The organisations that get the most from AI are not the ones that explore most aggressively. They are the ones that commit most deliberately.
Clay Parker Jones’s Permission to Move (2026) shows a clear bimodal distribution in AI adoption outcomes: organisations that have solved the structural permission problem for AI outperform those that have not by 1.44x (95% CI 1.27x–1.64x; Cohen’s d 0.85). Partial adoption — running pilots without committing to a process — is not a stable position. It trains people to be sceptical of AI investment, and leadership to be cautious of AI commitment.
Breaking that cycle requires more than a better tool. It requires a new way of working.
Most organisations that reach a commitment decision discover they do not have a sufficiently rigorous basis for making it. The business case was written to be approved, not to be tested.
Governance obligations — regulatory requirements, sector-specific rules, deployment constraints — are addressed at the headline level, not the operational level where implementation decisions are actually made. When the board approves, the commitment is already fragile.
The failure mode is not ignorance. It is approval-optimised reasoning: the case is constructed to clear the hurdle, not to survive contact with implementation.
The half-life of an approval-optimised business case is roughly the time between board approval and the first quarterly review. After that, the case has either survived contact with implementation or it hasn’t. There is no third outcome. The CFO who sponsored it knows which one is coming long before the board does.
The Commit stage produces a commitment statement that can be falsified.
A specific initiative, linked to a specific business process change, a specific performance metric, and a specific financial outcome. Not a declaration of intent — a statement that names what would constitute failure, and what conditions must hold for the initiative to succeed.
Alongside it: a full five-case business case (strategic, economic, commercial, financial, management) and board-ready presentation materials. Both validated against the Governance domain of the AI Value Intelligence before they reach the board — so nothing in the commitment needs to be walked back later. Board approval at this point of substance is the foundational act of executive governance — and what makes commitment durable rather than ceremonial.
Falsifiability is the test. A commitment built on it produces durable investment and genuine accountability. A commitment built to be approved produces a half-life measured in quarters.
Next: the AI Actionable gap — why AI Legible organisations stall, and the structural conditions that let practitioners move on AI initiatives without approval overhead.
Realise →