The intelligence community is actively grappling with the integration of AI into analytic production workflows. The professional consensus is converging on two requirements: prompt engineering proficiency and deep account mastery. Both are necessary. Neither addresses the structural problem this paper identifies: AI systems operating in intelligence production environments accumulate orientation drift, role fusion, and reasoning residue across sessions in ways that output-layer tradecraft standards cannot detect or correct. This paper distinguishes between output governance—the application of tradecraft standards to finished intelligence products—and orientation governance—the structural maintenance of a model’s authorized reasoning posture across the full production lifecycle. It then proposes an eight-layer AI-enabled intelligence framework that integrates orientation governance as a foundational substrate beneath the analytic production stack, and argues that without this substrate, tradecraft compliance at the output layer is applied to a model whose orientation may already have drifted from authorized parameters.
Narnaiezzsshaa Truong (Thu,) studied this question.