Artificial intelligence is widely viewed as essential to modern supply chains, yet most organizations are struggling to turn that conviction into operational results. A new study from Coupa suggests that many procurement and supply chain organizations are failing to scale AI initiatives or clearly demonstrate return on investment. The company’s Clarity AI Impact Report, based on a survey of more than 600 business leaders, finds that while enthusiasm for AI is high, most organizations lack the internal alignment needed to move beyond pilots.
According to the report, the central issue is not whether AI tools work, but whether organizations are equipped to deploy them at scale. A disconnect has emerged between executives approving AI investments and the operational teams responsible for implementation, leaving many initiatives stalled or unfinished. Although 86% of respondents say AI is critical to their business, only 29% report having a clear, company-wide strategy guiding its use.
AI Fluency Gaps Are Undermining Execution
Coupa’s findings point to a pronounced fluency divide across organizations. Just 5% of executive decision-makers say they use AI on a daily basis, compared with 57% of technical teams. That imbalance matters, the report suggests, because it shapes how goals, timelines, and investment cases are set.
When leaders lack hands-on familiarity with AI tools, expectations can become detached from operational realities. This is compounded by broader workforce readiness challenges. Only 21% of surveyed organizations believe they currently have the skills required to use AI effectively, while 69% say limited training and capability are slowing adoption.
In supply chain environments, where forecasting, inventory management, and logistics depend on tight coordination across functions, those gaps can prevent AI from delivering consistent value. Without shared understanding across leadership and operations, projects struggle to progress from experimentation to embedded decision support.
Data Complexity and ROI Pressure Stall Progress
The report also highlights what Coupa describes as “pilot purgatory.” Roughly 72% of AI initiatives fail to move beyond early testing phases, yet nearly half of executives still expect measurable business returns within six to twelve months. That tension places significant pressure on teams working within complex, often fragmented system landscapes.
Data quality and integration emerge as the most common obstacles. Seventy-seven percent of respondents cite fragmented or unreliable data as a major barrier to AI success. In supply chains, this typically means stitching together information from enterprise resource planning systems, warehouse platforms, transportation tools, and supplier portals, an effort that is both time-consuming and foundational.
Without reliable, integrated data, AI systems struggle to provide the end-to-end visibility and predictive accuracy that justify their investment. As Dennis Bruder, Chief Product Officer of AI at Coupa, notes in the report, expectations around AI have shifted. He argues that organizations can no longer justify funding AI initiatives based on potential alone, and that decision-makers are increasingly demanding measurable outcomes tied to business performance. Meeting those expectations, he says, requires moving beyond theoretical commitment toward platforms that support governance, adoption, and execution at scale.
From Isolated Tools to Governed Platforms
To address these challenges, many organizations are turning toward unified AI platforms rather than building bespoke solutions. Coupa’s research shows that 80% of companies now prefer to buy AI capabilities through external platforms. However, the way those platforms are used remains uneven.
Only 2% of AI investment is currently directed toward orchestration, even though 77% of leaders prioritize simple task automation. The imbalance suggests that many deployments remain narrowly focused, delivering localized efficiencies without reshaping end-to-end processes.
Governance is another unresolved issue. While 65% of executives favor “human-in-the-loop” oversight, more than half of respondents are unsure whether their organization has a formal AI governance policy in place. Without clear frameworks, human oversight risks becoming a brake on automation rather than a safeguard that enables trust and scale.
When AI Becomes Part of the Operating Rhythm
Evidence from recent enterprise deployments suggests that AI programs stabilize when their outputs are reviewed alongside the same metrics that already govern supply chain decisions. Teams that incorporate AI signals into standing processes, such as forecast sign-offs, inventory positioning discussions, or supplier scorecard reviews, tend to resolve trust and adoption issues faster than those treating AI as a separate layer of insight. The practical shift is subtle but consequential: AI stops competing for attention and instead becomes another input subject to scrutiny, adjustment, and accountability. Over time, that integration can matter more to scale than model accuracy alone, because it determines whether AI informs decisions consistently rather than intermittently.
