Enterprise AI Decision Architecture

Model governance

Model governance covers model selection, evaluation, bias testing, and update policy. It is about the AI system. Decision Architecture is about the decisions the system participates in.

Human-in-the-loop

Human-in-the-loop is a design pattern, not a framework. It says a human reviews the output. It does not say which outputs, under what conditions, with what authority, or with what accountability. Decision Architecture provides that specification.

RACI

RACI assigns roles: Responsible, Accountable, Consulted, Informed. It does not provide the logic for routing decisions based on stakes, reversibility, or exposure. When an agent is the Responsible party, RACI leaves the governance questions unanswered.

AI risk management

Risk management categorises risks and sets controls. Decision Architecture determines how those controls are applied at the level of individual decisions within a live process. One is strategic; the other is operational.

Marketing Implementation

The marketing-specific implementation. Classifies every marketing decision as AI-Primary, Collaborative, or Human-Primary based on reversibility, audience exposure, and frequency. Three zones, one named human owner per Human-Primary decision, no governance vacuum.

Classification Tool

The upstream classification step. For each process step, routes to automation (known path, stable rules), AI agent (variable path, recoverable outcome), or human (irreversible, legally exposed, non-delegable). Determines which steps enter the decision architecture and which do not.

Deployment Failure Mode

Identifies why AI deployments stall at the gap between controlled pilots and production. The Accountability Gap, one of six infrastructure gaps the Pilot Trap names, is a decision architecture failure: no named human owner for agent decisions post-deployment.

Buyer Trust Model

The buyer-side of decision architecture. Before enterprise buyers extend decision authority to an AI product, they evaluate three trust layers in sequence: Technical, Operational, and Strategic. Decision architecture cannot be sold to a buyer who has not crossed all three.

Implementing Guide

The decision-authority layer. How to assign autonomy tiers to agent decisions, specify the required controls at each tier, and define escalation and rollback procedures before deployment. Includes a pre-deployment governance checklist.

Implementing Guide

The architectural pattern. How agents, automation, and humans share a process: agents for variable-path steps, automation for deterministic ones, humans for decisions that cannot be delegated. Includes shared state design, auditability requirements, and vendor evaluation questions.

Seven-component cost model. Blue input cells. Produces net ROI, benefit-cost ratio, and payback period.

Printable one-page summaries of AMDA, Pilot Trap, Trust Architecture, and Belief Bridge.

Open-source templates and implementation tools for enterprise agentic AI go-to-market.

What is Enterprise AI Decision Architecture?

Enterprise AI Decision Architecture is the discipline of determining who or what decides at each step of an AI-mediated process, under what conditions, with what authority, and with what accountability structure. It is distinct from model governance, human-in-the-loop design, RACI, and AI risk management. Enterprise AI Decision Architecture is the operational layer that connects all of those disciplines into a coherent framework for running AI at scale.

How is Enterprise AI Decision Architecture different from AI governance?

AI governance addresses policy, compliance, and oversight at the organisational or regulatory level. Enterprise AI Decision Architecture is the practitioner-level implementation: for each specific decision in a specific process, who decides, under what conditions, with what review mechanism, and with what named accountability? Governance sets the rules. Decision Architecture determines how those rules are applied to every step of every AI-mediated workflow in practice.

What frameworks implement Enterprise AI Decision Architecture?

The Augmented Marketing Decision Architecture (AMDA) is the marketing-specific implementation. The Decision Boundary Matrix is the upstream classification tool. The Pilot Trap framework identifies the Accountability Gap as the most common decision architecture failure in production deployments. The Trust Architecture framework addresses the buyer-side question of how enterprises evaluate before extending decision authority to an AI product. The Belief Bridge framework provides the messaging model for closing the buyer belief gap during the commercial process.