Home / The Methodology

The methodology

From AI learning needs to capability at work.

A structured method for enterprise learning teams: diagnose role-specific needs, design and pilot relevant learning, assess practical performance and measure application in the workplace.

A facilitated workshop connecting learning objectives with workplace tasks — Illustrative workshop setting.

The principle

Start with the work people do.

An AI course is useful only when it addresses a real task. We begin with a defined use case, the people who interact with it and the decisions they must make. From there, we define what they need to know, practise and demonstrate.

Learning & Development leads the programme with the business owner. Governance and risk teams clarify responsibilities and controls; technology teams confirm what the platforms can support. We bring these perspectives into one learning design.

One traceable learning pathway

From requirement to action at work.

Each connection should be explainable: why a role needs a competence, where it is developed, how it is assessed and what shows its application in the workplace.

Requirement

Role

Outcome

Learning

Assessment

Application

The six phases

A practical route from diagnosis to improvement.

Each phase has a defined purpose, a tangible output and a decision point. The sequence can be adapted to the client's maturity and delivery model.

Align

Define the work and the intended change.

We agree what problem the programme should solve, who it affects and what responsible performance looks like.

What we do

Interview the L&D sponsor, use-case owner and relevant governance or risk owners.
Map employee groups, tasks, decisions, supervision and escalation points.
Identify applicable organisational requirements and the constraints of the operating environment.

Outputs and decision

Scope, stakeholder and role map, priority use cases and initial success criteria.

Decision: approve the population and tasks to assess.

Diagnose

Establish a credible baseline.

We examine current capability and the quality of the training and tools already in use.

What we do

Review curricula, materials, completion data, existing assessments and selected work samples.
Use interviews, workshops or practical tasks to identify knowledge and judgement gaps.
Examine whether the LMS supports the required journeys, assessments, feedback and evidence.
Record the source and confidence of each significant finding.

Outputs and decision

Training needs baseline, competency gap map, platform findings and prioritised constraints.

Decision: validate the gaps the programme must address.

Design

Turn gaps into a learning architecture.

We design the programme around observable behaviour and decide how success will be evaluated before delivery.

What we do

Define competencies and proficiency levels for each role.
Write learning outcomes that connect to specific workplace activities.
Design pathways, formats, content requirements and realistic assessment scenarios.
Set rubrics, feedback, remediation, platform requirements and an evaluation plan.

Outputs and decision

Role-based programme blueprint, assessment design, LMS requirements and measurement plan.

Decision: approve the pilot design and acceptance criteria.

Pilot

Test the complete learning experience.

A bounded group lets us check relevance, usability and operational fit before expansion.

What we do

Prepare the pilot with L&D, business owners and technical suppliers.
Review materials, access, accessibility, learner instructions and platform configuration against the design.
Run the learning and applied scenarios; collect participant, facilitator and manager feedback.
Track issues, exceptions and the effort needed to operate the model.

Outputs and decision

Pilot results, learner experience findings, issue log and revised programme elements.

Decision: determine what to correct before wider use.

Evaluate

Measure learning and application.

Course satisfaction and completion are useful signals, but they cannot alone establish workplace capability.

What we do

Review relevance and learner experience, then test knowledge and practical performance.
Observe selected tasks after training, using managers or appropriate process owners.
Compare relevant operational indicators with the agreed baseline where possible.
Separate learning gaps from barriers caused by tools, authority or processes.

Outputs and decision

Evaluation report, assessment results, workplace evidence and targeted remediation.

Decision: judge readiness within the agreed scope and evidence limits.

Improve

Refine, assign owners and scale.

Learning needs change with AI use cases, policies, roles and platforms; the programme needs an operating owner.

What we do

Prioritise corrections to content, assessments, workflows and platform support.
Define ownership for materials, evidence, review cadence and change triggers.
Plan expansion to further roles, business units, languages or markets.
Set the next measurement cycle and criteria for continued improvement.

Outputs and decision

Improvement backlog, ownership model and phased rollout roadmap.

Decision: expand only when the required conditions are met.

Evaluation and evidence

Measure whether the learning changed practice.

We agree the baseline, data sources and review points during design. Kirkpatrick's four levels organise the questions; measures remain proportionate to the role and use case.

01 / REACTION

Was it relevant and usable?

Participant feedback on task relevance, clarity, accessibility and confidence. We review the experience, not just a satisfaction score.

02 / LEARNING

What can people demonstrate?

Knowledge checks and practical scenarios with defined criteria, feedback and remediation. A certificate of attendance is a separate record.

03 / BEHAVIOUR

What happens at work?

Manager observation, sampled tasks and appropriate workflow evidence after training. We consider access, authority and tool quality.

04 / RESULTS

What improved for the organisation?

Relevant measures may include error patterns, rework or escalation quality. We interpret changes against the baseline and other influences.

The evaluation distinguishes evidence of learning from a claim that training alone caused an operational result. Findings include limitations and the strength of the evidence.

Illustrative application

An approved AI assistant in customer service.

Advisers use AI to draft customer replies. The programme aims to help them verify accuracy, protect information and recognise when to escalate.

Two colleagues reviewing an AI-assisted response against assessment criteria — Illustrative image of applied assessment.

This is an illustrative method example, not a client case study or a claim about a particular system.

Task

Review an AI draft before it is sent to a customer.

Learning outcome

Identify inaccuracies, sensitive information and escalation triggers.

Assessment

Work through realistic drafts using an agreed accuracy, privacy and judgement rubric.

Workplace evidence

Sample reviewed responses and manager observations after the pilot.

Improvement

Adjust learning, prompts, guidance or workflow according to the observed gaps.

Methodological foundations

Education first, connected to governance.

Each reference informs a different part of the work. The method is adapted to the client's context; using a standard as a reference does not imply certification or legal conformity.

LEARNING MANAGEMENT

ISO 21001:2025

Informs how we consider learner needs, objectives, responsibilities, delivery, feedback and improvement of educational processes.

DIGITAL LEARNING QUALITY

ISO/IEC 40180:2017

Provides a reference framework for quality in technology-enhanced learning, including review of the platform-supported experience.

EFFECTIVENESS

Kirkpatrick

Structures the evaluation of reaction, learning, workplace behaviour and organisational results.

Governance defines what people need to do

The EU AI Act and the organisation's own requirements help determine responsibilities in each use case. ISO/IEC 42001 and ISO/IEC 23894 support the management and risk perspectives that inform the learning objectives and operational controls.

References: EU AI Act Service Desk · ISO/IEC 42001 · ISO/IEC 23894 · ISO 21001 · ISO/IEC 40180.

Apply the method

Begin with one priority population or AI use case.

A bounded Learning & Capability Baseline gives the L&D team a role map, evidenced gaps, platform findings and a plan for a measurable pilot.

Discuss training needs