Medical record companies are racing to integrate artificial intelligence into multiple parts of their software, hoping to attract customers by simplifying a cumbersome user experience and keeping pace with competitors. In recent months, several large electronic health record (EHR) vendors, including Epic and Oracle, have announced or accelerated their modernization plans. The companies say their goal is to boost productivity amid rising costs in the healthcare system, while improving job satisfaction for doctors and nurses, enhancing patient care quality, and advancing medical technology. This shift marks a fundamental change for software often viewed by healthcare providers as a major source of frustration.

"Doctors don't want AI to play the role of a doctor and replace them," said Leigh Burchell, president of the Electronic Health Record Association. "What they really want is for AI to help with administrative burdens or reduce the time spent on documentation after visits. So, these are exactly the areas our members are most focused on."

The intersection of medicine and technology

EHR software enables clinicians to electronically access and update a patient's complete medical history, including diagnoses, lab results, and medications. More than a decade ago, U.S. healthcare providers were required to adopt EHRs to continue receiving full funding from federal programs such as Medicare and Medicaid. Software usage subsequently soared, with 78% of physicians and 96% of hospitals using EHRs in 2021, according to government data.

However, doctors often express dissatisfaction with EHRs. These systems can be difficult to use, filled with pop-ups and multiple screens, requiring clinicians to spend significant time entering and finding data. Another common criticism is that EHRs are designed more for billing than for clinical decision-making. Many doctors say these systems therefore distract from patient care.

Major EHR vendors have long claimed to be working to make their products simpler and easier to use. Recent advances in AI have significantly accelerated this process: for example, generative AI, which can understand and create original text and images; ambient listening technology, which combines microphones and machine learning to generate real-time records of clinician-patient encounters; and agentic AI, which can complete complex tasks without much human oversight.

In the past, using an EHR required extensive data entry and "one screen after another," said Girish Navani, CEO of EHR vendor eClinicalWorks. "If you look at the last two years, the emergence and rapid rise of generative AI has made many old paradigms obsolete."

Epic, the largest U.S. EHR vendor, "has been working to integrate generative AI throughout the software," said Seth Howard, Epic's executive vice president of research and development. The Wisconsin-based company holds more than a third of the hospital market share, followed by Oracle Health (22%) and Meditech (13%), according to data from health analytics firm Definitive Healthcare.

Epic says more than 3,100 hospitals use its system. According to Howard, about two-thirds of its customers are already using the vendor's generative AI products, including algorithms that help doctors write notes and summarize patient information for clinicians before visits.

Last year, Epic integrated ambient listening software from startup Abridge into its workflow, but Howard said the company is working to make its software compatible with more products that can transcribe patient visits in real time. The company is also working to make its documentation tools multimodal, so they can understand video, images, and other types of information beyond text. Over the past year, Epic has also begun building frameworks around agentic AI in its software to explore how digital assistants capable of summarizing information, interacting with patients, scheduling follow-ups, and more can work together. Overall, Epic has 125 AI use cases in development or live.

"We are investing heavily to ensure we innovate quickly and do it right at the same time." — Bharat Sutariya, Chief Health Officer at Oracle Health

Similarly, Oracle announced in October that it plans to release a new EHR platform this year integrating its clinical AI agents, featuring voice-activated navigation and search. The clinical agents can record and transcribe patient visits and draft notes in the EHR for clinicians to approve, the company said Tuesday, and are currently available for more than 30 medical specialties. Oracle Chief Health Officer Bharat Sutariya called the update "nothing less than a complete reinvention of the EHR." "We are investing heavily to ensure we innovate quickly and do it right at the same time," he said.

Meanwhile, Meditech and eClinicalWorks (a company providing cloud-based EHR software currently used in 70,000 care settings) are taking a collaborative approach. Meditech has long partnered with Google to accelerate clinician access to data and with Microsoft to automate documentation, including deploying Microsoft's ambient listening and clinical documentation tools into its EHR. Meditech also plans to incorporate Microsoft's latest clinical assistant when it launches this year, according to Rachel Wilkes, enterprise lead for Meditech's generative AI initiative. Through software integrations, Meditech customers can also use a variety of other AI documentation tools.

"We want to keep our existing customer base engaged," Wilkes said, "but 'yes, we also want to attract new customers with an innovative, scalable platform that can truly replace other competitors.'"

Screenshot of the welcome page for Microsoft clinical assistant Dragon Copilot User view of Microsoft clinical assistant Dragon Copilot.

eClinicalWorks is also partnering with Microsoft to use machine learning to match patient records, and with a company called Healow to create an AI medical scribe and an AI contact center.

Checks and balances for AI

Despite ongoing concerns about quality, privacy, bias, and a lack of federal regulation, EHR vendors are still reshaping their software with AI. Generative AI, in particular, is difficult to monitor. These models are known to hallucinate, providing factually incorrect or irrelevant answers, or omitting information entirely. Their performance can also degrade over time, a phenomenon known as model drift.

Nevertheless, record companies say they, along with customers, are carefully overseeing AI to ensure it does not fabricate any information or omit anything. Leaders at eClinicalWorks, Meditech, and Epic say they have invested heavily in risk management and established strict guardrails, including ongoing evaluation by clinicians. Meditech's Wilkes said there should always be human oversight in ambient documentation and generative AI products.

Meditech also adds a disclaimer to any record created by AI, stating it was generated by an algorithm and reviewed by medical staff. The company is working to introduce more formal citations in workflows to clarify what was generated and where the model pulled relevant information, Wilkes said. Epic applies the same principles for ensuring AI quality as it does to any other software, to ensure it performs well and does not cause harm, Howard said. But "the techniques used to understand 'whether it's performing well' may be different," Howard said.

Despite concerns about the models, vendors say hospital customers are still pushing forward with adoption. According to a survey conducted by HIMSS in the fall, 86% of respondents are already using AI in their healthcare organizations, primarily for administrative tasks, but interest in more clinical use cases is growing.

"Our customers are very enthusiastic about this. Even more so over the past 12 months," Wilkes said. "We've moved from a lot of educational influence to — now the question is, 'when? When can I get it? What's the cost? What's the plan?'"

"Technology always has early adopters, mid-stage adopters, and laggards. It was true for smartphones, it was true for color TVs, it's true for every product," said eClinicalWorks' Navani. "What I can tell you today is that we're seeing a remarkable growth in the number of customers we call early adopters." But "it's not just because the modern is new and good," Navani continued. "It clearly has an efficiency advantage. That's what's driving the change."