Somnus Explained · Issue Nº 03
How We
Use AI.
Where the intelligence lives, where it doesn’t, and who decides.
Editor's Note: This is part of Somnus Explained, a series describing who we are, what we make, and the principles that guide our work.
People ask us a direct question, so we will answer it directly: what does artificial intelligence do in Somnus, and what does it not do? Every company now says it is AI-powered. Almost none of them will tell you precisely where the AI is, where it is not, and what it is never allowed to decide. We will.
One framing shapes everything that follows: our wellness software operates under the FDA General Wellness Policy, and our software is built to support general wellness and to support the work of clinicians, not to replace them, and not to stand in for the judgment of the people who care for patients.
How we think about where the AI lives
The intelligence in Somnus lives in the software layer. BSA™, our acoustic engine, uses it to parse the night's signals on your phone: patterns in breathing and snoring, the shape of your sleep.
The engine is deterministic. Same night, same numbers: run the identical night through it twice and you get the identical result, with the reasons visible. A deterministic engine is the opposite of a black box, and most AI products cannot make that promise. And the Somnus Index™ is built on your own baseline, never a population average. You, measured against you.
The Somnus Index™ distills that picture into a single wellness measure you can watch move over time, and plain-language nudges help you act on it. If you have worn an Apple Watch, a Whoop, or an Oura ring, you already know this model: sensors feed a wellness layer, and intelligence makes the feed legible.
How we think about the implant
HYPNARA™ is a passive structural palatal implant. It holds tissue. It does not compute. It is not AI-driven, and it does not make decisions. We say this plainly because the distinction matters, and because it is easy to assume otherwise. The intelligence in our system is in the software; the implant is passive hardware, and we describe it that way because that is what it is.
How we think about software and judgment
MORPHEX™AI supports clinical decision-making; it does not diagnose independently. It is not the decision-maker. We are deliberate about this boundary. A clinician interprets, decides, and acts; the software's job is to support that work, not to perform it. We would rather describe that role narrowly and accurately than claim more than the software does. The honest position is that good software narrows the gap between information and judgment. It does not close that gap, and we do not pretend it does.
How we think about the Index
The Somnus Index™ is a general-wellness measure. It is meant to help people and their clinicians think about patterns over time, in plain and general terms. We describe it at that level on purpose. We do not present it as a diagnostic score, and we do not ask it to carry more meaning than a general-wellness measure should.
How we think about accountability
Two rules govern everything above, and we hold them as rules rather than aspirations. First: a human is accountable for everything our AI touches. No output of ours reaches a patient or a provider without a named human owner of its consequences; there is no “the algorithm decided” in this company. Second: we say what the AI is for, in plain language, before you use it, not after, and not only in a policy document.
How we think about saying this now
The register a company uses about AI when it is small is the register it gets held to when it is large. The field is currently full of products that blur the line between wellness insight and medical judgment, on purpose, for marketing reasons. That blurring is precisely what our posture exists to prevent: the wellness layer stays a wellness layer, and the clinical tools go through the front door of clinical regulation. That is not a limitation we are apologizing for. It is the design.
A note on information
If you want to understand how we handle information, our privacy policy and our terms are the governing documents. This installment states our operating principles in plain language; where those documents speak, they govern.
AI is how the loop gets sharp. People are why it exists. We do not intend to confuse the two.
We are glad to talk further about how we think and how we build. Reach us at [email protected].