Our article series, “The Path to AI-First Procurement: Closing the Gap Between AI Ambition and Execution,” based on new research from Ardent Partners and sponsored by Ivalua, continues today with a look at how procurement’s AI journey is being constrained less by a lack of interest in AI than by the tactical workloads, fragmented processes, legacy systems, and data challenges consuming organizational capacity. The path to AI-first procurement therefore begins with reducing that operational burden, thereby creating capacity for higher-value work and enabling greater speed, intelligence, and strategic impact.
In Part Eight, we examine how procurement AI governance remains immature, with 34% of organizations lacking formal policies, while 63% identify human-in-the-loop approvals as the most important governance requirement and 20% prioritize traceability. With 96% expressing moderate or high concern about black-box AI and 44% viewing limited transparency as a deal-breaker, explainability, operational visibility, and human oversight are essential to building trust and scaling AI across procurement.
Governance,Trust, and AI Control
As procurement organizations move toward more embedded and orchestrated AI operating models, governance becomes increasingly operational rather than administrative. Intelligent systems capable of influencing sourcing, supplier decisions, and workflow execution require a different kind of governance: one built for operational accountability, not just policy compliance. Current governance maturity across the market remains uneven and, in many organizations, still early.
More than a third of all procurement organizations report no formal governance in place for AI today (34%). Among those that do, governance tends to be broad and centralized rather than operationally specific. Enterprise- level oversight leads while the more precise execution controls, role-based access, escalation procedures, and model transparency requirements, remain less prevalent. Many organizations have governance in name. Fewer have governance in practice.
Procurement leaders appear highly aware of the governance challenge ahead. When asked to identify the single non-negotiable requirement for AI governance in procurement, the answer is not ambiguous.
Six in ten procurement leaders identify human-in-the- loop approval controls as the single non-negotiable requirement for AI governance (63%), making it the dominant answer by a substantial margin. Traceability follows at 20%. Together those two requirements account for more than 80% of all responses.
Procurement leaders are willing to embrace intelligent and agentic execution, but they are not willing to relinquish the ability to see what AI did, understand why it did it, and intervene when circumstances require it.
Human oversight is not a transitional preference that will fade as AI matures. It is a structural requirement for how procurement leaders define trustworthy AI operations.
This perspective becomes even clearer when examining procurement’s broader concerns surrounding opaque or black box AI capabilities.
Not a single respondent considers black box AI risk irrelevant to procurement operations. High and moderate risk concern together account for 96% of all responses. Nearly half of procurement leaders describe limited AI transparency as an outright deal-breaker (44%). Zero percent are unconcerned. For legacy software vendors marketing AI capabilities without explainability, operational visibility, or configuration transparency, these findings represent a significant and growing commercial risk. Procurement leaders are not debating whether transparency matters. They have already decided that it does.
Organizations that treat governance as an afterthought will likely encounter a scaling ceiling. Not because the AI fails technically, but because the organization itself will not trust those systems enough to operationalize them broadly. Governance is no longer simply a compliance requirement for procurement AI. It is a prerequisite.
This article is part of CPO Rising’s ongoing analysis of findings from Ardent Partners’ The Path to AI-First Procurement: Closing the Gap Between AI Ambition and Execution, based on the perspectives of 311 CPOs and senior procurement leaders across 25 industries worldwide.
Download this exclusive report from Ardent Partners.
RELATED ARTICLES
The Path to AI-First Procurement, Pt.1: What Does AI First Mean?
The Path to AI-First Procurement, Pt.2: The Execution Gap
The Path to AI-First Procurement, Pt.3: The Operational Reality Gap
The Path to AI-First Procurement, Pt.4: The Foundation Problem
The Path to AI-First Procurement, Pt.5: The Architecture Problem
The Path to AI-First Procurement, Pt.6: The Data Problem Underneath Procurement’s AI Ambition
The Path to AI-First Procurement, Pt.7: The Consumption Gap
