A bank is scaling multiple AI initiatives across retail banking, wealth management, and operations. Each business unit leader defines their own success criteria and governance approach, causing duplicated effort. What role would an AI center of excellence (COE) best serve in this situation?

A COE's core value in a scaling scenario like this is coordination — giving business units shared standards and reusable methods so retail banking, wealth management, and operations stop duplicating effort, while each unit still owns execution of the initiatives closest to its own domain expertise. Having the COE take over execution of every AI initiative goes too far the other direction; it removes business-unit ownership and turns the COE into an operational bottleneck instead of an enabling function that scales coordination across units. Requiring the COE to approve or reject every individual use case company-wide before any team proceeds creates an approval gate that slows initiatives down without matching benefit, since not every use case carries the same risk or needs the same scrutiny. Limiting the COE to occasional AI literacy workshops understates what it can offer; that role wastes its potential to solve the actual problem described, which is duplicated standards and governance, not a literacy gap. The operating principle: an effective COE enables and coordinates scaling through shared assets and standards, rather than absorbing execution authority that belongs with business units or shrinking into a narrow training function. Operational check: does the COE's involvement reduce duplication, or does it just add another layer business units have to get through?