Ambient Scribes
M

Mira Health

Mira Health sells AI built for orthopedic practices and surgery centers, from the first phone call to the paid claim. Mira Scribe listens to the visit on a phone or desktop and drafts a surgeon style note with laterality, range of motion, special tests and imaging findings, plus operative reports with implants, approach and modifiers. It suggests CPT bundling, modifiers, E/M levels and ICD-10 codes, queues spoken orders such as an MRI, and sends notes, diagnoses, orders and codes into Epic, athenahealth, ModMed, eClinicalWorks, NextGen or DrChrono, using templates for hand, spine, sports, trauma, total joints and pediatric orthopedics. Mira Concierge answers calls day and night in more than ten languages, books and reschedules in the EHR, checks eligibility, asks orthopedic triage questions, books urgent fracture slots and warm transfers urgent and post operative calls to staff with a spoken summary. Mira Operations handles intake, fax and paperwork, Mira Dictate covers addenda and letters, Mira RCM codes and bills orthopedic claims, and Mira also offers its own web EMR and patient portal.

The scribe runs on Deepgram speech recognition, Anthropic's Claude through AWS Bedrock and Google Gemini, hosted on AWS in the United States. Mira reports more than 10,000 encounters a month across more than 10 orthopedic clinics, and customers include Reno Orthopedic Center, Orthopaedic Specialty Institute, Ventura Orthopedics, Orthopaedic Associates of Maine and Baton Rouge Orthopaedic Clinic, where sports surgeon Matthew Fury writes all his operative notes in Mira. Pavan Pinnamaneni, Rohan Shah and orthopedic surgeon Sri Pinnamaneni founded the company, Mira Technologies, in New York, with orthopedic and spine surgeons from institutions such as the Hospital for Special Surgery among its clinical advisors.

AI Health Index verifiedNovember 20, 2026
Compare Mira Health with other vendors
Founded
—
Headquarters
New York, New York
Categories
ambient-scribes, patient-facing-voice-agents, rcm-and-prior-auth
Indexed Products
Mira Scribe, Mira Concierge, Mira Dictate, Mira RCM, Mira Operations, Mira EMR
Buyer Segments
Orthopedic practices, Orthopedic surgeons, Ambulatory surgery centers, Spine and neurosurgery practices, Orthopedic billing teams
Assessment

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

AI Capability
BB on AI CentralityThe model is the engine of a core module. The platform carries other value, but this capability does not exist without it.
Vendor Published

Models do the work in every product Mira sells on its main site. The scribe turns encounter audio into an orthopedic note through Deepgram speech recognition and Claude, Concierge holds spoken calls with patients, Operations uses Gemini to pull clinical information from faxes and attachments, and RCM suggests codes and modifiers from the note.

Mira EMR, a web record and patient portal on a separate site, is the exception: scheduling, transitions of care, a FHIR API and data export that run without the models. It widens what Mira sells beyond the AI products, so the models drive the products a practice comes to Mira for rather than everything it can buy.

BB on Autonomy and Oversight ModelThe oversight structure is described and one part is missing, commonly the threshold at which the system stops or what happens after it is wrong.
Vendor Published

The terms make the treating provider responsible for reviewing, editing and approving all generated content, and the desktop dashboard is where surgeons review notes and submit codes. Concierge has its routing written down: urgent and post operative calls are warm transferred with a spoken summary staff hear before accepting, urgent callers trigger a text to a staff phone, red flags go to the emergency room or a person under each practice's rules, and complex calls go to staff with context.

The missing part is the routine side: calls Concierge judges non urgent are booked without a person. The scribe page also claims 100% laterality capture and zero missed follow ups, figures without a method that the provider's review has to answer for.

BB on Model and Technology TransparencyThe approach or the suppliers are named without the version and update discipline behind them. Naming a supplier is the entry to this band both here and on Model Supply Chain Disclosure, which ask different questions of the same disclosure: who receives the data, and what produces the output.
Vendor Published

The suppliers behind each step are named. Deepgram converts encounter audio to text, Anthropic's Claude, reached through AWS Bedrock, writes clinical notes from the transcript, and Google's Gemini, through Vertex AI, extracts clinical information from documents, attachments and audio. Orthopedic templates for each subspecialty, configured before go live, shape what the notes contain.

