As procurement leaders face a surge in responsibility from the rise of generative AI, the very technology causing the overload could offer a path to sustainable relief—if integrated wisely.
Gen AI’s Promise Comes With a Heavy Operational Price Tag
Procurement has stepped into the strategic spotlight in recent years, thanks to global supply shocks, digital transformation, and boardroom-level demand for greater resilience. But with increased visibility has come increased complexity—none more acute than in the procurement of generative AI (gen AI) and machine learning tools.
Procuring gen AI solutions isn’t simply another digital sourcing task. These systems introduce a raft of new demands, from scrutinizing model training data to negotiating IP rights over generated outputs. Technical hurdles—like integration with legacy systems and ensuring interoperability across departments—are only part of the equation. Ethical, legal, and cybersecurity concerns also pile up, stretching procurement’s role far beyond traditional supplier evaluation.
In practice, the process begins like any other technology procurement: identifying business needs, surveying vendors, managing tenders, and negotiating contracts. But here’s where the complexity balloons. Procurement must assess not only performance metrics and pricing, but also how models were trained, what data underpins them, whether that data is biased, and what guardrails are in place against prompt injection or model inversion attacks.
Transparency becomes another sticking point. In mission-critical use cases, procurement leaders increasingly have to prioritize explainability over performance—an ongoing challenge with black-box LLMs. And even if these hurdles are cleared, post-purchase training needs and murky intellectual property rights can cause projects to stall. As PwC’s Philipp Rosenauer notes, reluctance from suppliers to cede IP over AI outputs remains a contractual minefield.
From Shadow AI to Strategic Overload
One growing concern is the emergence of “shadow AI”—unsanctioned deployments of generative AI tools within departments. These rogue purchases not only breach internal policy but threaten to derail company-wide interoperability goals. Procurement must now go beyond enabling strategic AI adoption; it must monitor and contain it, aligning disparate implementations into a coherent digital architecture.
The weight of these responsibilities is pushing procurement teams to their limits. Evaluating AI models alone is a major time sink—especially when frameworks for such assessments are still immature. And with AI adoption only accelerating, there’s no sign the workload will ease anytime soon.
Ironically, the solution may lie in the very systems adding to the burden. AI-powered procurement platforms—those automating source-to-pay workflows—could offer the bandwidth relief leaders urgently need. By streamlining routine tasks like invoice handling, contract management, and supplier onboarding, these platforms can free up capacity to handle more strategic AI-related decisions.
Procurement Must Automate to Survive Its Own AI Demands
Procurement isn’t just buying AI tools—it’s redefining its own operating model to accommodate them. That puts leaders in a bind: they must master the intricacies of AI procurement while still managing day-to-day operations, compliance, and cost control. The only viable path forward is to lean more heavily on automation within the function itself.
In essence, AI is creating the problem—and offering the only scalable solution. Procurement directors would be wise to act now, before the function becomes a bottleneck to broader digital transformation goals.
