Assess practical AI skills, deliver role-specific learning, standardise proven workflows, and measure capability growth across every team.
(a) The problem
Tool access proves procurement happened. It says nothing about whether a person can frame a problem well enough for AI to help with it.
Usage counts prove activity. They cannot distinguish a careful, reviewed output from a confident, wrong one pasted straight into a client document.
Training attendance proves an hour was spent. Reliable performance is only visible when someone does the applied work and someone else can check it.
(b) Operating model
Scenario-based assessment measures practical capability across all six dimensions, per person and per team.
Adaptive, role-specific learning paths target the gaps the diagnosis actually found — not a generic curriculum.
Proven prompts, guides and workflows are reviewed, approved, versioned and reused across teams.
Capability movement is tracked over time and exported as defensible evidence for boards and auditors.
(c) Capability model
Knows what modern AI can and cannot do.
Safe, lawful, privacy-conscious use.
Turns vague needs into measurable problems.
Designs repeatable human-and-AI workflows.
Gets real work done with AI.
Evaluates output before it is used.
(d) Role-specific learning
Problem Framing, Workflow Design
18 lessons
Applied Execution, Quality Judgement
14 lessons
Workflow Design, Applied Execution
16 lessons
Quality Judgement, AI Foundations
19 lessons
AI Foundations, Workflow Design
21 lessons
Responsible Practice, Problem Framing
13 lessons
Applied Execution, Responsible Practice
12 lessons
Quality Judgement, Responsible Practice
15 lessons
(e) Workflow library
Revenue Operations · 14 teams · 2,418 runs
Customer Success · 21 teams · 3,910 runs
Finance · 9 teams · 412 runs
(f) Governance
Nothing reaches the library without an owner, a risk classification and an approval. Every change is versioned, every review date is tracked, and every action is written to an immutable audit trail scoped to your organisation.
Security and governance →Named reviewer sign-off before publication.
Every workflow has an accountable owner.
Expiring content is flagged, not forgotten.
Full change history with diff and rollback.
Low, medium or high, with matching controls.
Categories that block publication outright.
(g) Outcomes
28%
improvement in assessed capability in 90 days
4.6 hrs
saved per employee per month
71%
of approved workflows reused by more than one team
Illustrative demo metrics from sample organisation data, not audited customer results.
(h) In their words
“We had licences everywhere and no idea whether anyone could use them well. The baseline told us in a fortnight, and the gap it found was judgement, not tooling.”
“Our audit committee asked how we govern AI use. We exported the evidence pack, workflow approvals and all, and the conversation was over in ten minutes.”
“The role tracks land because they use our own work. Support agents practise on escalation scenarios, not abstract examples.”
Composite examples using fictional organisations.
(i) Questions
One department, one assessment cycle, one defensible number to start from.
Start a capability baseline