For the people who
keep them alive.
EMTs, paramedics, combat medics, flight nurses. The job is judgment under noise — in a dark room, on a roadside, at a casualty collection point. We’re building a machine that runs the protocol so you can run the patient.
Medical concept of operations
You don’t need a chatbot. You need a partner that knows the protocol.
The medic we’re designing for is on a porch at 2am, in a Black Hawk at altitude, or under a tarp at a collection point with three patients and one set of hands. You’re running TCCC, NAEMT, or your local ALS protocol in your head — while you triage, while you talk to family, while you call the trauma room.
The cognitive cost is huge. Drugs, doses, contraindications, pediatric weight calculations, OLMC permissions. The protocol is the answer, but the protocol is also the bottleneck.
Vertex builds for the moment your hands are on the patient and the right next step is six sub-decisions deep in your training.
- Sepsis recognition delay. The 1-hour bundle assumes you knew it was sepsis at minute zero. You usually don’t.
- Pediatric dose math. Weight estimation, mg/kg conversion, and rate calc — under stress, with a screaming room.
- Multi-casualty triage on scene. START, SALT, JumpSTART — the algorithm is simple. Holding it across nine patients while one is decompensating is not.
- Drug-drug and allergy contraindications. A patient on warfarin, on a beta-blocker, allergic to morphine. You’re going to remember. You shouldn’t have to.
- Hemorrhage control in TCCC. The first two minutes of MARCH decide outcome. Time spent on the wrong limb is time the right one keeps bleeding.
- Documentation lag. The ePCR you write three hours later from memory is the legal record of care. Memory degrades.
Vertex is designing a medic-mode policy around published clinical doctrine—the same source-linked approach used by the ICU scenarios in Colosseum. The target system tracks protocol state; the clinician retains the patient and every clinical decision.
medic_call overrides any time.Every protocol step the policy proposes cites a real, published clinical source. Same schema as our reconnaissance rules — the only thing that changes is the citation file.
This is the trace schema we are designing toward—not a deployment record or clinical recommendation. Values below illustrate how patient state, proposed action, and a source citation would remain together for review.
We intend to build with medics through protocol review, sim-lab integration, and approved ride-alongs. Those relationships are invitations to co-design, not evidence of a current clinical deployment.
If your agency runs high-acuity calls and wants a longer conversation about field-medicine co-design, tell us.
Run the protocol with us.
Sim-lab integration, ride-alongs, ePCR co-design, and the long-term model-card collaboration on prehospital decisions.