For executives and strategy decision-makers

The discipline: what a well-made decision looks like

Before anything we sell you, this is the practice, and most of it takes a whiteboard and an honest hour.

Scope it first. Simon separated recurring decisions settled by an existing rule or template from novel ones that must be reached by judgement (Administrative Behavior, 1947; the programmed/non-programmed terms in The New Science of Management Decision, 1960). If a template reliably settles it, use the template; what follows is for the other kind.

Then: frame the decision and say what you are treating as fixed. Generate real alternatives, not one option and two straw men. Declare the criteria and their weights before anything is scored, because weights chosen afterwards are a preference wearing a method. Put the uncertain branches on a tree with expected values. State your risk posture explicitly rather than discovering it in the post-mortem. Ask what would have to stay true, and find the value at which it stops being true. Red-team the case, and name the bias each challenge is aimed at, Tversky and Kahneman's 1974 catalogue is still the working list. Then commit, and keep the reasoning rather than only the choice: Staw, in "Knee-deep in the big muddy" (Organizational Behavior and Human Performance, 1976), showed decision makers committing more resource to a failing course of action after the evidence had turned, particularly where they owned the original choice, which is why a decision needs a re-test rather than an anniversary.

And the frame that makes all of it bearable: the decision-analysis tradition growing out of Howard's work from the 1960s holds that a decision must be judged on how it was made (frame, alternatives, information, reasoning, commitment) because outcomes are contaminated by chance and arrive too late to steer by. You cannot control the result. You can control, and show, the quality of the reasoning behind the commitment.

The cost of doing this properly is never the framing. It is establishing what is true across systems no single function owns, and keeping it true afterwards. The expensive part is not drawing the model; it is building it and keeping it true.

What makes analysis decision-grade

The decision is framed as a system. The ecosystem it lands in is built out around it as a typed, parameterised network. The options are then tested with grounded analyses whose every claim cites a catalogued source. Three properties make that decision-grade rather than merely persuasive.

The reasoning is inspectable. An OWL-EL symbolic reasoner derives structural facts with the entailing axioms attached, kept visibly separate from LLM judgement and from simulation: a reader always knows whether a statement is judgement, proof or computation. Every AI interaction is audit-logged.

The reasoning is challenged before it reaches you. Every high-stakes output is red-teamed by adversary agents, so the load-bearing assumptions are hunted rather than hoped for.

And the decision itself is kept. A committed decision carries its rationale chain — the analyses, parameters, assumptions and sources behind it — and is re-tested when those parameters move materially. That is the difference between a decision you can revisit and a slide deck you can only re-read.

Two things we will say before you ask

First, SOSX supports the decision; it never makes it. The human stays in the chair, the stages that decide the model pass a human approval gate, and no part of this is autonomous.

Second, there are no measured decision-quality, decision-speed or outcome figures here; none exist, and we will not invent them. What the platform does is show its working, so a claim you repeat in a board paper is one you can trace back to its source, and Just Good Business points the same analysis at the decisions you take. We would rather show you it working on one of yours than quote a figure we cannot stand behind.

The governance point is not a compliance aside. In our live enterprise MBSE trials with a large global aerospace manufacturer, engineers interrogated the audit trail, source grounding and approval gates before the analyses, the governance substrate is what makes AI-generated analysis discussable in an engineering review at all. The same substrate is what makes it discussable in a board paper. And where the decision carries an AI-governance obligation of its own, the platform was built against the frameworks that obligation is judged by.

The question you are really asking

If we commit to this, what has to stay true — and how will I know when it stops being true?

The answer is not a recommendation. It is a rationale chain that stays attached to the decision and a re-test that fires when the ground moves: these were the analyses, these were the parameters underneath them, this is what has changed, and this is the part of the reasoning that no longer holds. A report ages quietly. A rationale chain tells you when it has.