One machine that looks safe when it isn't.
The CSM machine sits in a genomics environment where the hazards aren't abstract. It carries electrical, pneumatic, and laser-etching energy — often independently. It processes live patient blood samples. And it runs an automated capper whose probe cycles fast enough to cause serious crush injury. A locked main disconnect makes the machine look dead. It isn't.
At the same time, GRAIL's team was moving from an engineering-heavy group toward line operators, across a floor that works in both English and Spanish. Off-the-shelf video couldn't teach hazard recognition on this specific machine — and couldn't do it with equal rigor in two languages.
"Generic training tells people the rules. A simulation of their own machine builds the reflex — before the moment that needs it."
Discovery surfaced four specific gaps:
A locked main disconnect doesn't de-energize the laser etching circuit — and that circuit wasn't on the lockout list. Classroom video never surfaces documentation gaps like that.
A dropped sample tube shattering on the blade assembly is a formal bloodborne-pathogen exposure event — not a cleanup task. The instinct to "just clean it up fast" is exactly the wrong reflex.
People who know the machine deeply carry the highest overconfidence risk; new operators carry the highest inexperience risk. One training had to reach both without diluting either.
A meaningful share of the floor works in Spanish. Translated subtitles aren't parity — the scenario, the feedback, and the risk scoring all have to be native in both languages.
Photo-to-Sim, tuned to the CSM machine.
From GRAIL's facility photos, we generated a complete interactive training experience — not a video, an experience the technician moves through and can fail inside, safely, before they ever fail on the floor.
Every scenario is set on the CSM machine itself — the disconnect, the laser circuit, the blade assembly, the automated capper. Recognition transfers because the environment is theirs, not a stock facility.
A short, role-aware intro that frames why safety matters for the specific person taking it — Engineer, Operator, EHS Leader, or Supervisor — and sets up the recognition behaviors the simulation reinforces. Mobile-ready.
A first-person scenario built on their actual machine, delivered natively in English or Spanish. The trainee sees the exact equipment, PPE, and hazard context a CSM technician encounters — not a generic warehouse.
Three timed decision scenarios drawn from the machine's real failure modes. Wrong answers trigger constructive feedback; every completion feeds a live dashboard that surfaces worker-level risk, most-missed hazards, and response times.
Three scenarios. Built from the machine's real failure modes.
Each check is a real decision point on the CSM — the moment an incident actually starts. The wrong answer is the intuitive one, which is exactly what makes it worth training.
The main disconnect is locked and the machine looks dead — but the laser circuit has its own power and isn't on the lockout list. Stop and lock out the source documentation forgot. The single most-missed hazard in the pilot.
A sample tube shatters on the blade assembly, leaving biological material on the blades. Stop, don't touch, initiate the exposure protocol — the tempting "clean it up quietly" answer endangers the worker and destroys the incident record.
A colleague reaches into the running capper to check the probe. The "safe zone" people assume exists disappears at speed. Call out, hit the emergency stop, then report — watching carefully is not an intervention.
Two languages. Natively built — not subtitled.
The Spanish track isn't a voiceover dropped onto English footage. The scenario video, the decision options, the feedback for every answer, and the risk scoring all run natively in Spanish. A technician on the Spanish track gets a first-class experience with identical rigor — closing the exact gap where translated-subtitle training quietly fails.
The benchmark this pilot is built to reproduce.
The GRAIL pilot runs the same Photo-to-Sim model behind our anchor deployment at Trane Technologies — where the results were measured across a live, cross-functional cohort of new hires and experienced technicians.
Figures above are from the Trane Technologies deployment, shown as the outcome benchmark the GRAIL pilot is designed to reproduce.
Training output becomes an EHS risk signal.
Because every completion feeds the dashboard, a hazard being missed at scale stops being a training score and becomes a live finding an EHS leader can act on — before it becomes an incident.
From pilot to program.
The immediate step is closing out the CSM machine pilot with a full cohort. From there, the same Photo-to-Sim standard extends across additional equipment and sites — every module custom to the environment, native in the languages the floor speaks, and feeding one EHS risk view.
Generic training tells people the rules. Immersive, environment-specific simulation builds the reflex — and turns each cohort into a risk map its safety team can act on.
GRAIL didn't need more safety content. They needed a simulation of the exact machine their team services — in both languages their floor speaks. That specificity is what makes recognition transfer.
- Multi-energy hazard recognition on one machine
- Bloodborne pathogen exposure response
- Eng → operator readiness across experience levels
- Bilingual parity — not just translation
- Comprehension & response-time data, not just completion
- Simulation of the CSM machine from facility photos
- Role-specific tracks for 4 job functions
- Native English + Spanish — video, cards, scoring
- 3 timed scenario decision checks
- EHS dashboard: worker risk + most-missed hazards
- Photo-to-Sim — custom from facility photos
- Web-based, mobile-ready — no install
- Timed decision scoring & hazard logging
- Live EHS analytics dashboard
- Deployable in days, LMS-integration available