Specialty notes
SOAP notes in cardiology, derm, ortho, GI, and ophthalmology formats — signed within seconds.
Learn more
Ambient AI · specialty medicine
The ambient AI clinicianfor specialty practices.
Scribara listens to the encounter and finishes it — the specialty-format note, ICD-10/CPT coding with an audit trail, the prior auth, the referral, and the follow-up — before the physician leaves the room.
Listening · exam room142 ms
Dr. Rivera: Two-week history of exertional chest pressure…Patient: It eases when I rest, comes back on the stairs.
Signed chart · auto-drafted<300 ms
A/P: Stable angina, optimize medical therapy, stress test ordered.
Coding
I20.9 93000 · evidence-linkedPrior auth assembled · cardiology referral drafted
Verified & ready for clinician signature
Built for the workflows specialty medicine already runs on
Epic FHIRathenahealthNVIDIA InceptionHL7SOC 2HIPAA
01 / Listen
It hears the whole room.
Specialty-tuned, noise-robust speech recognition captures the encounter in real time — overlapping speakers, instruments, and dense clinical jargon.
02 / Reason
It understands the specialty.
A clinical-reasoning model drafts the note in your specialty's format and proposes codes grounded in guidelines and payer rules — with an independent verifier.
03 / Act
It completes the work.
Prior auth assembled and submitted, referral written, follow-up scheduled — every consequential step reviewed by the clinician and logged.
04 / Learn
It sounds more like you.
Every edit becomes labeled data, so notes, codes, and auths increasingly read as the physician would have written them.
What it does
One system, not one feature
Generalist scribes stop at the note. Scribara closes coding, prior auth, referrals, and follow-up — specialty-first.
Coding with audit
ICD-10/CPT/HCC with modifiers, each code linked to note evidence and a payer-rule trail.
Learn morePrior authorization
Evidence assembled to the payer's criteria, submitted, tracked, and appealed on denial.
Learn moreSees the findings
Specialty medical imaging — lesions, retina, X-ray — fused with the spoken encounter.
Learn moreHears the room
Clinician-grade ASR built for noisy, multi-speaker exam rooms.
Learn moreFollow-up & triage
From documentation toward autonomous follow-up and chronic-care management. [ASPIRATIONAL]
Learn moreRiva ASR · streaming142 ms
NeMo reason + verify98 ms
Clara imaging fuse54 ms
TensorRT-LLM serve<300 ms
Deep tech, not a wrapper
Owned clinical models on NVIDIA compute
Scribara owns its model stack instead of renting it. A specialty ASR on NVIDIA Riva, a clinical-reasoning LLM with an independent verifier on NeMo, and medical-imaging models on Clara/MONAI — served under 300 ms with TensorRT-LLM and Triton.
- Riva-based specialty ASR for accurate ambient capture [ASPIRATIONAL]
- NeMo reasoning + verifier with abstention via Guardrails
- Clara/MONAI imaging for derm, ophthalmology, orthopedics [ASPIRATIONAL]
- HIPAA on-prem via NIM on NVIDIA AI Enterprise [ASPIRATIONAL]
Agent workflow
Plan, perceive, reason, verify, act
Scribara runs an agentic loop with a human in the loop on every consequential action — reliability is engineered, not assumed.
Plan
The agent scopes the encounter and the work it implies — note, codes, auth, referral, follow-up.
Perceive
Vocalis transcribes; Visera reads specialty imaging; structured EHR context is retrieved.
Reason
A specialty model drafts outputs grounded in guidelines and payer rules via NeMo Retriever.
Verify
An independent verifier checks every output and abstains or escalates when unsure.
Act
Clinician approves; Scribara writes to the EHR and logs an immutable, reversible trail.
By the numbers
Built to return clinical time
0
Hours returned / week [ASPIRATIONAL]
0
Coding accuracy [ASPIRATIONAL]
0
In-exam inference budget
0
Specialties at launch
The ecosystem
Five products, one platform
One clinical intelligence platform — population health, chronic care, synthetic data, revenue analytics, and trial matching — on one GPU-accelerated backbone.
Nexar
Population health intelligence — RAPIDS analytics + NeMo risk scoring at health-system scale.
ExploreVigil
Real-time chronic care monitoring — Holoscan + Jetson edge AI between specialty visits.
ExploreAugur
Synthetic clinical data — NeMo + Clara/MONAI. [ASPIRATIONAL]
ExploreRevix
Revenue intelligence — deny-risk scoring and undercoding detection before submission.
ExploreOrbis
Clinical trial matching — NeMo Retriever + RAPIDS GPU connects patients to active trials.
Explore
Proof
What clinicians say
Outcomes are reported by design partners and marked illustrative [PLACEHOLDER].
“Charting ends at the office, not at home — the note is signed before I touch the keyboard.”
“It codes the way our best biller does, and shows its work for every line.”
Trust
Enterprise-grade from day one
Healthcare buys on trust. Scribara is built to clear procurement, security review, and compliance.
HIPAA / BAA
PHI handled under a Business Associate Agreement.
SOC 2 Type II
[ASPIRATIONAL] — roadmap to Type II within 12 months.
On-prem NIM
Run owned models inside your datacenter via NVIDIA AI Enterprise. [ASPIRATIONAL]
Immutable audit
Every agent action logged, reversible, tied to the clinician's signature.
Data residency
US, EU, UK regions; per-tenant keys (BYOK option). [ASPIRATIONAL]
ISO 42001
AI management-system conformity on the roadmap. [ASPIRATIONAL]
Answers
Frequently asked
No. Scribara owns its model stack — specialty ASR (Riva), reasoning + verifier (NeMo), and medical imaging (Clara/MONAI) — trained on its own encounter data and served on NVIDIA compute. Frontier APIs are only a bootstrap/fallback.
An on-prem path runs the models inside your boundary via NIM on NVIDIA AI Enterprise, so PHI never leaves the building. [ASPIRATIONAL] Cloud deployments use per-tenant keys and PHI redaction at ingress.
Cardiology, dermatology, orthopedics, GI, and ophthalmology at launch, with more added as specialty models are trained.
The in-exam inference budget is under 300 ms, served with TensorRT-LLM and Triton; a signed note is ready within seconds of the last word.
Give clinicians their evenings back.
See Scribara complete a specialty encounter end to end — note, codes, prior auth, and follow-up — in a 20-minute demo.