When two sites buy the same part from the same supplier at different prices, and nobody in either building knows, the waste is not a rounding error—it is a multi-million-dollar leak. That is the core insight from Priyanka Malla, an IT project manager specializing in AI-driven enterprise procurement and supply chain transformation.
“At scale, the multi-million-dollar opportunity in procurement is not smarter buying, it is unification,” she says. Malla manages a portfolio of capital and operating spend covering AI initiatives, enterprise procurement, ERP modernization, and infrastructure. Her largest program consolidated procurement across more than 50 geographically distributed sites onto a single standardized platform. Each site arrived with its own legacy process and supplier relationships, and there was no template to follow.
The obstacle, she found, was not the software. “When integrating systems that were never designed to talk to each other, the hard part is not the technology, it is reconciling how differently each site had been operating.” Reconciling fifty ways of buying into one was less an integration job than a negotiation with each site about how it worked.
She followed that with ERP modernization across more than 15 branches, bringing together supply chain and financial data that had lived in disconnected legacy systems. The two programs did the same thing from different directions: one standardized how the organization buys, the other standardized what it knows about what it has bought.
That order matters for artificial intelligence. Organizations that bring together sites or companies that each ran their own procurement “inherit duplicate suppliers, inconsistent pricing, and invisible spend.” An algorithm cannot optimize spend that nobody can see. “AI’s first real job is making that fragmented picture visible before it can make it optimal.”
Bringing AI into that environment created a problem nobody had solved for her. As an early AI-focused project manager, she needed a way to decide when an AI tool could be trusted in an enterprise adoption decision, and no criteria existed. She built them, and that work became the failure-mode evaluation frameworks she now applies to AI adoption decisions.
Her conclusion is blunt. “Most AI procurement rollouts fail at governance, not technology.” The risk is specific to scale. “Before scaling across dozens of sites, you need a shared definition of an acceptable failure mode. Otherwise every site trusts the algorithm differently, and you lose the standardization that justified the investment in the first place.”
It is a pointed observation, because it means a rollout can undo its own business case. The reason to consolidate fifty sites is to make them behave consistently. An AI tool that one site overrides constantly and another accepts without question reintroduces exactly the inconsistency the consolidation was meant to remove.
On where the technology goes next, she is specific about what automation should and should not do. “The next era of AI procurement is about routing decisions, not replacing buyers. The wins come from letting automation clear the high-confidence, high-volume transactions and escalating the exceptions to a human. That is where the cost curve actually bends.”
The pattern runs back through her career. Earlier, she worked on a product lifecycle management deployment across more than 500 global sites, another exercise in getting many independent operators onto one system. She is an IEEE Senior Member, holds PMP certification and a master’s degree in management information systems from California State University, Long Beach, and has written for the product management publication Mind the Product.
The headline promise of AI in procurement is an algorithm that finds millions. Malla’s account suggests the order is the other way round. The millions are already there, spread across suppliers, prices, and sites that were never compared. The algorithm only finds them once somebody has done the slow work of putting everything in one place.
