The decision support that stays out of the way.
Auracare CDSS is our clinical product: a clinical decision support system that works during the appointment, not after it. The twin you may already know walks into the clinic with the patient, and the clinician keeps their attention exactly where it belongs.
Clinicians already want AI. Today’s tools don’t fit the room.
The appetite is settled; the fit isn’t. The tools on offer weren’t designed for the consultation, and it shows in three ways.
The demand is already there
Clinicians want the help. What breaks is the moment they turn to a screen mid-consultation: the patient watches their doctor google, and trust erodes.
Generic by architecture
The general-purpose models these tools run on score around a third on specialty-level clinical benchmarks, with no personalisation to the patient actually in the room.
Disconnected from daily life
The record is a series of snapshots with long silences in between. Lifestyle and social history are slow to gather and easy to misremember, so the picture is always partial.
Everything arrives automatically, so nothing interrupts.
The CDSS is built around a single rule: every time a clinician has to break away to search, type or look something up, the consultation suffers. So it removes those moments.
-
The context is already in the room
The patient arrives with a complete lifestyle summary from their twin, and acute vitals stream live from our own devices, so nothing has to be dug up on the spot.
-
Nobody stops to type
The consultation is transcribed as it happens. No one breaks the conversation to write notes, and no one reads from a keyboard.
-
The clinician stays with the patient
The CDSS reasons in the background and surfaces what’s useful. The clinician never appears to search anything, so eye contact, and the relationship, holds.
A closed link, with no middleman.
Acute readings don’t get typed in or pulled from someone else’s integration. Our own devices feed the reasoning core directly: one closed hardware-to-software link, so the data the core sees is exactly the data the device captured.
Recording stethoscope
Captures heart and lung sounds as data the core can read, not just for the ear.
Blood-pressure monitor
Clinical-grade readings, streamed straight into the reasoning core.
Otoscope
A closer look inside, captured and passed on to the core automatically.
And more devices are on the way, the same closed hardware-to-software link, extended to new measurements.
How it reasons.
Most clinical AI sends a prompt to a general-purpose model and returns its answer; that is the 34% in the gap above. Here, every signal is encoded onto the clinical ontology first, a neuro-symbolic core reasons over the knowledge graph, and what comes back is ranked, sourced and auditable.
- Signals in The twin’s shared history, live vitals from our devices, and the conversation itself
- Encoder Each observation entity-linked to SNOMED CT and timestamped
- Knowledge graph The neuro-symbolic core reasons over 532,000 linked clinical concepts
- Traceable output Ranked differentials with sources attached; the clinician decides
Value-of-information loop: before it concludes, the core asks the single question, exam or test that resolves the most uncertainty for its cost, then reasons again.
The knowledge graph is live and explorable today; the reasoning engine that acts on it is in active development.
See the technology in fullThree outputs. One decision, and it’s the clinician’s.
The core doesn’t hand down an answer. It lays out what it has reasoned toward, ranked and traceable, and leaves the judgement where it has to stay: with the person in the room.
Ranked differentials
Possible diagnoses ordered by likelihood, laid out for the clinician to weigh. The call is always theirs to make.
Next steps, tailored
Guideline-aligned options that respect what’s available where the clinician works, including that practice’s own referral rules.
Documentation, automatic
Structured, formatted consultation notes written by the core, so admin shrinks and screen time turns back into patient time.
The figure beside each is clinician agreement, not model certainty: how often clinicians agree with that placement. The figures here are illustrative; the core is in development, and we will measure and publish the real rate from its first studies.
- Viral upper-respiratory infection 61%
- Acute bacterial sinusitis 22%
- Allergic rhinitis 11%
The core never reports a probability of being right. It lays out what it has reasoned toward, ranked and traceable; the clinician selects, and that selection shapes the next steps and the notes.
Auracare is our clinical decision support system, still in development. Its regulatory pathway is not yet confirmed and is under active, continuous review.
Building the clinical side? Let’s talk.
Clinical partners, health systems and investors: we’d like to hear from you as the CDSS moves toward its first trials.