As AI agents take root and automate what I call the “manual middle” of procurement, we’ll see a significant productivity dividend. As that operational capacity opens up, CFOs and other executives will understandably want to see the impact on budgets and calculate a clear ROI on AI. After all, if AI is doing the work, the question will inevitably become, “Why do we need the same headcount?”
I expect that pressure to be real, and felt across the enterprise. But I also see a real danger in moving too quickly. Many companies will be tempted to cut staff prematurely, assuming AI can immediately fill the gap. Those organizations may discover what I call the “AI Value Gap,” a point where autonomous agents, lacking institutional judgment and historical context, struggle to manage complex exceptions and unexpected situations.
We have experienced a “generational black swan” event nearly every year this decade. We should expect the next crisis to be just around the corner. When it arrives, procurement teams that have become overly reliant on AI agents may lack the sourcing expertise, institutional knowledge, and judgment needed to respond effectively.
The CPOs I believe will win are not the ones who simply trade human capital for short-term budget savings. They will be the ones who use AI to elevate their teams, expand procurement’s scope, and deepen the function’s strategic impact. The real opportunity is not to replace people with AI. It is to use AI to make procurement professionals more capable, more strategic, and more valuable to the enterprise.
A First Look at Procurement AI in Practice
Ardent Partners provides one of the first detailed views into how CPOs are actually deploying and using AI within procurement. The findings in our recent report are significant, but they are also early. The organizations responding have moved beyond theory, but most have not yet reached scale. In many cases, the percentages referenced apply only to procurement groups with some level of AI in place, which means broader market adoption remains smaller at both the capability and sub-process levels.
What I see emerging is not a finished picture, but a directional one. Procurement organizations are actively engaging with AI, testing it across a range of applications, and beginning to understand where it can deliver meaningful value. Clear patterns are starting to form, but I would not consider them fixed by any means.
What AI Is Actually Doing: Embedded Capabilities
Among procurement organizations with AI in place, the most commonly embedded capabilities are natural language interfaces and autonomous or semi-autonomous workflows, each used by just over 40%. I view these as practical entry points, focused on intake, search, and workflow execution. They can reduce friction and improve speed without requiring major process redesign.
The next tier includes content generation and third-party AI tools, each used by about one in three organizations. Tools such as Gemini and Microsoft Copilot are supporting drafting, summarization, and the augmentation of existing processes. More advanced capabilities, including predictive analytics, optimization engines, and risk detection, appear in roughly one in four organizations. These capabilities require stronger data environments and more mature governance, so their lower adoption reflects not a lack of interest, but the additional complexity required to deploy them effectively.
How AI Is Being Accessed: The Model Question
Across procurement organizations are a broad mix of AI access models, with many organizations using multiple platforms at the same time. For example, 63% rely on general-purpose tools or copilots that are often enabled at the enterprise level. Beyond that, the market becomes much more varied. About one-third of organizations are developing internal or customized models, while others are accessing AI through enterprise platforms such as ERP and analytics tools or through their core procurement systems. A smaller group is using standalone, best-of-breed solutions.
To me, this mix reflects just how quickly the AI ecosystem is evolving. The market is still in a trial-and-error phase, with multiple approaches developing in parallel, and there is no dominant model today. I expect there will ultimately be clear winners and losers, but for now, procurement organizations are testing different options, evaluating performance, and determining which approaches work best for their needs.
Where AI Is Delivering: The Visible Impact
When I look at where AI is delivering the most visible impact, operational efficiency stands out. Among procurement organizations using AI, 42% of CPOs identify operational efficiency as AI’s most visible impact. That aligns directly with the CPO focus on speed and cycle-time reduction. AI is helping teams process work faster, reduce manual effort, and improve consistency.
At the same time, 24% say AI’s impact is not yet clear. This most likely reflects the early stage of many deployments. Other areas, including decision quality, risk mitigation, and user experience, are cited less frequently today. I expect broader strategic benefits to emerge as AI usage expands and matures, but organizations will need clearly defined success metrics to track progress and demonstrate value.
