DAX Copilot Explained (Now Dragon Copilot)

Short answer: DAX Copilot is Microsoft’s ambient AI scribe. It listens to the clinician-patient conversation, separates the speakers, and generates a structured draft note the clinician reviews before signing. Its defining feature is depth of Epic integration — the note appears in the chart with no copy-paste step. It was rebranded Dragon Copilot in March 2025, which is why search results are confusing.

If you have been comparing “Nuance DAX”, “DAX Copilot” and “Dragon Copilot” and wondering whether they are three products, they are not. Here is the lineage and what actually matters.

The naming, briefly

Nuance built DAX, the Dragon Ambient eXperience. Microsoft acquired Nuance. DAX Copilot, sometimes written Dragon Ambient eXperience Copilot, was the ambient scribe product. In March 2025 it was folded into the Dragon brand and became Dragon Copilot, sitting alongside Dragon Medical One, the long-standing dictation product.

So: Dragon Medical One is dictation — you narrate, it transcribes. Dragon Copilot (formerly DAX Copilot) is ambient — it listens to the visit you were having anyway. Both run on the same underlying speech infrastructure, which is why the EHR coverage is shared.

Vendor materials, review sites and Microsoft’s own support pages still use both names. Assume they mean the same product unless the context is clearly historical.

One more conflation worth clearing up: DAX dictation is not a product. People search it because Dragon is famous for dictation and DAX carries the Dragon name, so the two blur together. They are opposite interaction models. Dictation means you speak to the software deliberately, narrating the note in the third person after the visit. DAX means you speak to the patient and the software listens to a conversation it was not addressed in. If you want DAX dictation, what you almost certainly want is Dragon Medical One, and the two are licensed separately.

What it actually does

The pipeline is the one every ambient scribe runs, executed well:

Ambient capture with multi-speaker differentiation, so patient statements are not attributed to the clinician.
Structured note generation across a reported 37+ specialties, producing the usual sections — HPI, ROS, physical exam, assessment and plan.
Note lands in the chart. With Epic this is the differentiator: generation happens inside the chart rather than producing text you paste.
Ancillary documents — referral letters and after-visit summaries generated from the same encounter.
Clinician review before finalisation. Nothing is filed unreviewed, which is the correct design.

In 2026 the ambient capture extended into nursing workflows, widening it from clinician-only documentation to broader care-team coverage.

What DAX does well is also the baseline any custom scribe build has to clear, so it is worth understanding precisely.

Where it fits

Epic shops. This is the honest headline. If you run Epic, Dragon Copilot’s integration depth is hard to beat and that alone justifies shortlisting it. Broader EHR support — athenahealth, MEDITECH, Oracle Health and 200+ others via the Dragon Medical One infrastructure — is real but generally shallower than the Epic path.

Organisations wanting a note. If your documentation burden is a well-formed clinical note, this is a mature product solving exactly that.

Enterprise buyers. It is sold through enterprise channels with a sales process, not self-serve. That suits health systems and frustrates small practices.

Where it stops

Your burden may be forms, not a note. This is the distinction that decides whether any ambient scribe helps you, and it is the most common scoping error we see. A well-generated SOAP note is genuinely useful if a SOAP note is what your chart requires. If you file a set of structured mandatory forms specific to your setting, the clinician still reads the note and re-enters the same clinical facts across those forms. The transcription was never the bottleneck. We went into this in what is an AI medical scribe.

Specialty and workflow edges. Broad specialty coverage means tuned for the common case. Unusual documentation, non-standard templates or workflows the product does not model are where you hit limits.

Pricing is enterprise-shaped. Not published, quote-based, with implementation fees and minimum terms. We put the reported numbers side by side in what AI medical scribes actually cost.

Data residency. Audio, transcripts and notes are all PHI. If yours cannot leave your boundary, that constraint is decided before features enter the conversation — see HIPAA-compliant AI architecture.

