We're past the point of debating whether AI will change procurement. It already has. The more useful question now is: what does that mean for the people doing the work?
At a recent roundtable, four procurement leaders — Pallavi, Canda, David, and Larysa — got specific. No hype, no forecasts. Just honest takes from practitioners on what's shifting, what's at risk, and where humans still matter most.
The skill that matters most right now
When asked to name one skill every procurement professional must build today, the answers were telling.
Pallavi went straight to data skepticism — the ability to look at AI-generated output and spot hallucinations or embedded bias before acting on it. Canda pointed to adaptability, and specifically the willingness to acknowledge your own biases as AI changes the landscape around you. David named emotional and behavioral intelligence — the kind of thing, as he put it, that "doesn't come out of university courses or professional bodies." And Larysa called out negotiation support and contract management as the capability AI will make table stakes fastest.
No one said "learn to use the tools." They all pointed inward.
The half-life of skills is accelerating
David brought data to the conversation. The World Economic Forum projects 92 million jobs displaced by AI by 2030 — but 170 million new ones created. The net is positive. What's harder to absorb is the pace of change underneath that.
In the early 2000s, the half-life of a professional skill set was roughly 10 years. By 2020, it had dropped to five. Within the next two to three years, it's projected to hit three. Skills that feel current today may be partially obsolete before the end of this decade.
"Dare I say," David noted, "with that then goes releasing old skills."
Where AI is actually being used
Larysa shared a breakdown she uses with clients — automating the repetitive, augmenting the strategic. On the automation side: purchase order creation, data cleansing, RFP spec matching. Pallavi described how LLMs can now take disparate vendor responses and normalize them into a consistent format for buyers — something that used to require significant manual effort and rigid supplier instructions.
On the augmentation side: category strategy, supplier performance management, contract opportunity identification. AI surfaces what's possible. Humans decide what to do with it.
David gave a vivid example from his own experience — a manual contract audit that took weeks and revealed millions in leakage. Suppliers had been adding packaging and shipping costs to invoices that were auto-approved within tolerance thresholds. "It ran into millions that we'd actually leaked out of the contracts." That kind of work — now automatable — used to either not get done or consume enormous bandwidth.
Reverse mentoring: the upskilling model that's working
One of the more interesting structural shifts discussed was the rise of reverse mentoring programs. Junior team members coach senior leaders on generative AI tools — market research, prompt workflows, data interpretation. Senior leaders teach juniors negotiation psychology and relationship management.
Both sides are vulnerable. Both sides are teaching. Companies that have implemented this model are seeing results, and it addresses something Canda raised early: fear. People with deep expertise in one area being asked to become beginners in another. Mutual vulnerability, as a design principle, helps.
Where humans stay irreplaceable
The roundtable closed on this question. The consensus was clear: negotiation.
Not just on price. Pallavi put it directly — "AI can calculate the optimal price, but it cannot look the supplier in the eye and gauge if they're hesitating because they're bluffing or because they're about to go bankrupt." That gut read, built over years of relationship context, is not something that shows up in a prompt.
David extended it to what he calls social contract skills — the relational capital that holds when conditions change unexpectedly. "When things like COVID kick in, it's relationships that will fix that stuff." Fixed-term contracts don't bend. People do.
The through-line across all of it: AI expands what's possible in procurement. But the leaders who will capture that value aren't the ones who hand off judgment to the system. They're the ones who get sharper about where their judgment actually matters — and invest there deliberately.
This roundtable was hosted by procurement.news.