Standing up a cross-functional steering body that standardizes AI tooling, training, evaluation, and governance across business units. In 2026, Centers of Excel
Standing up an AI Center of Excellence creates the organizational engine for scaled AI: a hub team owning platform, standards, enablement, and portfolio prioritization, with federated delivery in the business units. Done well it converts scattered pilots into compounding capability; done badly it becomes a bottleneck bureaucracy: the difference is operating model design.
CoEs die two deaths: irrelevance (an ivory tower business units route around) or chokepoint (every AI idea queued behind central staffing). Survivors earn pull with platform value, push delivery capability outward deliberately, and report business outcomes, not activity, to the executive sponsor.
Small and senior: 5–10 people covering platform, governance, enablement, and portfolio typically outperform large mandates. Scale follows demonstrated pull from business units, not org-chart ambition.
No: central delivery caps throughput at headcount. The proven arc is hub-and-spoke: CoE builds the platform and first lighthouse cases, then enables units to deliver on shared rails.