What DAX Copilot costs

Microsoft does not publish a price list, so every figure in circulation comes from third-party reporting and reseller disclosure. Treat the following as a range to open a negotiation with rather than a quote.

Reported pricing has sat around $450–600 per clinician per month at small volumes, tiering down as headcount rises. Setup fees have been reported near $650 for the first user and $250 for each additional, and some plans carry a 12-month minimum term. Reported list figures were cut sharply during 2026, so any number more than a few months old is stale.

Three things matter more than the headline when you ask for a quote:

The price at your headcount, not the list price. The spread within this one vendor is wider than the spread between vendors.
The setup fee and minimum term, which is where a cheap monthly rate gets its money back.
Whether the Epic integration is included at your tier or priced separately, since that integration is the reason most buyers are looking at Dragon rather than a cheaper ambient scribe.

For how this compares against Abridge, DeepScribe, Suki and building your own, the AI medical scribe cost breakdown runs the full comparison.

The alternatives, and who each one is for

Positioning below, not feature claims. Vendors change what they ship faster than any article can track, and accuracy numbers quoted in comparisons are almost never measured on documents like yours.

Product	Aimed at	The distinguishing thing	How you buy
Dragon Copilot (was DAX)	Health systems and large groups	Depth of Epic integration; the note lands in the chart	Enterprise contract, per clinician, setup fee
Abridge	Health systems	Structured output with coding support; enterprise partnerships	Enterprise, not sold to individuals
Ambience	Specialty groups and systems	Specialty-specific notes and coding emphasis	Enterprise
Suki	Mid-size groups, and developers	Ambient plus dictation and voice commands, with an API to build on	Quote-based, usually annual
DeepScribe	Ambulatory practices	Configurable note styles per clinician	Contract
Nabla	Clinicians and systems, including Europe	Self-serve entry with an enterprise tier above it	Both
Freed	Solo clinicians and small practices	Sign up and use it the same day, no procurement	Monthly, per clinician

Two things fall out of that table. The market splits between products you procure and products a clinician can start using this afternoon, and the right answer usually follows from which of those your organisation actually is. And every one of them produces a note, which is only the whole job if a note is the whole job. If your documentation burden is structured forms rather than prose, none of them is aiming at your problem.

How to evaluate it against alternatives

Comparison articles rank these on accuracy claims that are hard to verify. Run your own test instead:

Your audio. Your accents, specialty vocabulary, room noise, and the interruptions real consultations contain.
Your documents. Not “can it write a note” — can it complete what you are required to file.
Count what remains. Fields a clinician still fills by hand afterwards. That number is the product.
A difficult encounter. Does it flag uncertainty or produce something plausible and wrong? A confident wrong field is worse than a blank one, because blanks get noticed.
The full-term price, including implementation and renewal, in writing.

Point four separates serious systems from impressive demos, and no demo will show it to you unprompted.

When building makes more sense

We build custom AI medical scribes and we tell most people to buy one. Dragon Copilot is a good product and if it fits, buy it.

Building is right in four situations: your burden is structured forms rather than a note; your EHR is not meaningfully supported; data residency rules out the vendor; or per-encounter pricing has become your dominant line item. Our production build auto-fills more than ten mandatory forms from a single encounter with no clinical fact entered twice — that de-duplication is the thing products do not do for your form set, because your form set is yours. The full comparison is in build vs buy an AI medical scribe.

The takeaway

DAX Copilot and Dragon Copilot are the same product; Dragon Medical One is the separate dictation tool. It is a strong ambient scribe, strongest on Epic, aimed at enterprise buyers whose output is a clinical note. Test it on your own audio and your own documents, and count the fields left over — that number decides it, not the accuracy claim.

Sources

EpochC builds HIPAA-compliant AI medical scribes and clinical AI systems, including one auto-filling 10+ mandatory forms per consultation. Start a project — we will tell you honestly if a product solves it.

DAX Copilot explained: what it is, what it does, and the rename to Dragon Copilot