Dividing work among employees is a task that every manager or team leader performs daily, often without pausing to consider how many trade-offs it involves. Who has room this week, who has the expertise, who is due for a learning moment. The answer to whether AI can take this over is: partly, and the part that remains is precisely the part that the manager's role revolves around.
The familiar three-way split also runs through the division of work. Part of the process AI can already handle independently today: keeping track of who is working on what, flagging overload based on planning data, putting forward a proposed division based on available capacity in the project management system. That is agent work: a system that continuously monitors and comes up with a proposal before a human asks for one.
Another part works with oversight: AI proposes a division, a manager approves or rejects it, with reason. For example, when the system pairs two employees with similar skills to a task, but the manager knows that one has just come through a difficult stretch and needs rest, while the other is ready for a challenge. That trade-off appears nowhere in a planning tool.
And a part remains human work, full stop. The development needs of an individual employee, the political sensitivity of who gets a prestigious project, the conversation in which someone signals they would rather not do a task without saying so out loud: that calls for someone who knows the team, not an algorithm optimizing capacity.
The core of the issue lies in three axes. Judgment latitude is only moderately favorable for automation: dividing work is never purely a matter of calculating hours and skills, it is also a social and development-oriented decision. Structuredness is low: there is no fixed step-by-step plan that works the same way in every company, because teams, tasks, and personal situations always differ. Volume is favorable, however: it is a task that recurs often, which makes it worthwhile to support part of it with systems, even if not everything is automatable.
The remaining axes confirm this picture. Customer contact and physical presence play virtually no role, so they pose no obstacle. Creativity is only marginally needed, and the cost of errors is moderate: a wrong division does not immediately cause damage, but it can create friction or a missed development opportunity. Compliance comes into play as soon as the division of work touches on equal treatment or terms of employment; that is governed by its own legal requirements, separate from what a system can technically do.
A team leader at a service provider divides the work among eight employees every week. A planning tool can show exactly who has how many hours available and which certifications or skills fit which task. That saves the team leader from manually sifting through calendars and skills matrices. But the team leader also knows that one of the eight has just come through a difficult period and would prefer calmer work this month, and that another employee is ready for a task just outside their comfort zone in order to grow. Those two considerations appear in no system and change the division the system would propose.
In a company with a large, uniform team where tasks are highly standardized — for example, a call center with fixed scripts and clear KPIs per employee — the balance is different. There, dividing work is less a development question and more a capacity question, and a system can handle a larger share of the division independently, with a human stepping in only for exceptions. In a small team with a lot of interdependence and varying development paths, the share that remains human work is correspondingly higher.
The precondition is always the same: a system needs up-to-date insight into the team's skills and availability. Where that insight is missing or scattered across separate spreadsheets, automation remains limited to nothing, however favorable the rest of the profile may be. Where that insight is in order, AI can provide the overview and the initial proposal, leaving the manager with the trade-off that genuinely concerns people.
This shift also touches adjacent work. Whoever divides work often also needs to organize and schedule review cycles, and the outcomes of division and review sometimes feed into management reporting for the board or shareholders. Preparing management meetings and their agendas is also tied to who does which work and why.
This page describes a taxonomy, not a measurement of your own organization. Whether the share of automatable work in your case is higher or lower than sketched here depends on how structured your teams' work is and how well the insight into skills and availability is organized. The free quickscan from FTE TO AI consists of twelve questions, no account required, and gives an indication of what share of the hours in this profile can be taken over by AI today. The full work scan, which calculates the work of an entire company task by task into FTE capacity, is still under construction. Anyone who has already had a measurement done can read what a re-measurement shows about how that picture changes over time.
Vraag maar. Ik ken de kennisbank van deze site; wat ik niet weet, zeg ik erbij.
Answers come from this site’s knowledge base. Not tailored advice, and not a scan of your company.