The homepage says the AI is trained on orthopedic data and validated in orthopedic clinics, without saying whether that means tuned models or templates and prompts, which model versions are in use, or how a model change reaches a practice, and there are no release notes. The 97% first pass accuracy figure has no definition, sample or method behind it.

BB on Model Supply Chain DisclosureSubstantial partial disclosure, or a chain that is structurally short: an in house build, a cleared model that cannot be quietly swapped, or a deployment where the transfer does not occur at all. Naming the model provider exits the band below into this one; the axis rises from here on the completeness of the party list and on the terms that govern data once it arrives.
Vendor Published

The privacy policy, updated in September 2026, names each party and its role: Deepgram for speech to text, Anthropic through AWS Bedrock and Google Gemini through Vertex AI as language models, PostHog for product analytics without patient information, AWS for hosting and Firebase for sign in, with a separate subprocessor list.

It adds terms for the model layer: the AI providers process data for inference only, do not train on it, and audio and transcripts are deleted from their servers after processing, and the trust page says protected health information goes only to subprocessors bound by business associate agreements. No telephony carrier for Concierge's calls is named, and where Google, Deepgram and Anthropic process the data is not stated.

CC on Clinical and Operational EvidenceNamed customers, or vendor reported percentages with no method, denominator or reference standard. Scale of use is recorded here and is not treated as evidence of benefit.
Vendor Published

Mira reports more than 10,000 encounters a month across more than 10 orthopedic clinics, an 18% average revenue increase, 11% higher collections, more than 430 hours a year back per surgeon, about 12 minutes saved per note, denials falling from 9.4% to about 4.1% and clean claims rising to about 85% from 60% to 70%, none with a practice, period or method attached.

Named surgeons describe their own use: Daniel Gittings of Orthopaedic Specialty Institute estimates about an hour a day saved on documentation, Matthew Fury of Baton Rouge Orthopaedic Clinic writes all his op notes in Mira and bills surgeries the same day, and neurosurgeon Jason Liauw of Orange County Neurosurgical Institute replaced Dragon dictation and two EMRs. Kathryn Hanna, managing partner of Orthopaedic Associates of Maine, is quoted without a figure, and customer logos include Reno Orthopedic Center and Ventura Orthopedics.

CC on AI Safety and PHI StewardshipGeneral assurances of privacy and security that do not answer the questions artificial intelligence raises: what is retained, what reaches a model, and what happens to it there.
Vendor Published

Retention is stated precisely: audio, transcripts and notes are kept seven years by default, a practice can shorten that to deletion as soon as a note is finalized, data requested for deletion is removed within 30 days, and after termination a practice has 30 days to export before everything, backups included, is deleted. Mira never trains models on identifiable patient data, and its AI providers process data for inference only.

Under the business associate agreement, though, Mira may deidentify data to HIPAA standards and use it to improve its products and support clinical research, a grant over customer data that the training promise does not close. Hallucination checks, and the handling of a safety event in a note or a call, are not covered.

Regulatory and Compliance
BB on HIPAA and BAA PostureBusiness associate status is stated and supported by a substantive privacy document, with the agreement or its scope not fully published. For a vendor outside the United States, an equivalent regime documented to this depth grades here. A vendor whose product does not process protected health information grades here too when it says so plainly and explains the scope, with a service privacy document behind it: stating a position a buyer can rely on is the posture this axis grades, and the band above is closed to it because there is no agreement to publish.
Vendor Published

Mira acts as a business associate and executes an agreement with every covered entity before any protected health information is processed; the homepage says signed agreements are available before execution, and the trust page says a standard agreement is sent by email request within one or two business days.

