How is health IT leaders’ increasing reliance on Epic shaping their thinking about working with startups and other vendors? Global venture firm Redesign Health recently surveyed 112 senior executives at Epic-based health systems to understand how they’re thinking about Epic, AI and their willingness to buy from outside vendors. Redesign Health Director Kienan O’Brien responded via e-mail to Healthcare Innovation’s questions about the survey. 

In the survey, 71% of health system leaders describe their organizations as “Epic-first,” yet the research also identifies openings for companies that can outperform Epic in specific areas. The report is worth reading in full, but we asked O’Brien to expand on a few key points. 

Healthcare Innovation: Your report notes that the window for competing with Epic is narrowing. Is Epic itself also expanding out into new areas it hadn’t entered before — like ERP, for instance? 

O’Brien: I think you’d be hard pressed to find a company that ostensibly fits the “system of record” category trying to do more in the AI age than they were pre-AI. That now includes going after territory of other “systems of record,” which is pretty incredible. Right now if you’re playing exclusively within the data silos of the health system outside of truly deep clinical intelligence, assume Epic will come for you. Confidence is very high that Epic will be able to execute, but I still think in a world like ERP there should be a lot of waiting and seeing from potential customers.

HCI: Is the AI scribe space an example where there was a window and companies like Abridge took advantage. But now perhaps those tools will become commoditized and the customers will find Epic’s tool just as easy to deploy and have one less vendor to deal with? 

O’Brien: We saw this happen pre-AI as well. Epic came out with a surgical block optimization module that immediately encroached into what Qventus and LeanTaaS were doing and changed how CIOs approached those contracts because of exactly that. The survey would indicate leaders’ view is that fewer tools are better. Epic has benefited at every turn from “commoditization” of technology and that seems to be continuing in this new era. 

However, it’s important to note that Qventus and LeanTaaS haven’t gotten crushed by Epic. They built on their incumbent advantage with health systems to develop new capabilities and continue to provide more and more value. Startups should take lessons from them both in what they can expect from how Epic will react to you and how you can continue to win.

HCI: How can vendors that gain that early advantage in a new area hold on to those customers?

O’Brien: Speed is still your best friend. Assume Epic is coming for you even if there hasn’t been an announcement yet or a product they have that matches what you can do. Interact with the world outside of Epic, and ideally the health system. Go deeper into true clinical intelligence, and work with new data types Epic doesn’t process, like computer vision as a few examples of more durable strategic capabilities.

HCI: Although there are advantages to working with a single vendor for a lot of their IT solutions, do the health IT leaders recognize a downside to having all their eggs in one basket, so to speak?

O’Brien: The downside to locking yourself into the Epic ecosystem would be that you won’t be able to take advantage of the amazing other tools that will come out that are “better” in terms of ease of use, ROI, etc. On the clinical side, I don’t think healthcare has historically had a lot of health system workflow-related tools that blew away what Epic would eventually build to begin with. They might have been “better,” but frequently not worth the headache of dealing with another vendor, integration, or god forbid making clinicians switch between screens. At the moment, Epic is seizing on the historic perception they’ve cultivated to say it’ll be the same for all these AI applications. There might be some differences between Epic and an external vendor, but like we’ve seen in the past, it beats the perceived headache of dealing with an outside vendor.

ERP-type expansion is a different story. Early system users of this platform seem to be solving for a data siloing problem where Epic was over here, SAP over there, Workday somewhere else, financial systems there, etc. I’m not sure centralizing all your data within the walls of Epic is the right way to address the data fragmentation issue. Especially if you’ll then have to rely on them building new applications and workflows in a space they aren’t experienced in, which you would likely have to use over anything else. There are other ways you can solve the data problem. But Epic is getting out in front while they have health system trust before other players break down those data silos and maybe leaders start asking what they can do now that their data is accessible under one roof.

HCI: Your survey response had 52 percent for-profit community health systems, 40 percent from non-profit health systems, and only 8 percent from academic-affiliated ones. Could the propensity to work with startups or to develop solutions internally be higher in those academic-affiliated systems?

O’Brien: It’s interesting because we work with amazing academic-affiliated system partners that are very clearly leaning into how they can partner, buy from, or launch new startups, but the numbers in this survey (mind, we’re getting into smaller sample sizes now) indicate that the non-profit community hospitals are actually the most willing to work with newer startups of <3 customers in any capacity.

HCI: Your report notes that the health IT leaders say that AI solutions are likely to eventually transition to centralized budgets, potentially shifting purchasing decisions away from individual departments. Will this impact on the CIOs’ and CMIOs’ roles? And will this make it harder for startups to get their foot in the door?

O’Brien: CIOs still need to lead the work of bringing in all these vendors coming through more requests from department leaders, and manage how to start building more of their own tooling. At the moment, it’s a very hard job because their pipeline of IT projects, integrations, and vendor assessments has never been bigger. Startups might have an easier time getting a foot hold with department leaders who have more autonomy to get through contracting, but long timelines to deployment aren’t going away.

HCI: Does this report suggest that venture capital firms and venture studios need to work even more closely with health system partners that identify pain points for startups to work on — and then help co-develop them?

O’Brien: As AI capabilities continue to mature, the bar is higher than it has ever been for what is a defensible venture-backable idea. It’s not just what Epic will build, but now you have to consider what the health system itself will be able to build in the future. What will the frontier labs continue to develop inside of these health systems they are forward deploying into? 

Redesign’s goal as a venture capital technology firm is to back incredible founders pursuing venture-backable businesses. One of the ways in which we do this is by helping our health system partners in this new world to triage opportunities across the different build, buy, partner, or company launch pathways. It helps both us and our partners; it lets them quickly triage commercial opportunities and we get fantastic exposure to what really matters to health systems and understand deeply what actually is a venture-backable business.

HCI: Why do you think imaging AI and enterprise imaging add-ons tops the list of areas for external vendor competition? 

O’Brien: I think it’s partly rooted in how different radiology workflows have always been that left opportunities for startups to get traction with health systems even pre-AI boom (e.g., Rad AI, Sirona Medical). Plus radiology is probably the single largest area of clinical AI development for actually identifying and diagnosing diseases and risk factors that is not something Epic is doing, and it historically has stayed away from anything that brushes up against FDA regulation.

HCI: Anything else about the survey results that jumped out at you or surprised you?

O’Brien: Not surprising, but integration is the big buzzword still for why health systems have a hard time working with external vendors. It’s not just a technology problem, but a process problem. You hope this goes away given how much easier it should be today to integrate with systems exploring MCP frameworks or even just agentic integration tooling to replace legacy processes. We aren’t there yet, but if we could start reworking  those processes, this survey might look very different in the future.