Built so you stay in control

How we keep every AI draft safe for you to review

Nothing reaches a patient without you. The AI prepares a draft, flags anything that needs a second look, and waits for you to read, edit, and approve it.

Talk to our team See how it stays safe ↓

What happens with every AI output

1
The AI drafts
A note, summary, or message — never final.
2
It flags what to check
Risks and gaps are surfaced for your judgment.
3
You decide
You read, edit, and approve. Nothing moves without you.

You are always the final check.

Your safeguards

Three things that are always true

Whatever the AI produces, these never change. No setup, no fine print.

Nothing reaches a patient on its own

Every output is a draft. You read it, edit it, and decide — the system never finalizes a note, message, or order by itself.

It flags what needs a second look

Possible risks and escalation cues are surfaced for you to judge. It suggests; you decide. No automatic triage or disposition.

Drafts follow a checked structure

Notes and summaries are generated against a fixed format and checked for completeness before they ever reach your screen.

Want the engineering detail — architecture, routing, and the checks behind these safeguards? Open the technical detail →

How it stays safe for you

Nothing reaches a patient on its own

Every output is a draft. You read it, edit it, and decide — the system never finalizes a note, message, or order by itself.

It flags what needs a second look

Possible risks and escalation cues are surfaced for you to judge. It suggests; you decide. No automatic triage or disposition.

Drafts follow a checked structure

Notes and summaries are generated against a fixed format and checked for completeness before they ever reach your screen.

How the system is built

System layers

Experience layer

Prototype

Mixed demo status across Dr. Vita Scribe, ER Admin OS, Clinical Copilot review, and supporting specialty demos. See per-surface labels.

Workflow services

Prototype

Session orchestration, prototype safety review routing, and live-demo schema checks.

Workflow router

Prototype

Routes tasks to review-support vs structured generation paths.

Clinical services

Requires integration

Draft insight, note, customer-workflow artifact, and billing-cue services.

Telemetry + QA

Planned

Review events, edit history, and audit trail capture.

Request-to-draft flow

1. Capture

Live

Voice/text intake or clinician context.

2. Normalize

Live

Schema checks and basic safety guards.

3. Route

Prototype

Draft workflow routing for task type.

4. Generate

Live

Demo draft outputs with source/context notes where available; not clinically validated.

5. Review logging

Planned

Edit and acceptance logging planned; clinician review remains required.

What's live vs requires integration

Model and prompt configuration

Requires integration

Current demos label configured structured-draft and review-support paths without promising a fixed production model.

Any named model, prompt set, latency target, evaluation threshold, or fallback policy must be selected and validated during deployment review before production use.

Draft workflow routing

Prototype

Configured routing between structured-output and review-support workflow paths in the demo.

Deployment-selected model path, latency, and evaluation criteria must be locked through customer privacy and security review before production use.

EHR integration

Requires integration

Pilot scoping can start with browser review, copy/paste handoff, or integration discovery.

Any SMART on FHIR reads, vendor APIs, identity mapping, audit controls, and contracts depend on the customer EHR and are not live by default.

Epic FHIR R4 / SMART on FHIR integration

Requires integration

Pilot scoping can identify Epic FHIR R4 resources, SMART launch context, read scopes, and browser-only fallback paths.

Not live by default; requires Epic and customer authorization, FHIR endpoint access, identity mapping, audit controls, agreements, and deployment-specific privacy/security review.

FHIR sandbox discovery, vendor-specific access required

Planned

Sandbox-only FHIR discovery can support integration planning; production EHR, Epic FHIR R4, or SMART on FHIR access is not live by default.

Requires vendor access, customer authorization, contracts, identity mapping, and deployment-specific audit review.

EHR writeback

Requires integration

Current demos can prepare review-ready drafts and handoff artifacts.

Writing notes, messages, orders, or billing actions back into an EHR is not active by default and requires approved connectors, audit controls, and human authorization.

