Ninety days, one room,
at zero cost.
Gray Scrubs AI is pre-commercial. The detection engine works and what it needs is the clinical judgement of a team that lives this every day. The pilot runs at zero cost to you, and either side can end it with two weeks of notice.
What the system sees, before anyone writes a report about it.
One second of an operating room, built from cameras alone. Who is in the room, where they are standing, and what they are near. Every compliance question a pilot answers is a question about this, repeated for the length of the case.
Our own reconstruction of a public dataset of simulated surgery on a phantom. Held at one second, with the light of the seconds before it. Rendered in your browser.
The ninety days.
Install and co-design
- Cameras and edge device installed in one operating room
- Staff introduction and consent process, run with your team
- Baseline documentation of the protocols you already follow
- Interviews with the surgical team about where auditing hurts
Live monitoring
- Detection running on cases in that room
- Weekly compliance reports to the OR director
- Alert thresholds tuned against your feedback, not our defaults
- Fortnightly co-design calls with the engineering team
Readout and decision
- Detected events reviewed against your own incident record
- Full pilot readout presented to your team
- A written account of what did not work
Decision point: extend, revise, or stop
Our side of it
- The system at no cost for the pilot period
- Edge hardware, installed and maintained
- Weekly compliance reports and a live view in the room
- An engineer assigned to your team
- Co-authorship on any resulting publication
Yours
- One operating room, six to ten cases a week
- A champion: infection control officer or OR director
- Fortnightly feedback, 45 minutes
- De-identified outcome data, shared under the agreement
- Your judgement on where the product falls short
What we do not claim.
Gray-1 produces operational and compliance analytics. It is not a diagnostic device, it makes no clinical determination, and every flag is reviewed by a person. Clinical decisions stay with your surgical team, and the system stays advisory.
We have no outcome data. Nobody has run this on enough real cases to say that detected breaches predict infections at your site, and a pilot is how that question gets answered instead of assumed. The work so far has run against a public dataset of simulated surgery on a phantom, which we say plainly because you would find it out anyway.
Our regulatory posture follows from that. We are not seeking a diagnostic claim, and we will not describe the system as one to get a pilot signed. If a deployment ever needs a regulatory pathway, that gets settled in writing before anything is installed.