Artificial intelligence–mediated management increasingly influences how organizations allocate work, evaluate performance, screen personnel, prioritize cases, recommend actions, and structure managerial judgment. Yet governance arrangements often emphasize model accuracy, transparency, or nominal human oversight without specifying how employees can challenge consequential decisions, obtain relevant evidence, reach an authorized reviewer, and secure a reasoned disposition or remedy. This Original Governance Framework Article develops a proposed architecture for organizational contestability in AI-mediated management. It conceptualizes contestability as a practical governance capacity linking notice, challenge grounds, explanation, evidence access, appeal, escalation, consequential human review, remediation, case closure, and organizational learning. The article distinguishes contestability from generic transparency, employee voice, perceived fairness, and symbolic human involvement. It further proposes that meaningful appeal requires more than an opportunity to speak: employees must be able to identify a decision, formulate a challenge, access relevant information, reach an accountable review pathway, and obtain review by an actor with sufficient competence and authority. The framework also separates human presence from remedial authority and explanation from evidentiary sufficiency. Its propositions are intended to organize future empirical inquiry rather than establish causal effects or implementation readiness. The contribution is therefore a testable governance synthesis for analyzing whether AI-mediated managerial decisions are practically challengeable, institutionally reviewable, and capable of producing accountable organizational response under bounded technological, occupational, and institutional conditions.
