Answers, withoutthe hand-waving.

It captures the specialty encounter and finishes it — the specialty-format note, ICD-10/CPT coding with an audit trail, prior authorization, referral, and follow-up — with a human in the loop on every consequential action.

No. Scribara owns its model stack — specialty ASR (Riva), reasoning + verifier (NeMo), and medical imaging (Clara/MONAI) — trained on its own data and served on NVIDIA compute.

Cardiology, dermatology, orthopedics, GI, and ophthalmology at launch, with more as specialty models are trained.

An on-prem path runs the models inside your boundary via NIM on NVIDIA AI Enterprise, so PHI never leaves the building. [ASPIRATIONAL] Cloud uses per-tenant keys and ingress redaction.

HIPAA/BAA today; SOC 2 Type II and ISO 42001 are on the roadmap. [ASPIRATIONAL]

An independent verifier and NeMo Guardrails enforce abstention, and outputs are grounded via retrieval over guidelines and payer rules.

Against the administrative labor it replaces — per provider and per encounter — not per seat. Enterprise is custom with on-prem options.

Yes — a 60-day paid pilot at fixed price that converts to annual.

From multi-tenant cloud to VPC to on-prem NIM to air-gapped Scrivo appliances. [ASPIRATIONAL]