30% efficiency claim from AI experiment

A social-media post about an artificial-intelligence procurement experiment said the system found workflow “leaks” and lifted efficiency by about 30%. (x.com) Procurement is the part of a company that buys goods and services, and its delays often hide in scattered chats, draft quotes, and contract handoffs. McKinsey wrote in April 2025 that contract optimization and contract compliance remain major sources of procurement value leakage. (mckinsey.com) The post described “leak networks” as issues like stuck conversations and unowned supplier quotes that slow source-to-contract work. It presented the result as a social experiment rather than a controlled study or audited deployment. (x.com) That matters because procurement teams have been testing generative artificial intelligence in exactly these steps: spend analysis, contract work, purchase-order processing, and supplier support. The Hackett Group said 49% of procurement teams piloted generative artificial intelligence in 2024, up from 23% a year earlier. (thehackettgroup.com) The same Hackett report said early users saw improvements of up to 10% across productivity, effectiveness, and user experience, with some cases above 25%. That puts a 30% claim near the high end of what consulting and industry reports have described so far. (thehackettgroup.com) KPMG said in its procurement survey and simulations that generative artificial intelligence could produce large time savings in pilots, but it also said enterprise-wide impact “is far from clear” and that large-scale benefits are still outliers. The firm added that few companies had fully deployed the technology organization-wide. (kpmg.com) Adoption has moved faster in procurement than in many other business functions, according to a March 2026 summary of research from AI at Wharton and GBK Collective. That summary said 94% of procurement teams now use generative artificial intelligence tools, up from 50% in 2023. (artofprocurement.com) The open question is measurement. The post did not publish sample size, baseline cycle times, error rates, savings definitions, or a method for separating artificial-intelligence gains from process cleanup or closer management. (x.com) For procurement leaders, the claim fits a real pattern: artificial intelligence is being used to spot where work gets stranded between sourcing, quoting, negotiation, and contracting. The evidence in this case is still anecdotal, so any team trying it would need its own before-and-after data to know whether a 30% gain is real. (x.com)