Novaid
Every second counts, every step guided. A hands-free augmented reality system that turns an untrained bystander into a guided first responder, from a design-thinking sprint and a working HoloLens prototype to an AI-powered HUD that assesses the patient and overlays step-by-step 3D instructions.
- What we did
- Design thinking facilitation, interface design, prototype development, simulation testing
- Tech
- Unity · Microsoft HoloLens · spatial AI · facial recognition · dynamic 3D overlays
- Methods
- Personas, surveys, interviews, field observation, usability testing, high-stress simulation
- Arc
- Research prototype → AI-powered product
- Outcome
- Untrained bystanders guided through first aid, hands-free

The gap before the ambulance
In medical emergencies the first minutes are critical, and bystanders are usually first on the scene without the knowledge or confidence to act. First aid fails not because instructions don't exist, but because in panic people can't recall them, can't judge if they're doing it right, and can't hold a phone while both hands do compressions.
The design space was framed precisely: support the untrained bystander while waiting for the ambulance. Not replacing emergency services, bridging the gap. Hands-free, or it doesn't count.
Divergent, convergent, repeat
Surveys and interviews established how people actually behave around emergencies. Personas kept the untrained bystander, not the paramedic, at the center. Field observation grounded the interaction in real procedure, including the dispatch side of the EMS gap.


Proving it on HoloLens
The first generation was functional AR in Unity on HoloLens, not a video mock-up, so usability tests measured the real thing: legibility during physical activity, timing of instructions, whether guidance reduced hesitation. It did. Participants moved through the procedure with the overlay pacing them instead of freezing at the first step.
Both hands belong to the patient. The interface had to live somewhere else, so we put it in the air.
concept rationale
Then the system learned to see
The validated concept raised the next question: what if the system could also assess the patient? Novaid HUD uses spatial AI and facial recognition to read the patient's state, then overlays dynamic 3D instructions that adapt to what it sees. Tested in high-stress simulations, two rules drove refinement: never obscure the patient, and recognize the situation instantly. Guidance with latency is guidance that arrives too late.

Panic, turned into procedure
Ordinary bystanders empowered with real-time guidance, panic turned into structured action. It's the longest research arc in the studio's work and the clearest proof that emerging tech is only as good as the research underneath it.
