AI agents can execute sourcing work faster, but automation is also removing the repetitive assignments through which buyers traditionally learned suppliers, categories and commercial risk. Procurement therefore faces a less visible workforce challenge: replacing the experience that used to accumulate naturally through years of execution.
The pressure comes as companies already face a constrained talent pipeline. Accenture projects demand across core U.S. supply chain occupations will increase by 1.34 million roles between 2026 and 2035, while the available labor force adds roughly 221,000 workers. That leaves an implied gap of nearly 1.1 million roles and makes developing existing talent increasingly important.
At the same time, AI is taking responsibility for more of the work that historically developed procurement capability. Supplier identification, sourcing-event creation, bid analysis, data gathering and other transaction-heavy activities can increasingly be automated or supported by AI.
The productivity case is clear. The development model is less settled.
Automation Removes More Than Administrative Work
Procurement careers have traditionally been built through repetition. Junior buyers ran RFQs, collected quotations, compared bids, questioned suppliers and worked through sourcing problems. Much of that work was administrative, but repeated exposure also provided context.
A buyer learned what normal pricing looked like within a category. They saw which suppliers consistently delivered, where specifications created unnecessary cost and how apparently attractive commercial terms could create problems elsewhere. Judgment emerged from hundreds of relatively small decisions.
AI changes that learning environment.
When software performs supplier discovery, prepares events or conducts first-pass bid analysis, people can reach higher-value work faster. Fairmarkit, for example, says its agentic tools can support sourcing workflows including intake, supplier engagement and evaluation, while its 2025 review reported a 50% year-over-year increase in autonomous sourcing adoption among its customers.
That creates an important distinction between removing low-value work and removing low-value learning. The two are not necessarily the same.
Someone can receive an AI-generated comparison of supplier bids without having manually examined every submission. That saves time. But reviewing dozens of imperfect bids was also one way buyers learned which inconsistencies mattered, which questions needed to be asked and where commercial risk could hide.
The problem becomes more significant as organizations flatten teams and expect employees to assume broader responsibility earlier. Procurement can automate the task, but it cannot assume that the judgment previously developed through that task will automatically appear.
Training therefore needs to become more deliberate.
Exceptions offer one potential replacement for repetition. Instead of shielding less-experienced employees from complicated sourcing decisions, organizations can deliberately expose them to cases where an AI recommendation requires intervention. Supplier disputes, unusual pricing, specification conflicts, poor-quality bids and risk escalations become opportunities to understand why the standard process failed.
The decision history behind AI recommendations can also become part of the training infrastructure. Employees should not simply see what an agent recommends. They need visibility into the information used, assumptions made and circumstances in which experienced colleagues accepted, modified or rejected its recommendation.
That turns human oversight from a control mechanism into a development mechanism.
Procurement Needs a New Model for Building Judgment
The workforce issue is broader than whether automation reduces procurement headcount.
Accenture’s 2026 supply chain workforce research argues that technology deployment needs to be accompanied by role redesign, continuous skills development and greater talent mobility. Its modeling suggests the projected labor shortage cannot simply be solved by continuing to hire against growing operational complexity.
Procurement presents a particularly interesting version of that challenge because AI can increase the amount of spend or sourcing activity that one person oversees while simultaneously raising the sophistication expected of that person.
A junior employee may therefore encounter supplier strategy, category planning, stakeholder management and AI-supported decision-making considerably earlier than under a traditional career path.
The capability requirements change with that responsibility.
Data interpretation becomes important because employees need to determine whether an AI-generated answer deserves confidence. Commercial judgment matters because the lowest price may not represent the strongest outcome once service, risk, specifications and supplier capability are considered. Technology fluency matters because buyers increasingly need to understand how automated workflows are configured and when human intervention is necessary.
The result should not be a choice between human expertise and automation. It should be a development system designed around their interaction.
Rotations can help. Employees can spend structured periods reviewing sourcing exceptions, supplier performance problems, negotiations and category decisions rather than simply processing transactions. Scenario exercises can expose people to unusual commercial situations without waiting for those situations to occur naturally. Reviews of completed sourcing events can examine not only the final award but what information changed the decision.
AI itself can support this model. Historical sourcing events contain examples of negotiations, supplier responses, trade-offs and outcomes that previously remained scattered across systems and individual experience. Properly governed, that information can be used to help employees examine past decisions and understand why particular outcomes occurred.
Organizations also need to reconsider career progression.
If automation enables employees to perform sophisticated work earlier, traditional promotion models based largely on years spent at each level become harder to defend. Development may instead depend on increasing decision complexity, category breadth, supplier responsibility and ownership of commercial outcomes.
That also changes retention. Employees who reach substantive work earlier may expect broader responsibility earlier. Organizations that accelerate capability without creating corresponding career opportunities risk becoming effective training grounds for competitors.
The competitive implication extends beyond recruitment.
AI technology will continue to spread across procurement platforms, reducing the advantage created simply by possessing a particular automation capability. Nearly six in ten procurement organizations were already using or piloting AI according to research discussed by Fairmarkit and Ardent Partners in 2026, with sourcing, intake, supplier discovery and spend analysis among the areas seeing adoption.
As access broadens, differences in performance may depend increasingly on how organizations configure the technology, govern decisions and develop the people responsible for supervising it.
The Scarce Resource May Become Experience
Procurement should be careful not to measure automation only by the hours it removes. Every eliminated task should prompt another question: what did employees used to learn while doing it?
That does not justify preserving manual work simply as training. It means organizations need an intentional substitute for the experience automation removes. The strongest procurement functions may ultimately be those that use AI to compress administrative work while deliberately increasing exposure to exceptions, trade-offs and consequential decisions.
As technology makes execution easier to scale, accumulated judgment could become harder to replace. Building that judgment faster may prove more important than automating the next task.