Two Frameworks, Same Blind Spot
According to Procurement Insights, two AI governance graphics drew heavy LinkedIn attention this week. Vaibhav Aggarwal outlined seven pillars of an AI governance framework. Matthew Coombs mapped the governance documents companies need to standards including the EU AI Act, ISO 42001, and NIST. Both work as checklists for what to write, who should own it, and which rules apply.
The author argues neither graphic answers what happens when an AI outcome falls short: who can intervene, whether they can act before damage is locked in, what people are rewarded for, and whether anyone can rebuild the decision months later.
The Filing Cabinet Model
Together the graphics list dozens of policies, registers, inventories, and regulatory mappings. That is governance as designed, documentation that answers what should be on file.
What the cabinet model struggles to capture includes performance metrics, informal workarounds, and tools never formally approved. A March 2026 survey from TrustedTech and Censuswide covering 2,001 UK and US employees, as reported by CSO Online, found nearly two-thirds of senior decision-makers admitting they used AI tools their own organizations had not approved. No register in the cabinet lists those.
Procurement Insights notes the gap is less about AI technology itself than the operational strands AI touches.
Authority, Clock, Incentive, Record
Operational governance starts from how work actually runs. When AI falls short, leaders trace the strands behind the outcome, including hidden ones, and test four intersections:
Authority: who can act. Clock: whether they can act before the loss is final. Incentive: what the rule rewards. Record: whether the decision can be rebuilt later.
What happens in practice should feed the record, so documents describe live operations rather than a designed ideal. Outcomes loop back: the operation learns and rules update.
Why Procurement and Finance Should Care
Formal files still matter. Regulators expect them and auditors need them. But a filing cabinet is only as good as the operation it describes.
For teams running buying, AP, and supplier workflows with embedded AI, governance that stops at templates leaves shadow tools and tacit shortcuts ungoverned, exactly where failure costs often accumulate.
The piece closes by arguing teams should test at least one real decision against how operations run, not only against policy shelves.
