// Start here
Decision Engineering™ examines how institutional purpose and policy become actual human and automated decisions, where that chain breaks, and how control can be rebuilt.
Institutions do not usually fail because someone made one catastrophic choice. They fail because a decision made at the top quietly changes shape as it travels — through policies, teams, systems, and now AI agents — until what is executed at the front line is no longer what the institution decided.
The result may be operationally correct, technically compliant — and completely different from what was intended. As execution shifts from human judgment to automated and agentic systems, this happens faster, at greater scale, and with less visibility.
A board sets the purpose. Grow prudently. Treat customers fairly. Stay within risk appetite.
Policies translate it. The intent becomes rules, thresholds, and mandates.
Teams interpret it. People decide what the rules mean in their context.
Data and technology encode it. Interpretations become models, parameters, and system logic.
People and AI systems execute it. Thousands of decisions a day, each carrying — or losing — the original intent.
At every transition, the original decision can quietly change.
Each step in the journey usually has an owner. The board owns purpose. Compliance owns policy. Technology owns systems. Risk owns models. Every component is governed — on its own.
What is rarely governed is the handoff: the join where policy becomes interpretation, interpretation becomes code, and code becomes execution. Authority, intent, and accountability separate quietly at these joins — and no dashboard is watching them.
This slow separation is decision drift: the growing distance between what the institution decided and what its people and systems actually do.
Typically focuses on the design, use and oversight of AI models and systems, including safety, fairness and explainability.
Typically focuses on exposures, controls and the likelihood and impact of what could go wrong.
Focuses on improving choices through better evidence, analysis, judgment and decision processes.
Examines the complete journey through which institutional intent becomes a human or automated decision, including the connections between governance, policy, data, technology, execution and accountability. It connects the components across the full decision chain.
The Decision Integrity Chain™ (DIC™) maps the eight layers a decision passes through inside an institution — from institutional purpose, through authority, context, rules, data and execution, to accountability and feedback.
The chain gives boards and executives a shared map: where a decision currently is, which layer it is drifting in, and which join it broke at. Every case, paper, and engagement runs on this framework.
Read the framework introduction — The Foundation →In human-led decision systems, authority and accountability could usually be assigned to identifiable roles. As AI systems take on more authority, the system acts while accountability may remain dispersed, distant or undefined.
The wider that distance, the weaker the institution's ability to answer for its own decisions.
When an AI agent moves money, the institution must be able to reconstruct that decision later — the way a flight recorder reconstructs a flight. Not just what happened, but who or what had authority, which rules fired, what data was used, and whether execution stayed aligned with intent.
A log is not necessarily a decision record. Logs capture actions; a decision record reconstructs the authority, context and reasoning across the complete chain. If the chain can be replayed, control is real. If it cannot, control is an assumption — and the institution discovers the difference at the worst possible moment.
This is the argument developed in The Irrecoverable Institution: replayability, not explainability, is the governance standard for agentic banking. See the research →
The Decision Drift Audit™ maps all eight DIC™ layers in 10 days and delivers one board-ready finding. Fixed fee.
The Decision Integrity Blueprint™ defines what closing each exposure requires — written for the board, not IT.
Remediation runs through your teams or an agreed delivery partner. Closure is independently assessed against the DIC™.
Banking, healthcare, AI governance, and institutional transformation — in almost every documented failure, a decision problem underneath: who had authority, what the system was optimising for, who owned the outcome, and whether anyone could reconstruct what happened.
Knight Capital — when automated execution outran institutional control →
Silicon Valley Bank — when speed, evidence and governance stopped moving together →
When clinical intent and system logic quietly diverge →