What could ChatGPT in Epic mean for healthcare’s AI middlemen?

ChatGPT and Epic might seem like two great tastes that taste great together…unless you’re one of the growing number of healthcare AI companies that have been banking on selling components of the combination separately.

In early September, OpenAI launched a new integration that allows Epic users to connect their EHR with ChatGPT for Healthcare, enabling clinicians to conduct AI-assisted searches, synthesize disparate information, and perform other tasks without manually combing through authorized patient data.

It’s a big step toward potentially creating a much more conversational and powerful interface into the EHR – and that has big implications for both buyers and sellers in a burgeoning startup market that has focused on enabling exactly that functionality.

What new features are coming to Epic users?

The announcement from OpenAI and Epic contains several different elements, starting with the ability to integrate ChatGPT into supported Epic workflows.  

ChatGPT will be able to use the patient record as context for natural language queries, allowing clinicians to ask questions about recent changes to clinical history, such as new labs, medications, or notes. ChatGPT can synthesize the relevant information and then point the user back to the relevant data in the chart so the user can double check or explore further.

In some Epic instances, these features will be accessible directly in the EHR layout, so users won’t have to leave the chart to access the AI assistance. 

The key limitation is that the integration is read-only. Users cannot currently use ChatGPT to write into the record, place orders, or take any other action. It’s acting as an enhanced research assistant, not a scribe.

Complementing this integration is a new Healthcare Public Data plug-in, which gathers together connectors to nine official public healthcare sources, including ClinicalTrials.gov, CMS Coverage, RxNorm, DailyMed, and PubMed. 

“Teams can work with specific records, fields, identifiers, and versions while focusing the task on these authoritative sources. This makes it easier to compare and verify precise information such as trial eligibility criteria, medication identifiers, coverage policy versions, and provider records without searching each source separately,” OpenAI says.

However, the public data plug-in is separate from the chart search functionalities, meaning that users cannot synthesize actual patient data with these public data searches to conduct patient-specific, personalized mega-searches.

Doesn’t Epic already have its own AI interfaces into the chart?

Epic isn’t new to AI by any means, nor is the act of bringing ChatGPT into the fold likely to supersede the major investments the company has been making in native AI capabilities.  

Earlier in 2026, Epic said that more than 85% of its customers were active users of existing AI features, including Art for clinicians, Penny for revenue cycle management, and Emmie for patient-facing communications.

And the recent User Group Meeting brought even more AI tools, including an intelligence platform called Ergo and a new Agent Factory.

The difference is that the ChatGPT option represents another way into the data ecosystem: an independent-yet-integrated interface that offers an alternative to a fully native Epic AI experience.

That could be viewed as a notable departure from Epic’s insular, walled garden reputation.  And for startups whose main proposition was “we can do what Epic doesn’t easily let you do, which is allow you to work with Epic data without relying solely on Epic’s internal tools,” it could be a big blow to the pipeline.

What does this mean for healthcare AI vendors?

With ChatGPT on the scene, some AI vendors are going to be facing a new challenge to their pitch decks.  If “us or Epic” morphs into “us, Epic, or ChatGPT,” it might get more difficult to prove their value proposition to health systems, especially when CIOs are increasingly looking for ways to slim down their tech stacks and only focus on the AI capabilities that can give them the biggest bang for their buck.

As Hackensack Meridian Health CDIO Joel Klein, MD, said in a recent interview with Digital Health Insights, “We want to use all four corners of these [enterprise-level] solutions that we’ve already bought and paid for.”

“These smaller companies coming to us with the message that their products are better – and yes, sometimes their solutions may be incrementally better in certain dimensions,” he continued. “But if we adopted every single point solution using just that logic, we’d end up with this jumbled garage sale of stuff that’s all duct taped together.”

That doesn’t mean that all AI startups are under a similar level of threat from this announcement. The newly introduced capabilities are limited in their scope, and certainly don’t solve all the problems that EHR users are currently facing.

The biggest risk is likely to come to vendors whose main proposition is data summarization, retrieval, and presentation. Companies that focus on asking questions about the chart, summarizing longitudinal records, or surfacing relevant patient history within the workflow are most likely to find themselves under additional pressure during the sales cycle.

On the other hand, the billion-dollar ambient scribe segment may not start feeling as much of a pinch.  The read-only limitation of the integration means that ambient scribe systems – already in use in more than 62% of Epic hospitals – might not find their contracts under immediate threat by this announcement. 

Clinical “copilots” that offer some type of specialized reasoning or workflow function may find themselves somewhere in the middle. They will need to clearly demonstrate to decision-makers like Klein why they’re worth the additional investment.

Vendors who fall into the higher-risk categories might consider a different pitch: positioning themselves as “bridge” or “transition” partners that can help health systems fill the gap between current capabilities and embracing a fully Epic-native or Epic/ChatGPT-enabled future.

It’s something that CIOs are actively looking for, said Klein, especially as health systems wrestle with the inertia of extremely large platform vendors like Epic.

“If we wait for a handful of platforms to grind through their internal machinery, finally release something, and then go through a handful of versions of it before it’s really ready for primetime, we might be missing a window of opportunity to iterate and innovate in an area that really needs attention right now,” he said.

“Bridge vendors help us stay at the forefront of what we want to achieve without committing to a flea market of tools forever. If we can be intentional about it, and have clear expectations around what we’re bringing into the system and why, then we can have the best of both worlds so we can better serve our patients.”

A bombshell or a bump in the road?

Big names like Epic and OpenAI tend to produce a sense that the market will imminently be turned upside down, but that’s not usually the case in the healthcare industry.  

There’s certainly a ravenous appetite for generative AI capabilities, with 72% of physicians now using AI in their practice somehow. Epic users tend to be “early adopters and fast followers” compared to users of other EHRs, according to one recent study. With Epic’s massive market share, the sheer volume of interested users could result in relatively swift uptake of ChatGPT’s new offerings.

However, health system infrastructure doesn’t just turn on a dime. There aren’t clear, publicly available numbers on how many health systems have adopted ChatGPT for Healthcare or a ChatGPT Enterprise license yet. ChatGPT for Healthcare, for example, is brand new to the healthcare industry, having only launched in January of 2026 with just eight partners.  

An established relationship with ChatGPT has to be the first step before taking advantage of the new integration, which itself is simply one new option for the industry to do something that plenty of other vendors are vying to do.

The bottom line is that the proposition is intriguing, and similar vendors are going to need to raise the bar on what they offer their potential clients. But they probably don’t have to start packing up their offices just yet.  

Competition is likely to remain strong in a market where straight-up functionality isn’t the only differentiator for decision-makers. Personal relationships, privacy and security, and native healthcare experience all factor strongly into choosing a solution, and there are still plenty of vendors out there who may be able to out-maneuver even the largest companies in some of those areas.

For now, the announcement has the potential to cause waves throughout the vendor landscape, but there’s no guarantee it’ll result in widespread change. For vendors currently acting as AI middlemen, the challenge will be preserving their niche without getting squished by the changing forces from either side. 

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].