CASE STUDY · INDUSTRIAL MANUFACTURING · TEACH / BUILD
Crisis Simulation Training.
Leaders at a Fortune 500 industrial manufacturer face crises no slide deck prepares them for (a supply-chain disruption, a workplace-safety incident) where the cost of a bad first hour is enormous. We built an AI crisis-simulation pilot: a chat-based simulator with an AI crisis-master that drives a developing situation and adapts to the team's decisions, with roles assigned across participants and the whole thing structured as a recurring program.
The craft was tuning to the right signal: a crisis-master that escalates believably without railroading, pedagogically deliberate yet genuinely uncertain in the moment.
The problem
You can't read your way to good judgment under pressure.
Traditional crisis training is infrequent, scripted, and passive: you read the playbook in a room and hope it sticks when the plant floor is on fire. But a real crisis doesn't follow the deck. It develops, it surprises, and the cost of a bad first hour compounds.
The manufacturer wanted training that put leaders inside a developing situation and made them decide under pressure, repeatedly and at low cost, so the reflexes are built before the day they're needed.
WHY THE OLD WAY FAILS
The approach
Put leaders in the decision, not in front of it.
We designed a chat-based simulator built around an AI crisis-master: it drives a scenario forward, assigns roles across the team, and adapts the situation to the choices participants actually make. Instead of a fixed script with a known ending, every run is a live exercise that bends to the room.
We scoped it as a true pilot, a working technology preview with the dev cost credited toward the full program, so the manufacturer could evaluate it on real sessions before committing further. The whole thing was structured as a recurring program rather than a one-off event, so the learning compounds session over session.
- AI crisis-master drives and escalates the scenario live
- Roles assigned across the team, so decisions are shared
- Adapts to choices instead of following a fixed script
- Recurring multi-session program, so learning compounds
What we built
A working pilot, two scenarios, one recurring program.
Not a deck about training. A simulator leaders can sit down and run, with two leadership scenarios shipped in the pilot.
AI crisis-master
Drives the situation and reacts to what the team decides, introducing complications, pressure, and consequences so the exercise stays live rather than playing out a fixed script.
Two scenarios
A supply-chain disruption and a workplace-safety incident, each designed to be pedagogically deliberate while feeling genuinely uncertain in the moment.
Recurring program
Structured as a series of sessions, not a single event, with roles assigned across the team. Reps accumulate and the learning compounds over time.
THE ENGAGEMENT MODEL
A real pilot, not a sales demo.
We scoped a working technology preview (shippable, runnable on real sessions) with the development cost credited toward the full program. That structure lets the buyer evaluate the thing itself, on their own crises, before committing to the larger build. Lower risk to start, and the pilot dollars aren't thrown away.
What made it hard
Unpredictable enough to be hard, realistic enough to be believed.
A simulation is only useful if it holds two things at once. Too scripted and it railroads participants down a known path. They stop deciding and start guessing what the trainer wants. Too loose and it stops being believable, and the room checks out.
The real work was tuning the crisis-master so it escalates a scenario in a way that genuinely tests leadership judgment: reacting to decisions to keep the exercise live, while staying pedagogically deliberate underneath. Designing scenarios that are pedagogically deliberate yet feel genuinely uncertain in the moment was the craft. It came down to knowing which knob to turn so the pressure lands without the seams showing.
Outcome
Leaders in the decision, repeatably, at a fraction of the cost.
Directional outcomes, by agreement. But the shape is clear: a working crisis-simulation pilot delivered with two leadership scenarios and a multi-session program structure, scenario-based training that puts leaders in the decision at a fraction of the cost of live exercises.
OUT-01
Working pilot
A runnable technology preview, not a deck. Evaluable on real sessions.
OUT-02
2 scenarios
Supply-chain disruption and workplace-safety incident, both leadership-grade.
OUT-03
Multi-session
Recurring program structure so the reps accumulate and the learning compounds.
NEXT
Need your people sharp before the day it counts?
We build training that puts people inside the decision: repeatable, adaptive, and structured to compound. Names and figures are withheld by agreement; we're happy to talk through any of this in more detail under NDA.