Epic AI Charting
AI Charting is the ambient documentation feature Epic builds into its own record system, also marketed as Chart with Art. It listens during a visit, drafts the clinician's note, and queues orders from the conversation for the clinician to review. Clinicians can restructure current and future notes by voice, for example asking for the history of present illness as a bulleted list. It is one of the features Epic groups under Art, its AI for clinicians, alongside In Basket message drafting.
Epic announced AI Charting at its August 2025 Users Group Meeting and released it to customers on 4 February 2026 across several outpatient specialties. In March 2026 Houston Methodist became the first organization to use it for bedside nursing, and Epic said a home care version would follow. By the August 2026 meeting Epic reported it live in more than 70 specialties and across physicians, nurses, pharmacists and behavioral health, with the nursing version live at 11 health systems. Epic says more than 85 percent of its customers use Epic AI in some form.
Epic built AI Charting with Microsoft, which said it supplies its Dragon ambient AI technology for transcription, speaker separation and note components, running through Azure. Epic describes AI Charting and Microsoft Dragon Copilot as separate products.
Two facts frame any evaluation. It is available only to organizations running Epic, and because it sits inside the chart it can draw on the patient's medications, problems and history while drafting. No price, accuracy figure or independent evaluation of the feature is published.
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
The model is the engine of this module, and the platform around it carries its own value. AI Charting listens to the visit, drafts the note and queues orders, and none of that exists without the speech and language models behind it, so the capability being licensed is AI from end to end.
It is still a module of Epic's record system rather than a product sold on its own. It runs only inside Epic, it is switched on by organizations already licensing that system, and the record it writes into is the thing they bought first.
That places it at B: the capability does not exist without the model, and the platform carries other value.
The oversight structure is described in outline, and the part after an error is missing. AI Charting drafts the note and queues up orders based on the conversation; Epic's own words describe drafting and suggesting, so the clinician reviews the note and decides whether each order is placed. Because the feature sits in the chart, the draft is built with the patient's medications, problems and history in view.
What Epic does not publish is the boundary. Nothing says which order types can be queued, whether any are excluded, how queued orders are presented for approval, what the system does when it is unsure, or how errors found after signing are handled.
The supplier is named, without versions or an update practice. Microsoft said in August 2025 that it supplies its Dragon ambient AI technology to Epic's charting capability, and Epic told Digital Health News that Microsoft provides transcription, speaker separation and important note components while Epic does the listening inside its own Haiku and Canto apps. In February 2026 Epic said the feature uses multiple AI models accessed through Azure.
Which models draft the note, which version is in service, and how changes are announced are not published; Epic describes AI Charting and Microsoft Dragon Copilot as separate products without saying what differs underneath.
The model provider is named and the terms that govern data once it arrives are not published. Microsoft stated in August 2025 that it brings its Dragon ambient AI technology to Epic's charting capability, Epic said Microsoft supplies transcription, speaker separation and note components, and Epic described the models as accessed through Azure. A buyer can identify the party that receives the visit audio.
What is not public: a subprocessor list for the feature, which Microsoft services process the audio and in which region, whether audio and transcripts are retained or logged, and whether they are excluded from training. Epic's Data Privacy Framework policy says data passes to third parties only for uses customers authorize, but does not address this path.
Named customers and scale of use are published, and no outcome figures for the feature are. Epic's release says clinicians report staying focused on the patient and capturing a more complete record of the visit, with less work after the patient leaves, and it quotes a provider at Group Health Cooperative of South Central Wisconsin. At its August 2026 meeting Epic reported the feature live in more than 70 specialties and the nursing version live at 11 health systems.
Those are deployment counts and testimonials. Epic publishes no documentation time, note quality or accuracy result for AI Charting, and cites no independent or peer reviewed evaluation of it. The usage figures in the same release, 16 million monthly uses of Art's Insights summaries, describe a different feature.
General assurances that do not answer the questions ambient capture raises. Epic's Artificial Intelligence page describes a HIPAA compliant pipeline for its language models, and its Data Privacy Framework policy says Epic processes patient data for its healthcare customers, that staff access it only with permission, and that it passes to third parties only for uses customers authorize.
Nothing public states how long visit audio and transcripts are kept, by Epic or by Microsoft, whether they are excluded from training, or whether anything is de identified. For an ambient product those gaps matter most, because the input is a recording of the whole visit.
Safety engineering, meaning guardrails, hallucination mitigation and how a safety event is handled, is not published.
A privacy document reaches the service, with no business associate statement. Epic's Data Privacy Framework policy, effective 13 March 2026, describes Epic as a data processor or service provider to its healthcare customers and says staff access patient data only with the customer's permission, and Epic's Artificial Intelligence page describes its generative pipeline as HIPAA compliant.
No public statement of business associate status, agreement terms, or how those terms extend to the speech and language processing Microsoft supplies for AI Charting exists; those sit in each customer's Epic contract.
Controls are asserted with nothing independent published. Epic's public pages describe a HIPAA compliant pipeline for its generative features and link privacy policies, but name no SOC 2 report, ISO/IEC 27001 or HITRUST certification, penetration testing cadence or trust center; compliance documentation reaches customers through Epic's login only UserWeb.
