Even in the absence of a consistent connection structure between institutional and technical domains, organizations must still deploy and operate AI in actual business processes. As a result, adopters ultimately have to design the connections themselves: how institutional requirements are translated into policies and operational rules, how those requirements correspond to technical controls, and to which AI Decisions and Executions they apply.

In practice, many of these connections are not necessarily defined as an explicit Architecture. Instead, they may be supplemented by operational tacit knowledge through domain knowledge, individual experience, coordination across organizational units, existing workflows, and system-specific configurations.

As a result, even where institutional and technical elements are connected, the way they are connected may differ across Organizations and Systems, forming complex knots. Each time a new regulation, Model, Agent, technology, business process, or Risk is introduced, those knots may need to be adjusted individually.

The need for connection does not disappear. The burden of making those connections is transferred to the adopter.