The privacy policy covers the platform's handling of protected health information, and the trust page gives terms from the agreement: breach notice within 72 hours and a written report within 15 days, subprocessors bound by their own agreements, and cyber liability insurance. The agreement itself is not posted, so a practice reads it only after asking.

DD on Security Certifications and Trust CenterControls are asserted with nothing independent behind them, or nothing is published. Read the note before concluding anything: this is the grade most often corrected on a second pass, because assurance material frequently sits on a parent domain or inside an old announcement rather than on the product pages.
Vendor Published

Mira holds no SOC 2 report and says it is working toward one with Vanta; the SOC 1, 2 and 3, ISO 27001 and HIPAA attestations its trust page lists belong to AWS. The controls are described in detail: AWS WAF and Shield, rate limiting, a private VPC with isolated subnets and flow logs, AES 256 encryption with rotated AWS KMS keys, TLS in transit, Firebase authentication, role based and least privilege access, and CloudWatch alarms.

Penetration testing is said to happen regularly, with no cadence, tester or report named. The ONC disclosure page for Mira EMR lists security criteria, but its certification number and date are still placeholders, so no outside check is yet on record.

DD on FDA and Regulatory StatusThe positioning is unclear or unsupported: clinical claims with no regulatory position stated, or language implying a status the databases do not carry. An open enforcement action that names the product also grades here, whatever the positioning: an FDA warning letter with no close out letter posted, or a consent decree in force. The note says which of the two applies, and the grade returns to the positioning bands when FDA closes the action.
Vendor Published

Concierge triages orthopedic calls with no regulatory position stated. It asks about body part, side, mechanism and onset, checks weight bearing, swelling, numbness and open wounds, sets visit length by body part, books urgent fracture or sports slots and sends red flags to the emergency room under each practice's rules.

Urgent and post operative calls reach staff, but for the calls it judges routine the agent settles the disposition itself, and the protocol behind those calls is not given. The scribe sits on the documentation side, its notes and code suggestions drawn from the surgeon's own encounter and approved by the provider under the terms, which state that AI output supports rather than replaces clinical judgment.

DD on AI Governance and Bias DisclosureNothing published on how model behavior is governed or tested. Multilingual operation with no subgroup performance sits here when the vendor markets recognition quality as a strength, because a caller the system failed to understand leaves no complaint and no record.
Vendor Published

Concierge is sold as speaking more than ten languages natively, English, Spanish, French, German, Portuguese, Italian, Dutch, Hindi, Russian and Japanese among them, and no results are given for any language or accent. A patient whose description of a swollen knee or numb hand the agent misheard in any of them leaves no complaint and no record of the miss.

For the scribe, the 97% first pass accuracy and 100% laterality capture figures carry no method, sample or breakdown by subspecialty or note type. No evaluation process, model review before release, bias testing or outside audit is described, and the clinical advisors listed on the team page are not tied to any review of the models.

CC on AI Liability and RecourseMechanisms exist that let someone challenge an output, such as audit trails, source traceability or review before commit, with nothing standing behind the output and no route for the harmed party.
Vendor Published

Review before use is the check: under the terms the treating provider reviews, edits and approves all generated content, and the dashboard lets surgeons edit notes and codes before submitting them, while Concierge hands staff a spoken summary of each transferred call. Once a note or code is approved, the risk sits with the practice. The terms cap Mira's total liability at fees paid in the prior 12 months and give no warranty on the accuracy of notes, codes or call outcomes.

The 99.9% uptime commitment on the homepage covers availability rather than what the system writes or says, and no credit, correction obligation or remedy is attached to a wrong code or a missed laterality.

Integration and Deployment
BB on EHR and Interoperability DepthNamed systems with read access or one directional writing, or standards support with named deployments behind it.
Vendor Published

Notes, diagnoses, orders and codes from the scribe land directly in the chart, and Concierge books, reschedules and confirms appointments in the EHR, across Epic, athenahealth, ModMed, eClinicalWorks, NextGen and DrChrono, with new systems built in one to four weeks. That is two way work with named systems, but the method behind each connection, marketplace listings and integration guides are not offered, and no named practice is tied to a particular EHR in production.

