• integrated_82#Q1.11

RAIDT connects strategy to operational use by supplying a missing middle layer between organisation-level intent and one configured use of generative AI in work. The governance papers argue that strategy alone is too abstract and model documentation alone is too coarse. Organisations may have policies, lifecycle controls, and statements about responsible AI, yet still be unable to show what prompt version was used, what sources were retrieved, what tools were active, what safety settings shaped the output, or what human checks took place before the output was relied upon. RAIDT closes that gap by treating the run as the unit of governance and turning each material use into a reviewable evidence object.

In practice, strategic aims such as accountability, oversight, transparency, and dependable operation are translated into operational requirements through the run-level evidence pack and the score profile. The five pillars (Responsibility, Auditability, Interpretability, Dependability, Traceability) specify what must be evidenced if governance claims are to be inspectable at the point of use. Read this way, influence methods as governance interventions are the concrete means through which strategy acts on work: approved prompt structures, role assignments, human review steps, override rights, provenance capture, monitoring routines, and exception handling. Because these interventions are evidenced run by run, RAIDT allows managers to compare uses, sample audits, identify weak controls, and improve workflows without confusing strategic aspiration with operational proof.

A public-service team uses generative AI to draft explanations for housing-support decisions. The organisation's strategy says decisions must be reviewable, consistent, and accountable, but those goals remain abstract unless they alter daily work. RAIDT operationalises them by requiring a run-level evidence pack for each case: the approved prompt, the policy sources retrieved, the case data used, the model configuration, the reviewer who checked the draft, and any edits before release.

The score profile then shows whether the strategic controls are actually present in operation. If Auditability is strong but Dependability is only partial because retrieval sources vary too widely across cases, managers have a precise operational issue to fix. Strategy is therefore connected to practice through evidence-bearing routines rather than through policy statements alone.

• 15-RAIDT-IS-Governance_M_v07
• 13-RAIDT-Evidence-Review_M_v10