Scrivas
Scrivas sells SAI, an ambient clinical scribe built by physicians for practices, specialty clinics and health systems. SAI listens to an in person or telehealth visit, or takes dictation, prepares a draft in the clinician's own template and phrasing, and lets the physician revise it by voice before approving it and transferring it to mapped fields in the record with one click. SAI Assistant precharts from prior visits and, during the encounter, surfaces follow up questions, gaps in the history, differential diagnoses ranked by likelihood and tests or treatments to consider. The coding workflow suggests ICD-10 and CPT codes and an evaluation and management level, each confirmed by the physician before any claim goes out. Named connections include athenahealth, eClinicalWorks and Practice Fusion, with SMART on FHIR and FHIR R4 support, and documentation runs in several languages.
Pediatric emergency physician Fernando Mendoza founded Scrivas in 2012 as a human scribing company, and SAI is trained on more than 1 million visits from those years; practices can still pair it with human scribes for complex encounters. Essentials costs $115 per provider a month and Plus $175, with 15% off for annual billing, and Premium is quoted. Audio is deleted once the note is done, data stays on AWS in the United States, and a business associate agreement comes with every plan.
Capability Axes
An AI Health Index grade measures what a buyer can verify from public sources on the date shown. It is not a rating of how good the product is. A vendor can build an excellent system and grade low on an axis because it publishes nothing an outsider can check. How grades read
Drafting is what the subscription buys. SAI turns a visit or dictation into a structured note in the specialty template the clinician uses, revises it by voice and precharts from prior visits, and SAI Assistant surfaces questions, history gaps, ranked differentials, tests and codes, all of it model output. The patient lists, reminder dashboard and templates around it would not justify the price on their own. Practices can pair SAI with Scrivas's human scribes for complex encounters, but that service is sold separately, and without its models the software has nothing left to do.
The physician approves every note before it transfers to the chart and confirms suggested ICD-10, CPT and evaluation and management codes before any claim is submitted, while SAI Assistant's questions, differentials and test suggestions are offered for the clinician to consider. Templates and phrasing are set per physician, and voice editing lets the clinician fix a draft before sign off.
Scrivas markets its multilingual documentation as coming "without gaps in care", a completeness claim with no figure behind it. No threshold flags a low confidence passage or a language the model handles less well, and nothing is said about correcting a note after it is filed.
Training on more than 1 million visits, drawn from 14 years of clinical documentation by Scrivas's human scribing business, is the basis Scrivas gives for SAI. The assistant reads the current encounter and prior visits when it proposes questions, codes and diagnoses, and specialty and physician templates shape each draft. No model, supplier, architecture or version is named for the scribe, the assistant or the coding suggestions, and nothing tells a practice when they change.
All SAI data runs on AWS in US regions using HIPAA eligible services only, audio is deleted when the note is finished, and patient data never trains models without explicit written consent. The application agreement lets service providers handle content to run the service and names Wix or another processor for payments. No speech or language model provider is named and no subprocessor list is published, so the parties that receive encounter text, and the terms they work under, are not known to a practice.
Fernando Mendoza uses SAI in his own pediatric emergency practice, where the chart transfer workflow was built, and Scrivas reports 1.5 hours saved per day and documentation time cut by 50% to 70%, as physicians reported it; its ROI calculator assumes a 50% reduction. No other practice or health system using the software is named. The figures come with no dated cohort or measurement method, and no chart review checks the accuracy or completeness of SAI's notes.
Audio is captured locally during the encounter solely to generate the note and deleted once the note is finalized. Customers set how long transcripts and notes are kept, can request immediate deletion with written confirmation, and keep access for 90 days after cancellation before permanent deletion. Patient data never trains models without explicit written consent, access follows the minimum necessary principle, and data exchange is limited to the fields needed for note transfer.
The application terms still allow aggregate or de identified data to be used for improvement without saying how it is de identified, and no procedure is given for a harmful suggestion from SAI Assistant.
A business associate agreement is standard for every SAI customer at no added fee, and the security page states HIPAA compliance, minimum necessary access, US hosting and deletion rules for the clinical data SAI handles. The application agreement addresses protected information processed for the service and restricts what users may store as their own content. The agreement itself is provided on request rather than posted, so a practice cannot read its subcontractor terms before talking to Scrivas.
SOC 2 Type II is named, independently audited each year and covering security, availability and confidentiality, with the report shared under a nondisclosure agreement. AES 256 encryption protects data at rest and in transit, services are limited to HIPAA eligible AWS offerings, access is role based and monitored, and customers can request deletion with written confirmation. The report's period and exclusions, and whether it reaches SAI Assistant and the integration path into records, show only once the report is shared.
