Built by clinicians · for clinicians

The scan arrives already read.

OraliqAI segments the anatomy, flags the findings, quantifies them and drafts the report — so the clinician opens a prepared second read, not a blank slice.

CBCT · DICOM/ INTRAORAL STL·PLY/ PANORAMIC/ BITEWING · PA/ FDI notation/ audit hash-chained
The premise

Every scan today lands on a blank screen, and the clinician builds the read from nothing — one slice at a time, under time they rarely have. OraliqAI changes the starting point.

It works the way a trusted second reader would: quietly, before you sit down. The anatomy is already segmented, the findings already flagged and scored, the measurements already taken. What reaches you is a prepared read to confirm or overrule — never a diagnosis handed down.

A modern dental operatory with imaging equipment.
Fits the operatory you already run — no new hardware, no new capture workflow.
Who builds it

Built by clinicians. Reviewed by clinicians. Made for clinicians.

Most dental AI is built by engineers who have never signed a report. OraliqAI is the opposite: the people who read these scans for a living decide what it flags, how it says it, and where it must stay silent.

Model proposes a finding Clinician reviews Ground truth signed off Reviewed release → case no silent updates
Clinician-in-the-loop — every release
Built by
clinicians

Practicing radiologists and dentists define every finding, threshold and report line — not a data team guessing at clinical intent.

Reviewed by
clinicians

Every model change is signed off against ground truth by licensed readers before it can touch a case. No silent updates.

Made for
clinicians

It hands you a prepared second read to confirm or overrule — built to support the clinician's judgment, never to replace it.

Most dental AI
  • Trained by engineers on labels nobody clinical checked
  • Ships a black-box score with no evidence to review
  • Updates silently — you can't tell what changed
  • Positioned to replace the read
OraliqAI
  • Findings defined and validated by practicing clinicians
  • Every flag carries its slices, confidence and model version
  • Clinician-reviewed, versioned releases — nothing changes unseen
  • Built to prepare the read; the clinician always signs it

Reads the scans a practice already captures.

One engine across 3D volumes and surface meshes — nothing changes about how the study is acquired.

CBCT · DICOM

Cone-beam CT

Head-and-neck volumes: implant-site planning, inferior alveolar nerve canal proximity, sinus, TMJ and airway.

IOS · STL / PLY

Intraoral scans

Surface meshes assessed for occlusion, wear, restoration margins and arch metrics.

PANO

Panoramic radiograph

2D reads for caries, periapical lesions and bone-loss patterns across the full arch.

BW · PA

Bitewing & periapical

Intraoral films scored tooth-by-tooth, reported in FDI notation with per-finding confidence.

How a read gets built.

Four stages, each recording model version, input hash and slice-level evidence — an audit trail behind every finding.

Ingest

Validate & de-identify

DICOM and mesh files are checked and de-identified at the door, then graded diagnostic or limited before anything runs.

Detect

Segment & flag

Anatomy is segmented, pathology detected per site, each finding scored for confidence and localized to its slices.

Quantify

Measure & draft

Volumes, distances and asymmetry are measured; a structured, clinician-readable report is drafted from the findings.

Triage

Route & sign

Urgency is triaged and routed to the right queue; the clinician confirms, overrides with a reason, and signs.

A second read shouldn't slow you down. It should be waiting when you arrive.
— The design principle behind OraliqAI

Severity is reserved, never decorative.

Four levels drive routing and alerting — and the same scale colors every finding, overlay and report, by design.

RoutineNo action beyond the standard recall interval.worklist
Follow-upWatch-list finding; schedule a re-check.worklist · recall
UrgentNeeds timely clinician attention.ehr_inbox
CriticalEscalate now — the on-call reader is notified.on_call_radiologist
Oversight & assurance

The model prepares the read. The clinician makes the call.

OraliqAI is assistive clinical decision support — not an autonomous diagnostician. It is built around the boundaries a medical device is held to: review before action, a trail behind every claim, and patient data that never leaves its lane.

Nothing is asserted that a clinician has not seen, and nothing is finalized that a clinician has not signed.

Clinician-gated

Two-step sign-off

No report finalizes and no record writes back without an explicit confirm-then-sign.

Traceable

Immutable audit trail

Every model output and clinician decision is written to an append-only, hash-chained log.

Isolated

Tenant-scoped PHI

Patient data is isolated per practice at the database and storage layer — never commingled.

For licensed clinicians

See the second read before you read the scan.

Sign in to the clinician console, or talk to us about bringing OraliqAI to your practice.