Does Your EMR’s AI Exist Yet? A Vendor-by-Vendor 2026 Scorecard

The short version

• Shipped with receipts: ModMed Scribe 2.0 (240,000 visits in three months), Tebra AI Note Assist ($0.99 a note, public pricing), 4D Scribe, Meditab’s IMSGo transcript, Remedly’s claim predictor, and PatientNow’s Recura receptionist.
• Live with fine print: Nextech’s Cora Scribe is ophthalmology-only for now. Plastics and derm are announced. Get your specialty’s date in writing.
• Announced, not yet in release notes: Symplast’s five-feature AI suite, PatientNow’s AI Scribe, and EZDERM’s Eve ecosystem. Good companies, ambitious roadmaps… ask for dates, not adjectives.
• No specialty EMR ships generally available billing AI in 2026. Only 14% of providers point AI at denials while over half of health plans automate. Your denial queue is still a people problem.
• Buy on release notes, published pricing, and a reference customer using the feature. The AI page is where you start asking, not where you stop.

How we audited specialty EMR AI claims for 2026

In our 2026 EMR shortlist guide we promised a feature-by-feature AI status audit. Here it is. We read the release notes so you don’t have to… all of them, every vendor, every month back through 2024. It was a long week. Here’s how we scored it: a feature counts as shipped when there’s a release note, an app-store changelog, published pricing, or a user review describing it in use. A press announcement, a trade-show demo, or a marketing page counts as announced. An announcement is a save-the-date. A release note is the wedding.

Full disclosure before the list: we work inside most of these platforms every single day, several of them refer practices to us, and we genuinely like these companies. This is the same status check we’d run before any client signs an AI add-on, and this segment ships fast… so treat every “announced” below as a timestamp, not a verdict. Vendors: if we’ve missed a release note, tell us and we’ll update this the same week. Happily.

The specialty EMR AI scorecard: what’s live and what’s on deck at all nine vendors

1. ModMed has the receipts everyone else wants. Scribe 2.0 is live, trained on derm, plastics, and ENT visits, and logged 240,000 patient encounters in its first three months, with a vendor-reported million by mid-2026. Coding suggestions ride along in EMA, Klara routes patient messages, and Bonsai’s patient-reactivation AI came aboard in April. On deck: the Billing Assistant and Front Office Assistant announced last November. Ask for dates, not adjectives.
2. Nextech shipped Cora Scribe in January and the architecture is the interesting part: it writes structured findings into discrete chart fields instead of a summary blob, which is the difference between a note your biller can work with and a paragraph they have to decode. One catch: ophthalmology goes first. Plastics, derm, and med spa are next in line, so get your specialty’s GA date in writing… “over time” is not a quarter. The intelligent CRM and RCM-side AI are announced, not shipped.
3. Tebra publishes its AI pricing, and we wish that were less remarkable than it is: AI Note Assist runs $0.99 a note or $99 a month, live since June 2025. Reviewers grumble about the per-note charge, which is how you know they’re actually using it. Review Replies and Review Insights shipped even earlier. The Billing Assistant is early access for select customers, and the Voice Assistant is marketed as available but doesn’t have a documentation trail yet… talk to a practice using it before you budget around it.
4. Symplast has one shipped AI feature with a clean paper trail: video-visit transcription and summarization, live since September 2025, right there in its release hub. The wider suite on its AI page (scribe, ambient listening, chatbot, patient summary, daily huddle) hasn’t reached those release notes yet as of July 2026. Symplast documents what it ships, which is exactly why the release hub is the place to check. Ask the rep for dates, and if we’ve missed a note, we’ll gladly update this.
5. PatientNow is two AI stories in one platform. Recura, the acquired AI receptionist, is as real as it gets: named customers, live demos, and automatic booking-deposit collection since April. The AI Scribe on the homepage doesn’t have release notes, pricing, or docs yet. Buy the receptionist today. Get a date on the scribe.
6. EZDERM announced Eve in January and demoed it at AAD in March: chart prep, background note scrubbing, and two RCM tiers. It’s the most ambitious AI roadmap in derm, from the vendor with the best support in the business, and as of July it hasn’t reached the release notes. One number to keep straight in the sales conversation: the 95-98% collections figure attached to Eve RCM Complete belongs to EZDERM’s long-standing human billing team, who earned it the old-fashioned way. The humans have the track record. The AI hasn’t had time to.
7. 4D EMR shipped 4D Scribe in May on the back of its Knowtex partnership, which is the boutique-vendor playbook done right: partner with a scribe specialist instead of pretending to be one. The time-savings claims are the vendor’s own, so demo it on your visit types and judge for yourself.
8. Remedly is the sleeper, and I’ll admit a soft spot for this one. Its claim accept/reject predictor is live, documented, and tells you why a claim will bounce before you submit it, and ambient transcription entered beta in October. For a small shop, that’s a lot of shipped. The big platforms should be a little embarrassed.
9. CosmetiSuite inherits Meditab’s IMS AI: a scribe, a mobile transcript that handles English and Spanish, and a no-show predictor, all opt-in value-added services. Confirm they’re switched on in an aesthetic deployment before you count on them, and note the AI claim scrubbing on the parent blog is still written in the future tense.

