VERIFY Method (Organisational AI Capability)
A six-move method for building AI capability at the organisational altitude rather than the individual one. The premise: AI value capture comes from how work is organised, not from how skilled individuals are. Each letter is anchored to an independent evidence source, and each produces a concrete artefact. Runs as five sessions producing five artefacts, which assemble into a 90-day plan.
When you need to design a complete learning experience from scratch
You're planning a workshop, training, or learning session and need a proven structure to organize your content and activities.
Use when an organisation is answering the AI question with training and getting no measurable value. Use it before a tool rollout, not after. It is deliberately not an AI-fluency course: it assumes fluency is the floor and workflow redesign is the lever.
V — Verify the value: find where AI value actually sits in the work, before any tool decision
E — Establish the standard: write, in one page, what good looks like and who signs off
R — Redesign the work and the roles: rebuild the workflow as if AI were there from step one
I — Invest in coaching: budget for coaches embedded in teams, not trainers delivering curriculum
F — Front it from leadership: senior people use AI visibly, moving from we are doing to I am doing
Y — Yield lasting capability: sequence the five levers into one system with output metrics
Ensures your session has clear goals, logical flow, and measurable outcomes.
The six letters group into five phases of work: V (frame and verify the value); E+R (set the standard and redesign the work); I+F (coach the work and lead from the front); Y (sequence the system); and an integration phase that locks the plan and the accountability relationships. Leadership (F) is a thread from day one, not only a fifth step — F is where you make it systematic.
- 1Start by defining what success looks like at the end
- 2Work backwards from outcomes to activities
- 3Build in checkpoints to verify learning
- 4Allow time for practice and application
- Organisations past the pilot stage with no measurable AI value
- Leadership teams who have bought licences and seen nothing change
- L&D functions being asked to build AI capability with a curriculum
- Transformation leads who need artefacts, not enthusiasm
- Training is the wrong lever. Workflow redesign is the number-one organisational predictor of AI EBIT impact (McKinsey, ~2,000 firms), and only 21% have done it.
- The jagged frontier is real. Inside it AI lifts quality 40%; outside it, people with AI are 19 points more likely to be wrong than people with none. The standard is the judgement of when to trust.
- Capability lives in the system, not in the heads. Novices gained +34%, top performers ~0% (Brynjolfsson, 5,179 agents). Exposure happens by default; capability accumulates by design.
- Coach, do not train. A coach is a colleague slightly ahead, embedded in the team, with protected time. A trainer delivers content to a room.
- Leadership visibility is a forty-point swing. Positive sentiment moves 15%→55% on one behavioural variable (BCG).
- Measure output, never attendance. Licence uptake, completion rates and literacy scores corrupt the programme.
- Capability compounds. The first workflow redesign takes twelve weeks, the second six, the third three.
- It will not make participants AI-fluent. That is the floor, and a separate sibling programme.
- The bottleneck you find in V is usually political — it is somebody's job. That is why almost no one looks.
- Without a named sponsor who has agreed the programme is happening, the F conversation never gets scheduled and the method stalls.
- Apply pages are the deliverable, not a warm-up. Budget at least an hour each.