THE LEARNING EXPERIENCE ENGINE

Inside Madgwick

Argue the case. Play the counterpart. Run the simulation.
A conversational, AI-native environment that simulates experience, personalises at scale and delivers assurance of learning.

THE ANATOMY
01 · MADGWICK STUDIO · THE LEARNING ENGINE
STUDIO · LESSONS

Every lesson is a conversation

Dialogue, roleplay and simulation in focused 3-5 minute loops. Madgwick asks questions back, tests reasoning and stretches judgement - and learners demonstrate understanding before they progress.

Dialogue-based learning. Learners think out loud; the AI probes, never lectures.
Roleplay and simulation. Realistic scenarios with pressure, ambiguity and competing perspectives - judgement practised before the stakes are real.
Safeguarded from day one. A mandatory AI-readiness course before learners commence, and emotional-safety triggers if a learner starts engaging with the AI as though it were human - with escalation to real people.
Capability gates. Progression is tied to demonstrated capability, not calendar time. Start anytime, finish as fast as the evidence allows.
Learn Social Support
LEARNER CONTEXT
Registered nurse · Emergency · Regional
12 lessons · 4 units · context carried
MADGWICK

Welcome back. No need to introduce yourself again. Last time you acted on the trend rather than the third reading, and you set the threshold in advance.

Same ward, harder case this time.
MADGWICK

Then we will use your netball coaching for the part about calling it early in front of other people.

Type a message…
STUDIO · PERSONALISATION

The learning is the same. The context is theirs.

A chat alongside every lesson lets learners share their interests, industry and goals - and the learning reshapes around them, right there in Madgwick. A public health case for the nurse; a supply-chain case for the logistics manager. Same objectives, met through their world. Read it, watch it, or listen on the drive to work.

Profiled at onboarding. Interests, industry and goals captured from the first session and carried into every interaction.
Same standards for everyone. Objectives and assessment standards are constant; only the path to them adapts.
Read, watch or listen. Every lesson meets the learner in the format that suits them.
Personalisation compounds. Context persists across lessons and units instead of resetting - at a scale no human-only model could staff.
STUDIO · EARNED LEARNING

Remove the friction that blocks learning. Preserve the friction that builds it.

The aim isn't easier learning - it's making the right parts harder to bypass. Clunky systems and admin load get total AI support; working through a problem, defending a decision and navigating ambiguity stay with the learner, and judgement stays with the educator.

The Resistance Gradient. AI assistance is calibrated inversely to the cognitive demand of each task. Deep thinking gets Socratic questioning; logistics get total support.
Bounded AI. The AI asks questions back, tests reasoning and surfaces uncertainty - never a faster answer machine.
A published pedagogy. Formalised in the Calibrated Friction Architecture, led by academics and shaped by teaching staff.
FRICTION REMOVED
Clunky systems
Unclear instructions
Administrative load
FRICTION PRESERVED
Defending a decision
Navigating ambiguity
Improving through feedback
The Earned Learning principle · Read the essay
STUDIO · ASSURANCE OF LEARNING

Proven, not assumed

No surveillance, no single gate. Layered, non-invasive verification turns the unanswerable question - did they use AI? - into the answerable one: can they demonstrate the capability?

Two-lane assessment. Assurance Tasks are secure and verify genuine capability; Learning Tasks stay open and developmental.
The Swiss Cheese model. Multiple independent layers, each imperfect alone, aligned so what slips through one is caught by another.
The tutor is never the examiner. One AI supports the learning; independent layers assess it.
Humans hold the judgement. AI informs assessment. It never determines grades.
Visibility for organisations. Progress against objectives, engagement evidence and support needs - better timed, better targeted.
PROGRESS AGAINST OBJECTIVES
VISIBILITY
ULO1Capabilities and limits82%
ULO2Verification methodsNEEDS SUPPORT41%
ULO3Risk and mitigation68%
ULO4Responsible decisions63%
02 · THE COMMONS · THE HUMAN LAYER
THE COMMONS

Where the technology steps back

Education scaled at the cost of connection - and AI now makes it possible to wind the clock back. The Commons is where learners meet educators and each other: live sessions, study circles and mentorship, built into every unit. Humans focus on what they do best: contextualising knowledge, challenging ideas, forming relationships.

Smart pre-briefing. Before an educator engages, the system generates a learning narrative for each learner - patterns, breakthroughs, struggles - not just grades.
Intervention triggers. Academics learn who needs reaching before they disengage, not after.
Retention infrastructure. Belonging is the strongest predictor of persistence online. Connection is woven through every unit, not left to chance.
The human return. By absorbing the repetitive load, the model is designed to increase the share of academic time spent in direct engagement with students.
03 · FOUNDRY · THE EDUCATOR LAYER
Lesson 4.4: ROI frameworks for AI investments
Suggested update for AIB102
REVIEW
SUGGESTED UPDATE ADDED

In 2026, the evidence base for AI ROI has matured considerably. Returns are highly skewed. A small group of leaders captures disproportionate value, while many organisations report modest or unmeasurable returns (Harvard Business Review, 2026).

Text
applied
Podcast
queued to record
Video
queued to record
Accept changes Reject Edit manually
FOUNDRY

The educator holds the reins

Foundry is the professional workshop behind the learner experience. Educators and learning designers build, manage and govern everything learners see - from authoring Earned Learning sequences to approving the updates that keep programs current.

One precise map. Course → Teaching Units → Topics → Lessons → Learning Objectives. The AI tutor and assessor read from the same structure educators author.
AI tutor configuration. Educators shape how the AI tutors, prompts and responds - it never freelances.
Assessment design. Define how capability is practised and evidenced, lane by lane.
Permissions and analytics. Role-based access across the institution; engagement, outcomes and gaps in real time.
LIVING CONTENT

Content that never falls behind

If the field moved yesterday, the course can carry it today. AI surfaces new research, developments and real-world examples as they emerge, and the educator gets the notification in Foundry - accept, and it flows into the courseware; decline, and it never reaches a learner.

Surfaced by AI. Agents scan new research, industry developments and current events relevant to each program.
Verified by educators. Nothing reaches learners without expert review and approval in Foundry.
Applied immediately. Approved updates flow into lessons, examples and scenarios - every program stays current.
ADOPTION

AI involvement is a dial, not a default

Every organisation chooses its level - per course, per cohort.

LEVEL 1

A modern learning environment

AI is off. Cleaner interface, better infrastructure. Change nothing about how you teach.

LEVEL 2

Grounded student support

AI answers questions from your materials. Grounded, contextual, between classes. You keep control.

LEVEL 3

Earned Learning, live

AI becomes an active participant: Socratic dialogue, roleplay, simulation. A separate assessor verifies the capability is real.

GOVERNED THROUGHOUT
Sovereign-hosted Single-tenant option Conversations stay private AI never determines grades Emotional-safety safeguards White-label ready WCAG 2.1 AA

See it working, live.

Request a demo →