Vercel AI Gateway Realtime voice demo path

Live

Browser-based voice capture is available in current demo paths for supervised workflows.

Production use requires deployment-specific consent, retention, device, network, and privacy review.

SOAP generation public demo

Live

Demo paths can generate structured SOAP drafts and related packets for clinician review.

SOAP output is a draft artifact only and must be reviewed before any clinical use.

Voice workflows

Prototype

Dr. Vita Scribe and voice intake flows produce draft artifacts for clinician review.

Ambient listening, telephony, multilingual support, and clinical operations rollout require pilot verification before production use.

Canada / PHIPA-oriented deployment review planning

Requires integration

Pilot intake can scope applicable PHIPA and provincial privacy role mapping for health information custodian, agent, electronic service provider, or HINP roles only where customer counsel determines those roles apply, plus Canadian hosting/data residency, retention, access, audit logging, subprocessor, consent/disclosure, privacy impact, and security assessment inputs.

Readiness materials support customer-specific legal, privacy, security, and operational review only; they are not certification, legal advice, or a determination that a deployment satisfies PHIPA obligations.

Patient-language draft communications

Requires integration

Demo concepts can show patient-language summaries and follow-up prompts after clinician review.

Patient-facing communication requires approved workflow rules, clinician-reviewed source notes, and deployment-specific consent/disclosure review.

Review logging and acceptance analytics

Planned

The architecture describes edit deltas, acceptance rates, latency, and safety-trigger logging.

Review logging remains planned until customer workflow policy, audit retention, and reviewer identity mapping are configured.

Reference depth

Router + artifacts

Structured extraction

Live

Convert transcripts into schema-checked packets.

Review-support prompts

Prototype

Synthesize review prompts and draft assessment context.

Patient communication

Requires integration

Translate reviewed clinician notes into patient-language drafts.

Safety review routing

Prototype

Surface possible escalation cues for clinician review; no autonomous triage or disposition decision.

Service harness

InsightEngine

Prototype

Evidence-linked draft insights in the internal demo harness

MedicationReconciliation

Prototype

Medication reconciliation review prompts in the internal harness

NoteGenerator

Live

Structured SOAP draft generation in internal harness samples with schema section checks

PreVisitInterviewer

Live

Demo intake flow for structured capture and clinician review

Specialty coverage

Cardiology

Prototype

Pulmonology

Prototype

Emergency Medicine

Prototype

Primary Care

Prototype

Endocrinology

Prototype

Nephrology

Prototype

Psychiatry

Prototype

Oncology

Prototype

Want to see the workflows in action? Start with the labeled demos, then come back here for architecture and readiness details.

What's ready

What you can use today, and what needs setup

A plain look at where each capability stands. Anything tied to your hospital's systems needs setup and review first. Tap a row for the detail.

LiveImplemented in a current demo path without implying production readiness or clinical validation.

▾

Integration boundary: Readiness materials support customer-specific legal, privacy, security, and operational review only; they are not certification, legal advice, or a determination that a deployment satisfies PHIPA obligations.

Requires integration

Patient-language draft communications

Demo concepts can show patient-language summaries and follow-up prompts after clinician review.

▾

Integration boundary: Patient-facing communication requires approved workflow rules, clinician-reviewed source notes, and deployment-specific consent/disclosure review.

Planned

Review logging and acceptance analytics

The architecture describes edit deltas, acceptance rates, latency, and safety-trigger logging.

▾

Integration boundary: Review logging remains planned until customer workflow policy, audit retention, and reviewer identity mapping are configured.

For your technical team

The engineering behind the safeguards

Optional, deeper detail: how the system is built, the request-to-draft flow, routing and artifacts, the internal service harness, and our engineering checks. None of this is clinical validation.

Want to see it in your workflow?

We'll show you exactly how drafts are prepared, flagged, and handed to you for review — on your terms.

Talk to our team

Talk to our team Explore product demos

You always have the final say