Agentic AI in Procurement: High Potential, Higher Organizational Hurdles

---

Procurement is about as close as you can get to a horizontal business process tailor-made for the use of agentic AI: it involves complex decisions based on vast data quantities, cross-functional and repeatable workflows, and business volatility that creates enduring challenges.

Based on those factors, procurement professionals anticipate big benefits from agentic AI in the coming year. Unfortunately, they’re not well positioned to deliver on those benefits currently because of fragmented operations and systems, as well as major data integration challenges.

That means transformation across business functions – particularly procurement and finance – is required to realize AI benefits, according to IBM’s Institute for Business Value, whose Enterprise Business Operations Study yields powerful insights on procurement and AI. “Procurement cannot scale agentic AI until its operating models shifts from functional to enterprise-wide,” according to IBM’s analysis of that data.

The limitations it found in procurement – highlighted below – have similarities to the vertical industry challenges facing financial services, which we analyzed in a report earlier this week.

High Hopes for Procurement

Let’s start with the enthusiasm that Chief Procurement Officers (or CPOs) report in terms of how AI can impact their business: 78% of CPOs responding to the IBM survey believe AI will fundamentally reshape business operations by 2027.

More specifically, they forecast that AI can drive a:

• 58% boost in innovation
• 54% increase in security and risk performance
• 49% improvement in both procurement cycle time and supplier onboarding time
• 38% improvement in purchase order accuracy

These findings speak to agentic AI’s ability to reinvent core processes that are most valuable in procurement. If such a reinvention plays out – and that of course isn’t a given — IBM notes that procurement will evolve “from a cost control function to a real-time performance orchestrator.”

To make that evolution happen, procurement and finance need “a common data fabric spanning supplier records, spend taxonomies, commitment metadata, risk signals, policy logic, and compliance parameters.” But before that, there are barriers to be overcome.

Not Ready for Prime Time

Even with a clear set of data and operational practices needed to support agentic AI, the CPOs indicated their organizations aren’t optimally positioned to pull the trigger.

They’re impeded by fragmented or overlapping business processes and systems as well as governance gaps. These are concrete examples of where things aren’t working in support of autonomous processes powered by AI:

• Just 16% of organizations operate global, integrated business models
• Only 28% have workflow owners with end-to-end accountability, including on a global scale
• 83% cite partner data and technology integration as a major challenge

That last data point truly jumps out: if the teams running one of the most critical, partner-centric processes are reporting big data and tech integration challenges, that’s a troubling indicator of the hill they must climb to unlock the agentic AI results they’re anticipating. “Procurement cannot deploy autonomous workflows in isolation,” the report cautions. “It needs a  network of connected business capabilities.”

Other challenges speak to AI talent and organizational culture. For instance, 69% of CPOs name talent limitations as a gating factor as they seek to transform and, in a related finding, 48% identify a shortage of agentic AI expertise in their companies. In addition, half cite cultural resistance as a barrier to deploying agentic AI.

The time is clearly now for procurement to start advancing toward agentic AI in order to capitalize in 2027. IBM’s experts map out a 12-month roadmap that calls for leaders to start by mapping cross-domain workflows, establishing shared data definitions with finance, then progressing from domain-specific AI agents to multi-agent orchestration across domains.

Planning and executing all those advances between now and the end of 2027 is a tall order to say the least. But given that procurement executives recognize the potential payoff, and need to avoid being caught flat-footed as AI’s rapid advance continues, do they really have an alternative to moving ahead aggressively?

Community Summit North America is the largest independent innovation, education, and training event for Microsoft business applications delivered by Expert Users, Microsoft Leaders, MVPs, and Partners. Register now to attend Community Summit in Nashville, TN from October 11-15.