How AI is re-shaping the planning domain, value-add activity and role of the planner
Every planning organization is asking some version of the same question: as AI takes on more forecasting, order fulfilment, and exception management, what is left for the planner to do? For some organizations, this is already a practical question. For most, it is still a design question, since the data and process foundations needed to support AI at scale are still being built. Either way, the answer matters: it shapes the operating model, the roles within it, and future talent needs, not just the technology.
Much of the discussion on this topic settles on a simple principle: let AI handle the routine, humans handle the exceptions. That is partially true, but it misses an important part of the equation. Which decisions should remain human-owned by design? What happens to the skills planners stop practicing when routine work disappears? And how do you build the next generation of planners when the traditional learning ground is increasingly automated?
Where AI genuinely earns its place
The case for AI and agents in supply chain planning does not need much exaggeration. Technology is developing quickly, and investment is following. Gartner projects spending on supply chain management software with agentic AI capabilities to grow from under $2 billion in 2025 to $53 billion by 2030. At the same time, most planning organizations are still at an early stage. Some are piloting one or two use cases, while others are still working on the data and process foundations required to make AI useful at scale.
That is not a reason to wait. It is a reason to be precise about where AI creates value today, in tasks that are high-volume, repetitive, and well defined, such as:
- Demand sensing and first-pass forecasting: continuously processing POS, weather, market, and other demand signals instead of waiting for a monthly planning cycle.
- Routine exception triage: identifying and potentially resolving high-volume, low-complexity exceptions that currently consume a significant part of a planner’s day.
- Scenario generation: running multiple what-if scenarios quickly, allowing planners to compare alternatives instead of spending hours building them manually.
- First-draft replenishment and allocation: generating a proposed supply or allocation plan that a planner can review and adjust rather than building one from scratch.
These are valuable applications because they change where planners spend their time. The machine takes care of the obvious, while the planner can focus on situations where judgment adds value. But there is an important distinction here: freed-up time does not automatically mean better decision-making. If AI takes over the routine work, organizations need to actively decide what planners should do with the time that is created. Otherwise, there is a risk that the planner simply becomes a reviewer of machine output. And how value-add is that then still?
Not every decision needs the same treatment
Not every planning decision carries the same weight, and treating them as if they do is where most operating models get it wrong. Approving a replenishment quantity within known bounds and deciding how to respond to a sole-source supplier failing a quality audit are not the same kind of decision, yet many organizations put a human in front of both, or automate both, without asking why. The question worth asking of every planning activity is simple: is this a well-bounded decision the model can defend with confidence, or does it carry trade-offs, stakes, or consequences that call for judgment the model was never built to have?
Decisions that belong on the human side tend to share a few traits: they involve trade-offs across commercial, operational, and financial priorities that a model cannot easily weigh on its own; they respond to disruptions with no real precedent; they carry meaningful customer, reputational, or ethical consequences; or they rest on data too incomplete or complex, or a recommendation too unexplainable, to act on without someone checking it first. Everything else, the well-bounded, historically consistent decisions, is exactly where automation should run without a person in the way. Even better so. None of this is an argument against automation. It is an argument for being as deliberate about identifying what stays human as about what gets automated.
The quiet risk: judgment you stop practicing is judgment you lose
When a planner spends years making forecasts, investigating exceptions, and deciding how to respond to unexpected events, they build pattern recognition: what normal looks like, what signals matter, when something does not add up.
That learning happens through repetition, and repetition is exactly what disappears as routine work is automated.
This creates a paradox: the better AI gets at handling routine situations, the less often humans encounter the situations that build the judgment needed for the hard ones.
Supply chains are full of situations that fall outside historical patterns, exactly where AI is weakest and human judgment is required. Keep automating the reps that build decision-making capabilities, and you may risk the quality of those decisions precisely when it matters most.
The risk hits hardest at the entry level, where junior planners have always built judgment by doing the routine work that is now easiest to automate. A recent Gartner survey found that 55 percent of supply chain leaders expect agentic AI to reduce entry-level hiring needs. That is a problem long before it is a saving: if nobody starts at the bottom doing the work that builds pattern recognition, who grows into the senior, decision-heavy roles a few years from now?
Building this into the operating model, not bolting it on later
This raises the real design question behind any AI-enabled planning transformation: not whether AI will reduce the number of planning activities, but how the organization should be designed around the activities that remain. A few questions are worth answering before AI becomes deeply embedded:
- Who will develop the judgment that used to be built through routine planning work?
- How will the organization keep developing planning talent as more of the traditional learning curve is automated?
- Which existing roles should absorb new responsibilities, such as model oversight or scenario framing, and where does a genuinely new role make more sense? In practice this often splits into three distinct role archetypes on AI: who owns the data, who interprets its output, and who designs the logic behind it, what we think of as citizens, translators, and wizards of data and technology.
- How to govern a hybrid setup with human planner’s vs agents. Re-establishing accountabilities beyond the traditional planning models of the past.
Most organizations do not need to redesign their structure overnight, but these questions belong on the transformation roadmap from the start. The organizations that benefit most from AI will not simply automate existing processes. They will redesign how humans and machines work together, and build the operating model, capabilities, career paths, and change management approach alongside the technology, rather than after it.
The planning job isn’t shrinking. It’s re-shaping.
AI will change the planning domain, but the biggest change may not be the tasks that disappear. Instead, it should be about the skills and decision-making capabilities organizations need to build around the work that remains. Organizations that address this deliberately can use AI to create more capacity without losing the expertise their supply chains depend on. AI transformation should not only answer “What can we automate?” It should also answer, “What do we want our people to become better at?”
We have watched organizations answer that differently across dozens of planning transformations, and the ones that hold up a few years out are rarely the ones with the best technology, they’re the ones that got the operating model and the people plan right from the start. That doesn’t mean redesigning the structure overnight. But it does mean these questions belong on the transformation roadmap from day one: the organizations that benefit most from AI will not simply automate existing processes. They will redesign how humans and machines work together, and build the operating model, capabilities, career paths, and change management approach alongside the technology, rather than on top of it.
This blog is the third in a series around AI in supply chain planning. Interested to learn more about our broader vision on AI in planning then read our Digital Transformation With AI at the Heart whitepaper.