Video · 3:49 · synthetic data
Cath Copilot walkthrough
A day in the cath lab on synthetic data: ask what needs attention, clear pre-case flags with their sources, validate the in-room timeline, reconcile a value two systems disagree on, sign a cited note, send the handover and check the audit log.
Transcript
The full narration, as spoken in the voice-over. Headings mark each scene.
A case in the cardiac catheter lab touches half a dozen systems, and readiness checks, notes, audit returns and handovers are still pieced together by hand. Cath Copilot is a desktop copilot that joins them up, on the lab's own computer.
Your role decides what you see
Staff sign in with their hospital account; here, a demo picker stands in. The role decides what each person can see. This is a cath lab nurse, and the briefing already shows what is on today.
Ask what needs attention
Ask what needs attention, and the list is built from the record. Below it, the local AI adds a short summary, checked against the sources. The model runs on this computer; nothing leaves it.
Rule-based flags, each with its source
Is Bravo ready? Not yet: five open items, each raised by a local rule, not the AI, and each with its source. The nurse takes one, and resolves another with a note, both logged under her name.
A source on every field
The pre-case brief puts the case on one page, with a source on every field. A missing last dose is shown as missing, and any source opens the original record. Restricted fields are marked, not hidden: a nurse may not see the order details.
In the room: validate events as they arrive
In the room, the live timeline turns dim. Events arrive from the physiology system and the stock scanner, and a person validates each one. Haemodynamics are for display only: no alarms, and no interpretation.
Same patient, different role
Switch to the operator, and the same brief now shows the order details the nurse could not see. Same patient, same record. The role decides.
Two sources disagree: a person confirms
After the case comes reconciliation. The injector recorded one hundred and forty-five millilitres of contrast; the nursing record says one hundred and forty. A person confirms one value, with a reason.
Unsupported statement: signing locks
The procedure note is drafted from the record, with a citation on every statement. Change one forty-five to one forty, and it no longer matches its source: it turns red, and signing locks. Put it back, add the findings and the plan, and sign.
Audit return pre-filled, confirmed by a person
The national audit return is pre-filled and cited. A person confirms each field, and accepts or rejects each suggested code. The system never codes on its own.
A cited handover: facts and gaps, no instructions
The handover to the coronary care unit is written from the record: closure time, the anticoagulant with no recorded last dose, and what is still open. Facts and gaps, no instructions.
Every step, by name, in the audit log
The quality lead sees every step in the audit log, by name. In the desktop app, the log is append-only and hash-chained, with no edit and no delete.
Recommends, drafts, routes. Never decides.
Ask for clinical advice, and it declines. Cath Copilot is a prototype on synthetic data. It is not a medical device, and it is not for clinical use. It recommends, drafts and routes; it never decides. And the AI runs on the machine.
Cath Copilot, from Zapx Labs. Every fact cited, and every action confirmed by a person.
Cath Copilot is pilot-stage software. Everything in this video runs on synthetic data. Cath Copilot is a prototype running on synthetic data only. It is not a medical device and is not for clinical use. It recommends, drafts, and routes; it never diagnoses, orders, or decides. A clinical safety and information-governance process precedes any use with real patient data.
See Cath Copilot on synthetic data
Cath Copilot is a pilot. Book a demo and we'll walk through it on synthetic data, then talk about what a pilot under your own governance would need.
Book a demoOr email hi@zapxlabs.com.