athenaAmbient

The model is the engine of this module, and the record system around it carries its own value. athenaAmbient listens to the visit and drafts the note, possible diagnoses, recommended orders and prescriptions, and none of that exists without the models behind it.

It is still a feature of athenaOne rather than a product sold on its own. It is delivered through routine athenaOne updates at no extra charge, it runs only for athenaOne customers, and the practice management, billing and record system it sits in are what those customers bought.

That places it at B.

Review is asserted and the controls around the clinical suggestions are not described. athenahealth presents athenaAmbient as an opt in part of the encounter that drafts the note and, in its own words, automates draft notes, diagnoses and prescriptions, with recommended orders and care gaps surfaced in the chart. athenahealth says clinicians review and edit Ambient Notes drafts.

What is missing is the mechanism for the parts that act on the patient. Nothing published says how suggested diagnoses, orders and prescriptions are presented, whether each must be accepted individually, what the system does when it is unsure, or how errors found after signing are handled. Review of a note is described; review of a prescription is implied.

Ask how diagnoses, orders and prescriptions from athenaAmbient reach the chart, what must be accepted one by one, and what athenaAmbient will not suggest.

The architecture is described in general terms and nothing is identified. athenahealth describes an Advanced Intelligence Layer that combines engines, large language models and machine learning with an ambulatory care dataset, and says its cloud platform lets it iterate on AI features quickly.

No model, version, training data description or change practice is published for athenaAmbient, and quick iteration across a shared platform makes the update question more pressing rather than less.

Ask which models produce the transcript and the note, how customers are told when they change, and whether a practice can pin a version.

The architecture is described and no model provider is named for athenaAmbient. athenahealth names the partners whose engines a practice can choose instead inside Ambient Notes, Suki and Microsoft Dragon Copilot among them, but for its own model it describes an intelligence layer of engines and large language models without saying whose.

athenahealth's help page says it binds third parties that access customer data with business associate agreements, and no subprocessor list or retention and training terms for a model provider are published.

Ask whether athenaAmbient sends audio or text to an outside model provider, who it is, and the terms that govern data once it arrives.

Figures are published, and none isolates athenaAmbient. athenahealth's August 2026 release reports network results since late 2025 of nearly six fewer minutes per encounter on chart preparation and documentation, more than two workdays a month per clinician, and up to 30 percent more charts closed the same day among early adopters. Those describe its AI capabilities together, and no method, denominator or baseline is given.

The Ambient Notes page carries more method: a customer survey of 153 respondents in February and March 2026, in which 90 percent reported a meaningful reduction in burnout, and an OrthoLoneStar case study reporting 36 percent less after hours documentation and 3.6 minutes saved per encounter. Those results come from Ambient Notes across its engines, most of them before athenaAmbient was released, so they are not evidence for this model. A family medicine owner in Colorado is quoted on the release.

Ask for documentation time and note accuracy results for athenaAmbient itself, with the number of clinicians and the comparison used.

One element is stated specifically and the ones ambient capture raises are not. athenahealth's patient facing privacy notice says it de identifies protected health information under the HIPAA expert determination or safe harbor methods, and athenahealth describes its intelligence layer as drawing on a data network built from its ambulatory customers.

Nothing published says how long visit audio and transcripts are kept, whether athenaAmbient's models are trained or tuned on customer encounters, identified or de identified, or what reaches any outside model provider. For a feature that records the whole visit, those are the first questions.

Safety engineering, meaning guardrails, hallucination mitigation and how a safety event is handled, is not published, as it is not for any record in this index.

Ask for audio and transcript retention, the training position for athenaAmbient in writing, and whether de identified encounter data feeds its models.

Business associate status is stated and supported, and the agreement is not published. athenahealth's privacy notice for patient facing services describes its role as a business associate to the healthcare provider, and its website privacy policy says processing for provider customers is governed by agreements that may include business associate agreements. The athenaOne help page on security and HIPAA says athenahealth will have business associate agreements with any third party it hires that needs access to customer data.

The agreement itself sits in each customer contract, and nothing published says how it covers ambient audio or the model processing behind athenaAmbient.

Ask for the agreement terms that apply to athenaAmbient, and whether a model provider is among the subcontractors bound.