For self hosted customers much of the security picture belongs to the health system's own environment, but the audio and transcription path runs through Microsoft, and Epic does not say which certifications cover it. Epic's chief security officer spoke about patching and an industry AI security initiative at the August 2026 meeting, which is posture rather than attestation.
No clearance is claimed and no regulatory position is published. The note drafting needs none. The order queuing is clinical content, but it is drawn from a conversation the clinician took part in and is placed only when the clinician accepts it, so the basis of each suggestion is in front of the person accountable for it.
Epic has not said publicly how it classifies the feature or whether queued orders are positioned as decision support.
Nothing Epic publishes says how AI Charting's behavior is governed or tested. The release, the Artificial Intelligence page and the privacy policies carry no evaluation of draft accuracy, no responsible AI policy specific to ambient documentation, and no results by language, accent, specialty or patient group; product documentation sits behind Epic's customer login.
The gap matters more for ambient capture than for text features, because recognition quality varies with speech, and a part of the visit the system misheard leaves nothing in the note to flag it. With the feature live across more than 70 specialties, the evidence to measure it exists inside Epic's customers.
Review before commit is the mechanism, and nothing stands behind the output. The note is a draft the clinician edits and signs, and queued orders wait for the clinician, so an error can be caught at the point of use with the chart beside it.
Epic publishes no error rate for drafted notes or queued orders, no limitations list, and no warranty, indemnity or remediation commitment for the feature, and no route is described for a patient harmed by an order that began as a suggestion.
The feature is part of Epic's own record system: it listens inside Epic's apps, reads the patient's medications, problems and history, and writes the note and queued orders into the same chart, with no interface to build. Its reach past that chart depends on the record system's exchange with systems outside it.
Epic reported at its August 2026 meeting that 9.3 billion patient records were exchanged across 50 states and 10 countries, half of them between Epic and other record systems, and that 90 percent of Epic organizations are live or installing on TEFCA. Its interoperability page documents Care Everywhere, FHIR APIs and public interface specifications on open.epic. The chart AI Charting draws on can therefore include records received from other systems.
In the other direction, the feature runs only inside Epic and cannot be bought by an organization on another record system.
The record system runs where the customer's Epic runs, on its own infrastructure or with Epic Hosting, named in Epic's Data Privacy Framework policy. The speech and language processing comes through Microsoft on Azure, and Epic publishes nothing on where that processing runs, in which region, or how one customer's audio is separated from another's.
For the part of the feature that matters most to residency, the visit recording and its transcription, location is implied rather than committed.
No price, and the buying route can be established. AI Charting is available only to organizations running Epic, which license it from Epic as part of their existing relationship; Epic said in August 2025 that customers would be able to license it in 2026.
Whether it carries a charge beyond the existing Epic license, whether any charge scales by clinician or encounter, and what implementation work is needed are not published. The comparison a buyer faces is between a feature added to a system already licensed and a separately contracted scribe with its own price and its own integration.
Coverage is stated clearly and nothing published validates it yet. Epic's release says its developers worked on site with physicians across many specialties, and the feature launched in February 2026 across several outpatient specialties. In March 2026 Houston Methodist became the first organization to use it for bedside nursing, and Epic said a home care version would follow. At its August 2026 meeting Epic reported AI Charting live in more than 70 specialties and in several clinical roles, physicians, nurses, pharmacists and behavioral health among them, with the nursing version live at 11 health systems.
Those figures say where the feature is switched on, not where it has been shown to work. No results are published by specialty, by setting, or for inpatient against ambulatory use, and the feature is limited to organizations running Epic.
Alternatives to Epic AI Charting
Other products that do the same kind of work as Epic AI Charting, and what each one does.
Abridge is an enterprise ambient AI platform that turns the conversation between clinician and patient into a structured, billable note inside the record.
athenaAmbient is the ambient scribe athenahealth builds into its own athenaOne record system.
Microsoft Dragon Copilot is the ambient clinical documentation assistant Microsoft created in March 2025 by merging DAX Copilot with Dragon Medical One, both from its 2022 acquisition of Nuance.
Oracle Health Clinical AI Agent is the ambient documentation and assistant layer built into the Oracle Health electronic health record, the platform Oracle acquired with Cerner, and it is now marketed as part of Oracle Health AI Agents.
Cerbo AI Scribe is the ambient documentation feature built into Cerbo, an EHR and practice management system for functional and integrative medicine, direct primary care, cash based practices, IV and hormone therapy clinics and medical spas.
iScribeHealth is a Nashville company founded in 2015 as a mobile documentation platform that combined speech to text dictation with human virtual scribe and transcription services, and that moved to ambient listening and generative AI as iScribe AI in 2023.
Compared With
Each comparison carries a written verdict, the buyer conditions that favor each vendor, and a graded side by side. Pairs that cross a category boundary are grouped separately, and their verdicts state where the boundary sits rather than manufacturing a head to head.
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
|
Licensed by Epic customers as part of their Epic agreement; terms not published. | Not published. Business associate terms sit in each customer's Epic contract, and nothing public says whether they extend to the Microsoft audio and transcription processing. | Not published. | Vendor Published |
No price is published. AI Charting is available only to organizations running Epic, which license it from Epic alongside their existing agreement. Whether it carries a separate charge, whether any charge scales by clinician or encounter, and what implementation work is billed are not stated.