The billing-AI gap: no specialty EMR ships autonomous billing in 2026

Now the part I care about most, because billing is where we live. Count the shipped billing AI in this segment on one hand: Remedly’s claim predictor and Tebra’s early-access assistant. That’s the list. ModMed’s Billing Assistant is a 2026 roadmap item. Nextech’s RCM AI is a phrase sitting on top of a rules-based scrubber. EZDERM’s Eve RCM is announced. Symplast, PatientNow, and 4D have no billing AI story at all, and to their credit, they don’t pretend to.

Meanwhile the payer side automated years ago. Over half of health plans use AI in administrative workflows, and only 14% of providers use AI to reduce denials, while 41% of providers now see denial rates above 10%. Flip that: the entity deciding whether your claim pays is running algorithms. The software defending the claim mostly isn’t yet. Payer algorithms have denied claims in roughly a second apiece. Your appeal takes longer. That’s the asymmetry in one sentence.

And the “autonomous coding” headlines you’ve seen, the 95-98% accuracy numbers, come from radiology, pathology, and ED work at health systems. Generally speaking, no production model is making cosmetic-versus-functional calls on a septorhinoplasty or arguing a denial appeal. We covered where AI actually earns its keep in billing: it moves humans to the exceptions. It doesn’t remove the humans watching your clean claim rate and days in AR.

So for 2026: weigh the documentation AI heavily, because it’s real and it’s good. Don’t pick an EMR on billing AI, because none of them ship it yet.

Five questions that separate shipped EMR AI from roadmap AI

Take these into the demo. Repeat after me: receipts, not roadmaps. A vendor with shipped software will enjoy these questions, and that reaction is itself the answer. Watch the rep’s face on number one… you’ll learn everything you need in about four seconds.

1. “Show me the release note.” Shipped features live in release notes and changelogs. If the feature only lives on the AI page, you have your answer and a friendly follow-up: “so what’s the GA date for my specialty?”
2. “Show it running in my specialty.” An ophthalmology scribe demo doesn’t document a rhinoplasty consult. Ambient AI is trained by visit type, and surgical-subspecialty satisfaction with generic scribes runs at about half the primary-care rate. Your visit mix or it doesn’t count.
3. “What does it cost, in writing?” Tebra publishes $0.99 a note. That’s the bar now. “Included in your quote” is not a price.
4. “Connect me with a practice using it.” A reference using the specific AI feature, not the platform generally. Vendors with adoption numbers volunteer them… ModMed leads with its visit counts for a reason.
5. “What are the model’s HTI-1 source attributes?” Certified EMRs with predictive AI are required to disclose training data, logic, and risk controls. A vendor with real AI has this document ready. It’s a fair question, and the good ones treat it that way.

Then hold the answers against your cosmetic split. The scribe buys the doc minutes. The billing operation protects the revenue. Staff the second one with people who read release notes for fun… we know a few.

Frequently asked questions

This article is for general informational purposes and is not coding, billing, or legal advice. Verify current rules and your contractor policies before making operational decisions.