McKinsey: agentic AI in sourcing stalls at pilots because decision workflows stay unchanged

McKinsey's October 28, 2026 article on agentic AI in strategic sourcing documents a widening gap between procurement AI ambition and operating reality. In its Trends in Procurement Organizational Design survey of 200 chief procurement officers and senior procurement leaders, fielded in June 2026, 80 percent of respondents plan to increase procurement technology and AI investment over the next 12 to 18 months. Yet only 2 percent report AI embedded at scale, and 55 percent remain in pilot mode.

The article's central claim is that the constraint is organizational, not technical. Most functions deploy agents against individual use cases, such as drafting category strategies or summarizing supplier data, while leaving the end-to-end sourcing workflow and its decision points untouched. The value McKinsey associates with agentic AI sits upstream: continuously assembled market and supplier data, earlier identification of opportunities and risks, and sourcing decisions rethought around what agents prepare and what humans decide.

Dataleo reads this as a decision-allocation problem. Scaling an agent means deciding, in writing, which decisions it prepares, which it executes and under what thresholds, and who is accountable when a supplier-facing commitment fails. Without that allocation, pilot mode is not hesitation; it is the rational response to unassigned risk. The survey numbers describe intent and self-reported maturity, not audited outcomes, and the article's prescriptions reflect McKinsey's consulting vantage point.

Image disclosure: the header illustration is a thematic Pexels stock photograph of a business meeting (photo 36765714), free license verified on the source page on October 31, 2026.