Mira EMR's disclosure page lists transitions of care, a standardized FHIR API, electronic health information export and public health reporting criteria, with its certification number still to be added.

BB on Deployment Model and Data ResidencyOptions and residency are stated with isolation or the processing path left open.
Vendor Published

Data is stored in encrypted databases in US based AWS data centers inside private VPCs with isolated subnets, and Mira states that data never leaves United States jurisdiction. The model processing path is the open part: audio goes to Deepgram, transcripts to Anthropic through AWS Bedrock, and documents and audio to Google Gemini through Vertex AI, and where Google, Deepgram and Anthropic process that data is not specified, though all three delete it after processing. Mira runs as a hosted service only, deployed on a phone, desktop or browser, with no on premises option, and deployments on supported EHRs go live within about 48 hours.

Commercial
CC on Commercial TransparencyNo price is published and the posture is discoverable: a buyer can establish how the product is sold and what drives the cost before contacting the vendor. Most of the index sits here.
Vendor Published

No price is shown for Scribe, Concierge, Dictate, Operations or RCM, and the route in is a booked demo. How Mira charges can be partly pieced together from the cost disclosure for Mira EMR: a recurring subscription priced per practice with certified features included, a one time implementation and data migration fee quoted before signing, per interface fees for third party connections, and no fee for the patient API or data export.

Whether the AI products are priced per surgeon, per encounter or per practice, and whether they are sold separately or as a bundle, goes unsaid. Mira's headline value claim is more than $300,000 per surgeon per year in recovered revenue and saved overhead.

BB on Setting and Specialty CoverageCoverage is named with validation behind part of it.
Vendor Published

Orthopedics is the whole focus: subspecialty templates for hand, spine, sports medicine, trauma, total joints and pediatric orthopedics, a surgery center offering for the operating room, facility claims and PRO-PM reporting, and reference tools covering more than 2,100 CPT codes and 1,322 musculoskeletal ICD-10 codes.

Named users are orthopedic practices, among them Reno Orthopedic Center, Orthopaedic Specialty Institute, The Orthopedic Center of St. Louis and Baton Rouge Orthopaedic Clinic, plus a neurosurgery institute. Concierge answers patients' calls around the clock and RCM codes and bills orthopedic claims, the patient facing and revenue cycle jobs beside the scribe. No surgery center is named as a customer.

Alternatives

Alternatives to Mira Health

Other products that do the same kind of work as Mira Health, and what each one does.

EzyScribe is an ambient AI medical scribe from Pennhealth Informatics LLC of Trevose, Pennsylvania, a clinical documentation and business process outsourcing company working since 2017.

Marvix AI, based in Boston, sells clinical documentation software built for specialty practices, where visits run long and histories span years.

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.

Uncovr turns operating room video into the operative report and the procedural codes that follow it.

Decoda Health sells an AI native electronic record and operations system for elective care practices, including medical spas, wellness clinics and weight management programs.

Comprehend Health, also sold as Comprehend PT, is an AI scribe built for physical therapy. It listens to a patient visit, or to a short summary the therapist records afterward, and writes the daily note, evaluation or progress note in the therapist's own style.

Commercial

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
Not published.
Mira sells after a demo and shows no price for Scribe, Concierge, Dictate, Operations or RCM. Its cost disclosure for Mira EMR describes a recurring subscription priced per practice, with certified features included. Whether the AI products are priced per surgeon, per encounter or per practice is not stated. Mira signs a business associate agreement with every covered entity before any protected health information is processed. The agreement is sent on request. Mira EMR carries a one time implementation and data migration fee quoted per practice before signing, and third party interfaces carry a setup fee per interface. Deployments on supported EHRs go live within about 48 hours, and new EHR integrations take one to four weeks. Vendor Published

Mira EMR's patient API and electronic health information export carry no fee. Mira's value claim is more than $300,000 per surgeon per year in recovered revenue and saved overhead.