What HIMSS 2026 Told Us About Where Ambient AI Actually Stands in Specialty Care

Marvix HIMSS blog title image
Bhavya Sinha
April 8, 2026

A year ago, ambient AI conversations at health system events were broad and exploratory. This year in Las Vegas, people were showing up with receipts. A lot of people had already run pilots and wanted to talk about where specialty care was still slipping through.

We also started noticing a shift in who we were speaking with. Many of the conversations were with CIOs, CMOs, and similar roles representing 20 to 40 provider groups, looking at how something like this would actually roll out across a practice.

The Gap That Kept Surfacing

Specialty care runs on longitudinal context, and assembling that context accurately is a harder problem than most ambient AI solutions have solved for. A patient's history sits across prior notes, imaging, and outside records, and documentation built only from today's encounter tends to leave coding accuracy and claim outcomes exposed in ways that surface slowly.

What struck us at HIMSS was how precisely people could describe this.

What Drew People In at the Booth

A Marvix AI team member giving a demo of the product to a booth visitor at HIMSS
  1. Patient Recap and Composite Note

    Patient Recap
    was the first thing that stopped people. Previous notes, imaging, lab reports, and outside records are pulled in from the EHR and organized into a structured AI-generated summaries (Patient Recap) before the visit even begins..

    Then the same history is carried into the current visit’s composite note and placed into the relevant sections, so it reflects the patient’s full clinical picture rather than starting from scratch each time.

    In some situations, people mentioned they already had access to this information, but it lived across systems and formats and took time to piece together. Here, it showed up in one place, already structured, and ready to be used during the consult.
  2. Billing Codes Automation

    The coding conversation opened up from there. Marvix runs coding logic inside the workflow, so E/M codes, ICD-10-CM codes with laterality, and HCC/RAF scores are ready when the note is. The MDM rationale sits inside every note, which meant people could see exactly how the coding decision was reached.

    That's what made the audit-ready output feel real rather than assumed. First-pass acceptance rates were the metric that kept coming up as the actual downstream measure of whether any of it was working.
  3. Custom Templates

    Another feature that impressed people at our booth was custom templates. In a lot of systems, everyone within a specialty ends up using the same structure, which doesn’t always reflect how individual physicians actually document.

    With Marvix AI, each physician can have custom made templates that match how they write. This is done via neural style transfer which learns from the way a provider documents (structure, phrasing, formatting, brevity etc) and carries that through into every note and other kinds of documents being used by the provider.

The Question That Surfaced Toward the End of Every Conversation

By the end of most conversations at the booth, the question had shifted from what Marvix does to how quickly it could be up and running. In most cases, this was coming from teams thinking about rollout across multiple providers, which is where integration and workflow questions started to matter more.

If there are two-way EHR integrations for example with Advanced MD, eClinicalWorks, Veradigm, Athenahealth, and other major systems, how data is pulled and how it is pushed back into the EHR. The real-time collaboration across physician and medical assistant roles within the same note also drew interest, particularly from people with bigger teams

What We Took Away from Las Vegas

HIMSS 2026 made something clear about where ambient AI in specialty care is heading. Health systems are now asking whether the technology works specifically for the complexity of specialty workflows, and that's a meaningfully different question from where the conversation was a year ago. The people walking the floor had already done one round of evaluations and arrived knowing exactly where those fell short. That made the conversations at our booth feel less like introductions and more like a continuation of something already in progress.

Want to see how Marvix AI handles specialty documentation? We offer a 30-day free trial of Marvix AI, integrated with your EHR from day one, for your entire team.

Start a free trial
No items found.