Method How the human layer is read, and where it changes.

Which of these sentences have you said?

Read eight signals along four tension lines. Pick a sentence to see what may sit underneath it and where a leader can start.

It doesn't replace an adoption program. It completes one.

The human layer plugs into an existing AI adoption program at three points.

What your dashboard doesn't show

A usage dashboard tells you how many people open AI. It doesn't tell you where the work really stands. The Adoption Quality Map reads one workflow with three questions and places it in one of five spots: two healthy, two that need a closer look, and one hidden risk.

The map describes the work, not the person. No score; you get a place on the map and the questions you couldn't answer. Your answers never leave your browser; no email needed.

A decision for each workflow

The Human Layer Sprint exists to reach a decision, not to produce a report. Evidence and field tests lead to one of five decisions:

Adoption breaks in two directions. Too much carries its own cost.

Some teams keep away from AI. Others cling to it: they want to be seen using it, hand over the decision, and try to solve everything at once. On a dashboard, the second looks like success.

Adoption shows up in nine moments

The eight points above are what a leader feels. What actually shows up in the field is nine moments: three in the under direction, three in the over direction, one in either, and two healthy.

A card names what a leader sees; the seven axes ask where it may come from. Each moment also produces a response in the leader. The diagnosis belongs to the moment, not the person: the same person may show a different pattern next month.

One loop, three points of intervention

A leader sees an AI-related behavior. Reads it. Makes a move that fits the reading. That move produces a new behavior on the other side — and that behavior seems to confirm the leader’s first reading.

The critical point: this loop runs in the leader and the employee at the same time, and each feeds the other. The employee’s silence grows the leader’s impatience, and the leader’s impatience grows the silence. Most of the time nobody acts in bad faith; the system repeats itself.

If you’d like to start with yourself

A short twelve-question self-assessment. What it does not measure: tool knowledge, prompting skill, technical competence. What it looks at: your habits in diagnosing and responding to AI-related behavior in your team.

The map looks at the work; this check looks at the leader’s own share. It gives no score or band and is not a psychometric test: it shows which habit is strong, which needs attention, and one behavior to try this week. In the program this check is never used alone: it is read together with team interviews, workflow data and leaders’ field tests. Your answers never leave your browser, and no email is required.

We look at three points

Good programs measure skill before, at the end and ninety days later. At the same three points we look at the workflow: who keeps the decision, have the conditions changed, is it safe to say "I don't know" in the team?

A plateau is not a failure report; it is a diagnosis. It usually points to a condition: no reinforcement, no support from the manager, no habit formed. Learning-transfer research shows that one of the decisive conditions for persistence is support from the participant’s own manager.

So the work doesn’t end in the room: a weekly check card runs for ninety days, and the participant’s manager receives a separate sponsor briefing.

What we say, and what we don't

The figures below are evidence of the problem, not of our method. The method is new and in its founding-pilot stage: we designed it, and we will test it with organizations in the first pilots.

What this is not

Knowing the scope up front prevents disappointment halfway through.