Healthcare — on the Lab bench, not in the market
AI is already in
your clinical workflows.
Governed. Evidenced. Safe.
We publish the analysis because the questions are already live for the people working in it. We sell nothing here yet — and we say so plainly.
Why this sector
Chosen by consequence: here, what breaks is patient harm.
Clinical decision support is where AI risk in healthcare concentrates. Triage tools, risk stratification models, early-warning scores and diagnostic support systems sit directly in the pathway between a patient and a clinical judgement. When they are wrong — or right on the population they were trained on and wrong on yours — the consequence is not a compliance finding. It is patient harm. That is why healthcare is on our list — and not yet a market we serve.
This page is analysis, not a service. If that changes, this page is where it shows first.
Where this sector sits
On the Lab bench — working the governance questions through.
No separate rulebook — and that is the problem
Healthcare providers do not get a dedicated AI rulebook. The Care Quality Commission assesses whether services are safe, effective and well-led, and treats technology deployed in care delivery as the provider’s responsibility, not the vendor’s. If a tool is shaping triage decisions or drafting clinical correspondence, an inspection team is entitled to ask who approved it, how it is monitored, and what happens when it is wrong. “We hadn’t catalogued it” is not an answer a well-led provider can give.
What we publish here
Analysis, in the open, before anything is for sale.
The Lab bench
We are working the governance questions through before taking this sector anywhere — publishing as we go, because they are already live for clinicians and boards. When it moves out of the Lab, this is the ground it will stand on.
Visit the Lab
Oversight, designed as a discipline
Every agent we ship is inspectable and stoppable, with a named human accountable for what leaves the building. In a clinical setting, that is the line between decision support and decision-making.
See how we work
Frameworks exist. The work is the mapping.
ISO 42001 gives a management-system spine; the NIST AI RMF a structure for risk; the EU AI Act treats AI safety components of medical devices as high-risk. The work is mapping a real estate against them.
See Citadel
What discovery looks like, elsewhere
Shadow AI in clinical settings looks like a junior doctor pasting a discharge summary into a consumer chatbot after a long shift. The starting point is discovery — what Sentinel does in the sectors we do serve.
See Sentinel
What we won’t claim
No services sold here — yet. Stated plainly, up front.
- We take no healthcare clients today. This page is Lab analysis, nothing more — if anything here implies otherwise, treat it as an error and tell us.
- We claim no clinical expertise, and will not sell governance advice into a clinical pathway we have not worked in.
- We make no claims about any product’s fitness in a clinical setting: nothing we build has been validated on a patient population.
- We do not know when this sector moves out of the Lab, and we will not pretend to a date.
Bring us the question you can’t get answered.
Thirty minutes with a founder. Tell us what breaks in your sector if it goes wrong and we will tell you where the evidence has to come from — and whether we are the right people to gather it.
Price agreed before we start · Founder-led · NDA available