A bridge that catches problems sooner — not an emergency service.
An AI calling high-risk patients has to be honest about what it is and what it isn't. It's a calm check-in between visits that listens for early warning signs and gets the urgent ones to a person on your team. It does not diagnose, prescribe, replace 911, or stand in for a clinician — and we built it to behave that way, not just to say so.
- HIPAA-compliant
- BAA-ready
- CPT 99490-native
- Transparent AI
I'm not feeling right at all — my chest is really hurting and I can't catch my breath.
That sounds serious, and I want you to be safe. Please hang up and call 911, or have someone with you call right now.
It catches the slow slide, before it becomes a crisis.
Most decline doesn't happen all at once. It builds quietly at home between appointments — a missed dose, a symptom creeping up — until it lands someone in the hospital. A regular check-in call is a chance to notice early and get a person involved while there's still time to act.
Where it sits in care
Think of it as an extra set of ears between visits — not a replacement for any of the care a patient already has.
- It isA between-visit check-in that listens for early warning signs and routes the urgent ones to your team.
- It is notAn emergency line, a diagnosis, a prescription, or a substitute for a clinician.
Patients keep their doctor, their pharmacy, and 911. This adds a regular call in between — nothing it touches is taken away.
The four lines it will not cross.
Being useful and being safe both depend on the AI staying in its lane. These are the limits we hold it to — on every call, for every patient.
It does not diagnose
The AI notes what it heard and how concerning it sounds. It never names a condition, never decides what's wrong, and never tells a patient what their symptoms mean. A clinician does that.
It does not prescribe
It won't start, stop, or change a medication, and it won't give medical advice. If a patient asks, it points them back to their care team — it doesn't answer for them.
It does not replace 911
It is a between-visit check-in, not an emergency line. If something on a call sounds like a crisis, the AI tells the patient to call 911 — it doesn't try to handle it.
It does not replace a clinician
It extends the reach of your care team by listening at scale and surfacing the few who need a person. The judgment, and the care, stay with your clinicians.
It extends the reach of your care team — it never replaces a clinician, diagnoses, or prescribes.
If something sounds serious, the AI points the patient to 911.
An emergency is the one moment a check-in call must not try to manage on its own. If a patient says something on a call that sounds like a crisis, the AI doesn't keep going with its questions. It clearly tells them to hang up and call 911 — or to have someone with them call — and it pages a clinician on your team. The fastest help is emergency services, and the AI sends them straight there.
- Directs to 911 firstEmergency help is faster than anything the AI could do. So it points the patient there plainly, without delay.
- Pages your clinician tooAt the same moment, a person on your team is alerted — so the care team knows, not just the patient.
- Doesn't play doctorIt won't try to assess, talk a patient down, or decide whether it's 'really' an emergency. That call belongs to 911 and a clinician.
The AI surfaces a concern. A clinician decides.
The point of an alert is to get a person involved — not to act in their place. The AI's job ends where judgment begins: it puts a clear picture in front of a clinician on your team, and that person makes the call.
The AI listens and flags
On a normal check-in call, it pays attention to how a patient is doing and turns what it heard into a clear, reviewable signal — a short summary and a sense of how urgent it is.
Urgent goes to a person, now
When something can't wait, the platform pages a clinician on your team in real time — while the concern is fresh. The routine is logged and trended instead of crowding the queue.
A person decides
The AI never makes a care decision. It hands a clear picture to a clinician on your side, and that person decides what it means and what to do — every time.
In short: urgent pages a clinician now; worth-a-look goes to a review queue; routine is logged and trended. A human is the one who decides what any of it means — the AI never closes the loop by itself.
Patients are told they're talking to an AI.
No tricks, no pretending to be a nurse. On the first call, the AI says plainly that it's an automated assistant calling on behalf of the patient's care team. Trust starts with people knowing what they're talking to.
“Hi, it's the Family Guardian AI assistant calling on behalf of your care team. Is now an okay time for a quick check-in?”
- Says it's an AI, up frontThe patient knows it's automated from the first call — not partway through, and not never.
- Names whose team it's forIt's calling on behalf of the patient's own care team, so the call has a clear reason and source.
- Asks if now is a good timeIt's a check-in, not a demand. The patient can say it's not a good moment — and that's respected.
Wording is illustrative. The opening line is plain by design — older patients shouldn't have to guess who, or what, is calling.
No clinical content reaches a patient until a physician on your side signs it off.
No workflow runs with real patients until a supervising physician signs it off — enforced in the database.
This is the most important safety control we have, so we're plain about where we are today: the workflows in the product are drafts built from published guidance. They are scaffolding for a physician to review — not content any physician has approved yet. We won't claim otherwise, and the system itself won't run an unsigned workflow on a real patient.
A hard stop, enforced
Until a supervising physician signs a workflow off, the system blocks it from being assigned to a real patient. The block lives in the database — not in a policy someone has to remember.
Your physician owns the content
A clinician on your side reviews the draft, edits it to your standard, and signs it off. The clinician layer is a seam your medical staff fills — partner medical staff may be that supervising layer.
On the record
Which physician signed which version, and when, is logged alongside the workflow — so the clinical content behind any call can always be traced.
See the safety behavior for yourself. In the demo.
In a twenty-minute walkthrough on your panel, we'll show how a call sounds, what happens when something serious comes up, how an urgent concern reaches a person on your team, and where the sign-off gate stops unsigned content cold — the real product, not slides.
A real person replies — usually the same business day. No sales sequence, no obligation.