AI Learning Consultant and Product Architect

Incident 9:42: Stabilize the Stack

Case study

TECHNICAL TRAINING

A technical incident-response simulation that turns a production outage into guided job practice.

Accessible incident-response training simulation with voiced coaching, guided onboarding, xAPI tracking, and technical L&D case study depth.

Client

Role

Sole builder, brief to handover

Scope

Branching simulation · voiced coaching · xAPI

Stack

Web simulation · Voiced coaching · xAPI events · Learner dashboard

4

PERFORMANCE PHASES

Find the evidence, stop the spread, recover the working version, explain the decision.

3

DIFFICULTY PATHS

Guided mode translates the technical language so a beginner can still make the call.

25

VOICED RESPONSES

Voiced coaching, hints and randomised decision cards.

100
%

ATTEMPTS TRACKED

Each attempt emits xAPI-compatible events: score, time, hints, retries.

chain:Evidence,Spread,Recover,Explain | tiers | dots | grid

How it’s built

Point at a step

Four job behaviours, agreed up front.

The run

transcript

1

Mission brief

A checkout outage at Relay Nine Commerce, failed requests multiplying.

The run

coach

2

Find the evidence

Diagnosis, with an interactive dashboard and a service topology.

The run

safety

3

Stop the spread

Containment, before the failure spreads further.

The run

loop

4

Recover

Recovery: back to the working version.

The run

persona

5

Explain the decision

Stakeholder communication, in plain language.

Coached and tracked

voice

6

Voiced coaching

Coaching and hints are voiced, with captions and audio controls.

Coached and tracked

bars

7

xAPI dashboard

Faster, more accurate, more independent: not just finished.

01

/ 01 · Every reply, live

The rep

Practises the pushback their next call is likely to bring.

Process

Discover · Scope · Build · Prove

Teams outside engineering get pulled into incidents they cannot read. The scenario starts where that hurts: a checkout outage at Relay Nine Commerce, failed requests multiplying, and no obvious first move.

Find the evidence, stop the spread, recover the working version, explain the decision. Guided mode had to translate the technical language so a beginner could still make the call, and the whole thing had to work with captions, resizing and audio controls.

A mission brief, an interactive dashboard, a service topology, voiced coaching, hints, randomised decision cards and repeatable attempts. Every modality earns its place by removing confusion, not by adding polish.

Each attempt emits xAPI-compatible events: score, time, hints, unsupported actions, phase progress, retries. The dashboard answers whether a learner is getting faster, more accurate and more independent, not just whether they finished.

More work

All work

Scaling complex technical systems but your execution is lagging?

Book a consultation

© 2026 HADRIEN LEE /// AI LEARNING CONSULTANT

© 2026 HADRIEN LEE /// AI LEARNING CONSULTANT