in Convenient Percentages
There is another problem with productivity math that looks much cleaner on a spreadsheet than it does inside an organization.
Suppose AI can automate roughly 25 percent of the tasks performed across a team of ten employees. It's tempting to look at that and see the equivalent of two and a half positions.
Real work usually isn't distributed that neatly.
The tasks affected by AI may be scattered across all ten jobs. Some of the remaining responsibilities may require specialized knowledge or human judgment. Workloads may fluctuate during the year. Certain responsibilities may still require human oversight even if they occupy relatively little time.
You can't simply remove 25 percent of ten people.
This is where task-level productivity assumptions eventually collide with job-level workforce decisions.
Perhaps the remaining responsibilities can be recombined into redesigned roles. Maybe employees can take on work from another function, or the organization can use the capacity to support growth. In some cases, fewer positions may genuinely be required.
But none of those outcomes happens automatically because technology eliminated several tasks.
Someone still has to design the future work.
The Workforce Assumption Isn't Always About Numbers
It's also easy to assume the workforce conversation is primarily about headcount.
Often it isn't.
An AI strategy may depend on employees becoming effective users of new tools, managers learning how to supervise AI-enabled work, teams redesigning processes, and employees developing enough judgment to recognize when an AI-generated answer is wrong or inappropriate.
Those are capability assumptions.
Buying the technology doesn't automatically create them.
If the expected ROI depends on employees working differently, then learning and development are part of the AI investment whether they appear in the technology budget or not.
The same is true for managers. Leading AI-enabled work may require them to evaluate output differently, rethink workflows, coach employees through changing responsibilities, and make more nuanced decisions about where technology should and shouldn't be used.
If the business case assumes all of this capability will simply emerge after implementation, there is another workforce assumption hiding in the model.
Managers Will Have to Make the Strategy Real
Whatever strategy executives choose, managers will eventually translate it into everyday work.
They're the ones employees will approach with questions about how jobs are changing. They'll see which tasks technology handles well and where it creates new problems. They'll notice whether the promised time savings are actually occurring and whether employees are using that capacity productively.
This is why bringing managers into the conversation only after the major decisions have been made can be a missed opportunity.
The people closest to the work often know where the real inefficiencies are. They understand which activities customers value, which tasks consume time without producing much benefit, and where automation may create consequences that aren't obvious from an executive presentation.
Managers shouldn't determine enterprise AI strategy on their own, of course.
But they can provide valuable intelligence about whether the assumptions underneath that strategy match the reality of the work.
The Technology Budget and Workforce Budget Need to Meet
Consider an organization that invests heavily in technology specifically intended to reduce administrative workload. The investment is approved because leadership expects significant productivity gains.
Then every vacancy in the affected function continues to be automatically backfilled. Headcount plans remain essentially unchanged. Job descriptions stay the same, and nobody meaningfully redesigns the workflow.
Perhaps that's intentional. Maybe growth requires every bit of capacity the technology creates.
But if nobody has examined the relationship between the technology investment and the workforce budget, the organization may find itself funding both the new way of working and the old staffing model at the same time.
That can make transformation very expensive.
Technology planning, financial planning, and workforce planning can no longer operate as completely separate conversations when technology is specifically intended to change human work.
If the organization expects the technology to change work, the workforce plan should eventually reflect where leadership believes that change will occur.
Otherwise, the budgets may be telling two entirely different strategic stories.
One Productivity Number Probably Won't Fit the Whole Organization
There is also a temptation to apply broad productivity assumptions across an enterprise.
AI will improve productivity by 10 percent, 15 percent, or 20 percent.
But across what work?
The impact of AI on a software developer's work may look very different from its impact on a nurse, salesperson, accountant, executive assistant, plant manager, or customer service representative. Even within the same job, some tasks may change dramatically while others barely change at all.
A single enterprise productivity assumption makes financial modeling easier.
It can also create false confidence about workforce implications.
The more useful analysis happens closer to the work. Leaders need to understand where employees actually spend their time, which activities technology can realistically change, what capacity might be created, and what the organization wants to accomplish with that capacity.
That approach may produce a less dramatic spreadsheet.
It may produce a much better strategy.
Employees Are Making Their Own Assumptions Too
While executives discuss AI strategy and ROI, employees are having a somewhat different conversation.
They're wondering what the technology means for them.
Will it make the job easier? Will expectations increase? Will fewer people be needed? Will new skills become important? Will experience that took years to develop still matter? Will the job they have today look anything like the job they'll have several years from now?
Leadership may not have complete answers to those questions.
That's understandable. In many organizations, nobody knows exactly how quickly the technology will develop or how extensively work will change.
But silence doesn't eliminate the questions.
Employees will fill in the blanks themselves, and workplace rumor has always been remarkably efficient at filling information vacuums.
Executives don't need to pretend they can predict every workforce outcome. They can be transparent about what the organization is trying to accomplish, what leadership currently understands, what remains uncertain, and how future workforce decisions will be approached.
That kind of clarity becomes particularly important when employees are being asked to help implement technology they suspect could eventually change their own jobs.
Make the Workforce Assumption Visible
None of this means an organization should delay AI investment until it can perfectly predict every workforce implication.
That's unrealistic.
Technology will evolve. Employees will discover uses nobody anticipated. Some productivity gains will exceed expectations, while others won't materialize at all. Strategy always contains assumptions.
The important thing is knowing what those assumptions are.
When executives review an AI investment, they should be able to explain what needs to happen in the workforce for the projected return to become real. Perhaps the model depends on handling greater customer volume without equivalent hiring. Maybe it assumes lower contractor spending, different jobs, new employee capabilities, fewer future hires, redeployed capacity, or eventually lower labor costs.
Once the assumption is visible, leadership can ask whether it's realistic.
Without that conversation, an organization can approve a remarkably sophisticated technology strategy while maintaining a remarkably vague plan for the people doing the work.
The Most Important Decisions May Come After the Technology Works
Perhaps the most interesting part of all this is that the hardest workforce questions may arrive when the AI implementation succeeds.
Imagine the technology performs exactly as promised. Work that once required ten hours now takes six.
That's good news.
But it isn't the end of the strategy.
Leadership now has to decide what the organization wants to become capable of doing with those four recovered hours. Maybe the answer is growth. Perhaps it's better service, lower cost, redesigned jobs, new capabilities, or additional capacity for strategic priorities that have been waiting for attention.
The technology can't make that decision.
That's leadership work.
AI can change how much human effort a task requires. It can change how quickly work happens and what employees are capable of producing.
But someone still has to decide what happens to the people, jobs, capabilities, and capacity surrounding that work.
So the next time an AI business case arrives with an impressive productivity projection, look underneath the number.
Somewhere inside that spreadsheet is a workforce assumption.
Make sure you know what it is before the ROI depends on it.