Recognized certifications are named and the scope question is left to the buyer. athenahealth's certifications page lists HITRUST CSF certified status, a SOC 1 report under SSAE 18, PCI DSS, EHNAC accreditation, DirectTrust accreditation and ONC 2015 Edition certification, achieved, in its own words, for all or part of its products and services.

No report or certificate is retrievable, no dates are given, and the page does not say whether the HITRUST scope reaches athenaAmbient or the processing behind it; it directs buyers to sales or their customer success manager. The SOC 1 report addresses financial reporting controls rather than security operations.

Ask for the HITRUST letter with its scope and date, and confirmation that it covers athenaAmbient's audio and model processing.

No clearance is claimed and no regulatory position is published. The note drafting needs none. The feature also infers possible diagnoses and recommends orders and prescriptions, which is clinical content, but each suggestion is drawn from a visit the clinician took part in and from the chart in front of them, so the basis is reviewable by the person accountable for it.

athenahealth has not said how it classifies the diagnosis and order suggestions. Held at C with that gap stated.

Ask athenahealth for its regulatory analysis of the diagnosis and order suggestions, and whether any can reach the record without a clinician accepting them.

A process is named and no evaluation is published. athenahealth says an EHR AI CoLab of clinicians and superusers within its athenaInstitute validates model outputs and workflow fit, alongside thousands of alpha and beta testers across more than 70 user groups, before tools reach broad release.

That describes who looks at the output, not what was found. No accuracy evaluation, fairness method or results by language, accent, specialty or patient group are published for athenaAmbient.

Ask for the CoLab's evaluation criteria and results, and for any accuracy figures broken out by language and specialty.

Review before sign off is the mechanism, and nothing stands behind the output. Drafts and suggestions reach the clinician in the chart, where an error can be caught before it is signed.

athenahealth publishes no error rate for athenaAmbient's notes, diagnoses, orders or prescriptions, no limitations list, and no warranty, indemnity or remediation commitment for the feature, and the Ambient Notes page notes that individual results may vary. No route is described for a patient affected by a suggestion that was accepted in error.

Ask what athenahealth commits to when athenaAmbient drafts something that was not said or suggests a wrong order, and for any measured error rate.

athenaAmbient is part of athenaOne itself: it drafts into the patient's encounter and works from the chart, with no interface to build. Being inside the record is not what this grade measures; the record system's exchange with systems outside it is.

athenahealth publishes ONC 2015 Edition certification of its health IT and DirectTrust accreditation as a health information service provider, which are standards based exchange credentials with athenahealth's own network behind them. The pages that describe the feature do not set out named exchange networks or volumes, which is what the band above asks for.

The reverse side, stated plainly: athenaAmbient runs only inside athenaOne. Ask for the exchange networks athenaOne is live on and how records received from them reach the chart athenaAmbient drafts from.

A single hosted service, with location implied. athenaOne is a cloud based, multi tenant record system run by athenahealth, which describes itself as headquartered in the United States, and athenaAmbient runs inside it. No hosting region, residency commitment or description of where ambient audio is processed is published, and athenahealth's privacy policy carries an international transfer section without placing this processing.

Ask where visit audio is processed and stored, whether any step uses an outside model service and in which region, and how one practice's data is isolated from another's.

The pricing basis is published. athenahealth says athenaAmbient is delivered through its routine update process and included in the athenaOne subscription at no additional cost, so an existing athenaOne customer can size the incremental cost before any conversation: none.

What stays unpublished is the subscription it rides on. athenaOne's own price is not on athenahealth's site, so a practice not yet on athenaOne cannot size what it would pay, and partner engines inside Ambient Notes carry their own terms.

Ask whether any part of athenaAmbient, such as order or prescription capture, is tied to a higher athenaOne tier, and what partner engines cost by comparison.

Coverage is stated and nothing published validates it for this model. athenaAmbient is available to the more than 170,000 clinicians on athenaOne, which serves ambulatory practices, on desktop and mobile. athenahealth has said Ambient Notes as a whole has been adopted across nearly 50 specialties, but that count covers all its engines.

No results are published by specialty or setting for athenaAmbient, and the alpha and beta testing across more than 70 user groups is described without outcomes.

Ask which specialties have templates and measured results for athenaAmbient, and how it performs against the partner engines a practice could choose instead.