The Decision Architecture Discipline: A Proposal and Reference Model for the Age of AI
Every person and enterprise that makes consequential decisions already has a structure for making them, and most of these structures were never designed. They accreted from habit, hierarchy and the last crisis, and their cost accumulates as decision debt. Generative AI makes the problem urgent: when analysis becomes cheap, the frame, what the decider wants and the commitment become the scarce parts of a decision. This paper proposes the Decision Architecture Discipline, whose object is the structure of deciding: the standing structure in which consequential decisions are made, held and revised. It sets out a concept system for that structure. It states a foundational principle, the Dual Lens Principle: the desired future state is described before the present is read, and neither is complete until both are held together and compared. Seven further principles follow, among them the principle of AI participation: AI proposes, sorts, classifies, surfaces and assembles, and the decider elects, confirms, adjudicates and rejects. The paper's reference model, the Decision Architecture Reference Model, has fifteen parts and seven load-bearing choices settled in order, and two models are fitted to it. Solo Decision Architecture, whose framework is specified in usable form, is for a consequential decision held by one person. Business Decision Architecture is a proposed model for decisions made on behalf of an enterprise, not yet operationally specified or demonstrated. The paper does not claim the term "decision architecture," in print since at least 1986, nor to be the first to design a standing structure of deciding, which behavioral strategy, organizational decision quality and decision hygiene already do. Its claim is that, in the literature it examines, no established field takes that structure as its central object for one person and for an enterprise as one thing. It states its claims about effects as hypotheses, with the findings that would count against them.
Authors
- Monica M. Hernandez (ORCID: https://orcid.org/0009-0006-3687-9530)
- Daniel A. Montero (ORCID: https://orcid.org/0009-0007-9131-8557)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2026-11-20
- DOI
- https://doi.org/10.5281/zenodo.23626397
- Primary Topic
- Leadership, Behavior, and Decision-Making Studies
- Type
- article
- Field-Weighted Citation Impact
- 0.00