Scrivas's application terms name the non device clinical decision support exclusion in section 520(o)(1)(E) of the Federal Food, Drug and Cosmetic Act and address professionals able to review the basis of a recommendation independently. SAI Assistant surfaces patient specific differential diagnoses, ranked by likelihood where applicable, and suggests further tests or treatment beside the note, and the physician signs before anything is filed. The ranked differentials do not show how each suggestion was weighted, which is the independent review condition the exclusion depends on.
The physician controls note format, edits every draft and decides what reaches the record, templates are set per specialty or physician, and access is logged with clinic level controls. Multilingual capture is marketed as holding practices to the same standard of care whatever language the patient speaks. No results are published by language, accent, speaker or specialty, and no policy describes how SAI's models are evaluated or monitored for uneven performance.
Physicians can correct a draft before signing it and review suggested codes before a claim is filed. The application agreement disclaims errors and any duty to correct defects, generally caps liability at $100 subject to applicable law, makes the user responsible for professional services and requires indemnification for certain claims arising from generated output. Customers can request removal of data and receive written confirmation. A mistake in a generated note, a faulty code suggestion or an unsuitable patient specific recommendation has no remedy of its own.
Approved notes go to mapped fields in the chart with one click. Scrivas names athenahealth, eClinicalWorks and Practice Fusion connections and supports SMART on FHIR and FHIR R4, with EMR integration included from the Plus plan up while Essentials is aimed at practices with limited integration needs. Mendoza runs the transfer in his own practice. The integrations write notes into the record; whether each named system also feeds prior visit data back for precharting is not stated.
One hosted service with a committed location: all SAI data is hosted on AWS in US regions using HIPAA eligible services, and no data is processed or stored outside the United States. Audio is captured locally during the encounter and deleted when the note is complete, and notes move to the practice's record only after approval. No choice of region, private deployment or description of isolation between practices on the shared service is offered.
Essentials is $115 per provider a month on monthly billing, with precharting, note creation, voice editing, basic templates and a patient list dashboard. Plus is $175 and adds two concurrent SAI Assistant sessions, EMR integration, fully customized templates, analytics and implementation support, and Premium is scoped to the practice with five concurrent assistants and dedicated support. Annual billing saves 15%, showing $98 and $149. There are no long term contracts or usage fees and the first five notes are free, but every plan starts with a contact form rather than a self serve signup.
Individual practices, growing groups and health systems are the buyers. SAI handles in person and telehealth visits across emergency medicine, primary care, surgical specialties, oncology, cardiology and pediatrics, with note format and phrasing set per physician, and multilingual capture aimed at primary care, pediatric and community health practices serving patients who do not speak English. Mendoza uses it in his own pediatric emergency work. No results are broken down by specialty, care setting or encounter language.
Alternatives to Scrivas
Other products that do the same kind of work as Scrivas, and what each one does.
ScribeRyte AI is the ambient scribe of Nivaran, the new name for ScribeEMR and CodeEMR, a Woburn, Massachusetts company that also sells live remote scribes, medical coding and billing.
CureMD sells a specialty EHR, practice management and billing platform to physician practices and now presents it as an agentic intelligence platform, with AI across documentation, phones and coding.
Sunoh.ai is the ambient AI scribe from healow, an eClinicalWorks company, and the documentation option built into the eClinicalWorks record system.
Tali AI is a Canadian AI platform for primary care built around an ambient scribe.
ClaireMed makes Claire, a Toronto built AI tool that interviews patients by phone before their appointment and hands the physician a structured history and a draft note.
Augmedix sells ambient medical documentation to health systems, hospitals and clinics, with AI drafting the note and trained medical documentation specialists adding as much human support as a customer buys.
Pricing
Vendor-published figures are labeled as such. Figures labeled “Estimated” are derived from third-party sources and have not been confirmed by the vendor.
| Entry Price | Pricing Basis | BAA Tier | Implementation | Source |
|---|---|---|---|---|
|
$115 per provider per month
$115 baseline
|
Essentials is $115 per provider per month with monthly billing, and Plus is $175 per provider per month. Annual billing saves 15%, which shows as $98 and $149 per provider per month. Premium is priced for each practice. There are no long term contracts or usage fees, and each plan starts through a contact form. | A business associate agreement is included at every tier without a separate fee. | Plus includes implementation and support. Premium adds dedicated onboarding, account management and ongoing adjustments. | Vendor Published |
Plus adds an electronic record integration option, customized templates and analytics. Premium adds capacity, analytics and dedicated support. The first five notes can be tried without charge through a demonstration.