Microsoft expanded its Dragon Copilot artificial intelligence clinical assistant to nurses, designing its ambient AI technology specifically for unique nursing workflows.
The tech giant codeveloped the ambient-enabled AI capabilities along with nursing leaders and front-line staff from leading organizations such as Mercy and Advocate, according to the tech giant. The solution also will be available in Epic’s Rover app.
The company also announced updates to enable the integration of partner AI apps and agents directly into Dragon Copilot.
Microsoft unveiled the Dragon Copilot AI assistant for clinicians at HIMSS 2025 in March, billing it as an all-in-one technology that combines voice dictation, ambient listening and generative AI. The tool combines the natural language speech recognition capabilities of Dragon Medical One with the ambient listening of DAX Copilot.
The nursing solution streamlines documentation by recording nurse-patient interactions and integrates those conversations into flow sheet documentation that’s automatically filed in the electronic health record. Nurses can then preview and edit the documentation for accuracy. Nurses also can use the solution to access medical content directly in their workflow, reducing tab-hopping, according to Microsoft executives.
The Dragon Copilot AI assistant also takes on routine tasks to help draft notes and summarize patient interactions, which helps reduce clicks and close the loop faster.
“This was natural for us, as we have decades of experience supporting radiologists and physicians with productivity tools,” Mary Varghese Presti, corporate vice president, health and life sciences at Microsoft, told Fierce Healthcare.
Microsoft’s Dragon Copilot tool for nurses was designed for nurses’ specialized workflow and documentation style after spending time shadowing bedside nurses.
“Nurses document in flow sheets. A physician can have a conversation; it’s a narrative. The technology for supporting nurses is very different because you have to identify, from the conversation, very discrete data elements that have to go into very distinct fields on a flow sheet or even multiple flow sheets,” said Varghese Presti, who started her career as a pediatric nurse at Johns Hopkins Hospital before moving into health tech.
Nurses represent the largest workforce in healthcare, and 65% of nurses report high levels of stress and burnout, a study found. More than a quarter of a nurse’s shift is consumed by documentation and administrative tasks, opening up opportunities to use AI and technology to alleviate some of that burden.
“That 25% of time on documentation is 25% of their shift not spent on patients. And they went into the profession to be patient-facing,” she said. “The more we sent teams in to observe and shadow, the more we appreciated the depth of the challenge, and how might we look at the jobs to be done as a nurse on a unit. How might we bring technology and design in a way that really augments the human capacity in that role?”
As nurses adopt AI ambient technology, it does require some change management to shift to a “care out loud" approach, as Varghese Presti refers to it.
“When I was nurse, I'd go in and I’m clocking a lot of things at once. I'm clocking that the food has not been eaten. I'm talking to the patient and saying, ‘From a scale of one to 10, what's your pain?’ I'm checking to see if they've gotten out of bed. These are all things you document, but I’m not saying them all out loud,” she said. “Care out loud does require a bit of engaging with the patient. What's interesting is that the nurse leaders are super excited about this because it is bringing the patient into that dynamic and it’s actually engaging the patient far more, and it's allowing the patient to have a lot more insight about what's going on in that stay as well.”
Ten health systems partnered with Microsoft to develop the tool and have deployed it as early adopters.
Over the course of a shift, a bedside nurse could enter a patient’s room 20 to 40 times, and the technology is capturing information from all those interactions.
“There's a level of insight about what's happening for that patient that never would have been captured before,” Varghese Presti said. “I'm really excited to see the knock-on effects as it scales and if we can really show what happens between a nurse and a patient, the value, the outcomes, the quality, the depth of the interaction, the quality of care and support that's actually happening on a shift."
This builds on industry efforts to effectively measure the value of nursing care, specifically acute and critical care nurses, as with the Blueprint initiative.
The ability for nurses to document in near real time using ambient AI also could help clinicians flag issues sooner and improve clinical outcomes, she noted.
Often, nurses will take care of several patients on the floor and then document an hour or two later. The AI clinical assistant accelerates documentation.
Varghese Presti cites an example at one of the early adopter health systems when a neurologist placed an order minutes after a nurse entered data into the flow sheet. “The neurologist realized that there was an action that he wanted to take. You have these knock-on effects where you detect patterns or are compelled to move to action on something faster than you normally would,” she said.
“We could have started someplace else, something that's a little bit easier. I don't think any reasonable person is going to enter a space by picking the hardest problem to solve first. Actually, that goes against all the principles of software development. But we really felt very convinced that if we were going to spend our time on this, we needed to be able to actually mitigate this really big pain point for nurses,” she said.
Microsoft's second announcement, around new partner capabilities, will make AI innovations available to Dragon Copilot customers, Microsoft executives said.
Microsoft’s partnerships include Elsevier, OpenEvidence and Wolters Kluwer UpToDate for medical references within Dragon Copilot. Other partners include Atropos Health, Canary Speech, Lightbeam Health Solutions, Panagaea Data and Press Ganey. To streamline revenue cycle management, Microsoft also partners with Ensemble, Regard, Cohere Health, Humata Health, Rhyme and RhythmX. The company’s partnerships with Artisight and hellocare.ai expand smart hospital and virtual care capabilities.
“At Microsoft, partners are in our DNA. We really are a partner-first company. When you think about what's going on in healthcare and what we all want to be supporting in terms of innovation, there is such a barrier to entry to get in and actually get to the point where you can actually drive value. It’s really about the ecosystem,” Varghese Presti said.
“I don't think a few point solutions is going to be enough. It’s been paramount to us in our strategy to ensure that we extend the value of Dragon, make it an extensible platform,” she added.
Baptist Health is exploring how to integrate Canary Speech’s vocal biomarker analysis tech into its clinical workflow through the Dragon Copilot integration.
“The ecosystem Microsoft is creating with Dragon Copilot is essential for scalable innovation in healthcare,” said Brett Oliver, M.D., chief medical information officer for Baptist Health, in a statement. “Rather than relying on isolated point solutions, we’re able to adopt multiple tools—like Canary Speech—within a unified, ambient workflow. This approach allows us to advance diagnostics and care delivery without disrupting the clinician’s experience.”
"These latest updates broadly support Microsoft’s approach to healthcare AI solutions," Varghese Presti noted.
“At Microsoft, we look at this as more about, how do we help the healthcare industry become much more frontier? With ambient scribing, these are very early viral use cases for generative AI, important ones, but it's just the tip of the iceberg,” she said. “We look at it as, how do we enable clinical intelligence on tap? And that means moving through ambient to generative AI to agentic AI.”
The next level of healthcare AI—agentic AI—opens up opportunities for more sophisticated capabilities.
“We’re looking at what are those things that agentic services can do so that you can actually return the human being to clinical nuance, physical touch, judgment. We really look at it much more expansively around intelligence,” she added.