What leaders need from AI training: decisions, not demos

Why does leadership support decide whether AI is adopted?

Because staff take their cue from what leaders fund, protect and ask about. Where employees see a clear AI strategy from their leaders, 80% report measurable impact from AI; where that clarity is limited, 55–60% do (BCG, AI at Work 2026; Indicative). The tools are the same in both groups. The difference is whether someone senior has said what AI is for and what it may do.

The results follow the same pattern. Nearly nine in ten organisations now use AI regularly, yet only 37% attribute any earnings impact to it (McKinsey, The state of AI in 2026; Indicative). The organisations getting most from AI are twice as likely to have senior leaders who are committed to it and a way to measure its impact (same source; Indicative).

Commitment, in practice, means decisions: which ideas are funded, who owns them, how much the system may do, and how success will be judged. That is what executive AI training should produce.

What must a leader be able to judge?

Not how a model works. A director general does not need to write prompts well; they need to judge proposals well. Four judgments matter:

• Value. Which problem does this idea solve, for whom, and how would we know it worked? A measure taken before and after, not a demonstration.
• Authority. How much is the system allowed to do before a person signs? At Cader Group we use the Delegation Ladder for this: four levels (Assist, Draft, Advise, Delegate), each tested on whether a mistake would be recoverable or binding.
• Readiness. What must exist before the idea is safe to fund: an owned data source, a named approver, a record of past outcomes, a way to reverse errors.
• Ownership. Who is accountable for the idea after the session, and by when they report back.

A leader who can make those four judgments can say no to a polished demo and yes to a modest idea that will work. That is the capability that matters at the top.

What does a one-day format that ends in decisions look like?

The structure we use for leadership sessions has three parts, and each produces something.

Morning: from many ideas to a few. Departments bring their ideas in advance. The committee tests each against value and readiness, merges duplicates and drops what cannot be measured. The output is a shortlist of five or fewer.

Midday: place each idea on the ladder. For each shortlisted idea, the committee agrees its level and its consequence, names an owner and lists what must exist before funding. Ideas that are unsafe as Delegate are often fundable straight away as Draft.

Afternoon: decide on the first one. The committee works through one idea in depth and writes a Decision Brief: the problem, the level of authority, the approval point, the measure of success, the owner and the next step. The session ends with a decision: fund now, fund once named conditions are met, or stop.

This is the format of AI Decisions for Leaders: one day, ending in a prioritised AI idea list and a Decision Brief on the top idea.

What should leaders bring on the day?

Preparation decides how far the day gets. Ask for four things in advance:

• The ideas, written down. One paragraph each from the departments: the problem, who it affects and what would change.
• The rules. Which AI tools are approved, and which data may and may not be used.
• Current measures. For the processes the ideas touch: volumes, turnaround times, error or rework rates, where they exist.
• The people who can say yes. If the person who controls the budget is absent, the session produces recommendations, not decisions.

Chiefs of staff who book these sessions can do most of this in a week. Our Readiness Snapshot is a quick way to establish the starting point beforehand.

What happens after the session?

A Decision Brief is only useful if it moves. Three things should follow within weeks, not quarters:

1. The owner reports back on the conditions attached to the decision, such as an owned data source or a named approver.
2. The team that will do the work is trained on it. The people who will use the agent or workflow learn to build and check it on their own cases.
3. The idea is measured against the baseline set in the brief, and moves up or down the ladder on that evidence.

Where the question is larger than a single idea, such as how roles, structures and processes should change, a two-day program such as Organisation of the Future produces a Future Organisation Blueprint and a 12-month roadmap with owners.

Measure where your own team stands: take the free Readiness Snapshot →