why cliro

Not a scribe. Not an EMR.
The structured record neither one gives you.

Everyone is building a better scribe. A scribe writes down what was said — useful, and it leaves the hard part where it was: turning what was said into evidence a clinician can act on.

Cliro works one layer down: it turns each session into structured evidence — every line linked to where it came from — and hands every decision to a clinician.

The note is a by-product. The record is the point.

FIG. A — what a scribe hands back
…and she said the pain’s worse getting up from the chair, been about three weeks, worse in the mornings, tried some paracetamol…
FIG. B — what Cliro hands back
  • obssit-to-stand pain
  • obs3wk onset
  • obsAM pattern
  • medself · paracetamol
FIG. C — what makes it a record
signed · dr. e. marsh
08:41:44 · linked to session 12:41–13:03
published to record · immutable
on scribes

A scribe is genuinely useful — it saves real time in the room.

on emrs

An EMR is genuinely necessary — the record has to live somewhere.

Neither touches the hard part.

four things
  1. 01

    Structure, not transcription

    A scribe hands you back a paragraph of what was said. Cliro hands you back a structured record — each observation named, tied to the moment it happened, ready to reason over. Text is what happened. Structure is what it means.

    …and she said the pain’s worse getting up from the chair, been about three weeks, worse in the mornings, tried some paracetamol …
    record · observations
    sit-to-stand pain3wk onsetAM patternself-med · paracetamol
  2. 02

    Evidence you can check

    Every recommendation carries its reason. Every number traces back to a recorded moment — a citation, a timestamp in session audio, a source document. When the system can’t know, it says so, plainly. Verify before you trust.

    every claim traces to its source · verify before you trust
    observation · citation
    • TUG · 14.2sCited · Podsiadlo 1991
    • gait velocity · 0.82 m/s▶ 14:22 · session audio
    • sample too short (0:41)— · could not be assessed
    connected speech · intelligibility — could not be assessed · sample too short (0:41) — needs ≥2:00 conversational speech
  3. 03

    A clinician gate, in the architecture

    The AI can only propose. Nothing enters the record until a clinician confirms and signs. It isn’t a policy we’re asking you to trust — it’s how the system is wired.

    proposedgatesigned
    wired in, not promised
  4. 04

    One spine, every discipline and language

    Physio, speech, mental health, OT — the same underlying record, richer with every session and result. The words the client used, in the language they used them, alongside any translation.

    one spine
    physio
    speech
    mental health
    OT
    client said · EN · 中文 · हिन्दी
the table
An AI scribe
What it does
Transcribes a session
The evidence
A block of text
The decision
Writes for you
Tomorrow
Another note
An EMR
What it does
Stores what you type
The evidence
Fields
The decision
Holds your typing
Tomorrow
The same fields
Cliro
What it does
Structures the whole journey into signed, source-linked evidence
The evidence
Every line links to the moment it came from; every recommendation cites its research
The decision
Proposes; a clinician confirms and signs — enforced, not promised
Tomorrow
A record that’s richer with every session and result, across disciplines and languages
what each thing actually does
what a record carries over time · sample

We’re not trying to be a better scribe. The scribe writes the note; we’re building the record the note was always just one piece of.

the design

Cliro supports clinicians.
It does not replace them.

That isn’t a limitation we apologise for. It’s the design — and it’s the reason a clinician, a funder, and a patient can all trust the same record.

01 captured

session recorded · evidence extracted

02 draft
AI · proposed

draft findings · nothing entered yet

03 the gate

required · a clinician must act

04 signed

awaiting signature

nothing leaves without a signature

proof · rehab practice · in development

We ran it on one real client’s records (identifying details removed). It proposed the same first assessment the treating clinician chose — and flagged a treatable voice problem that usually goes unreferred. The clinician made every call. We’re building this inside a working rehabilitation practice, with the clinicians who use it.

n=1 · real de-identified client · the clinician made every call