This is the third in a four-part series examining the critical strategic dilemmas healthcare executives must navigate in 2025. Read part one and part two.
Epic's AI pipeline is hugely ambitious, covering 100+ projects across clinical documentation, revenue cycle optimization, and patient engagement. But many of these capabilities remain in early stages of deployment, with limited scalability, while specialized vendors are already delivering mature solutions that integrate with your EHR.
This creates healthcare's third fundamental AI dilemma in 2025: Do you wait for Epic's roadmap to materialize, or do you partner with specialized vendors to move faster?
(Before we dive in, a few disclaimers: First, while we focus here on Epic as the most common EHR across AI Catalyst members, the strategic trade-offs are similar if you use a different EHR vendor. Second, as with all of the dilemmas presented in this series, you'll likely need to apply some combination of “Epic-first” and “best-in-class” approaches. However, limited budgets, attention spans, and capacity force tough choices about your emphasis.)
The case for ‘Epic-first’: Trust and maximize your biggest tech vendor
It’s tempting to simply align your AI strategy with Epic’s AI roadmap – and there are plenty of compelling reasons to do so.
Epic's healthcare expertise may translate to safer, more reliable AI implementation. Epic understands clinical workflows deeply, they know your organization intimately, they've tested solutions across thousands of facilities, and they have significant insight into healthcare data security and stakeholder needs. All that experience makes many organizations far more comfortable betting with Epic than with an unknown new entrant.
Out-of-the-box integration arguably matters more than cutting-edge capabilities. Many health systems prefer seamless workflow integration over the absolute latest features. This preference reflects both immediate workflow benefits and a strategic bet that today’s feature gaps are only temporary. As one CMIO explained, "Epic might only be 80% as good as the niche vendors [but] give them a year and they'll get there."
Having fewer clinical AI vendors strongly correlates with better outcomes. Our recent AI Use Case Audit
revealed a striking pattern: health systems rating their clinical AI implementations as effective average just 3.6 vendor partnerships, while those rating them ineffective average 5.4 vendors. This suggests that making concentrated bets — such as going all-in with Epic — might be the right approach for clinical AI. Plus, many health systems acknowledge they haven't maximized their existing Epic functionality, so your easiest path forward may be to tap Epic’s potential before venturing elsewhere.
Using Epic lets you leverage rapid learning across the company's massive customer base. In AI, more data usually translates to smarter, more reliable results – so with hundreds of health system clients and 260 million patient records covering 80% of American patients, Epic has an inherent advantage as it develops AI capabilities.
The case for 'best-in-class': Act faster with better tools
Despite Epic's advantages, specialized vendors are delivering solutions today that address specific needs with greater sophistication. If you’re facing urgent challenges, these focused solutions may offer a better fit.
Time spent “waiting for Epic” can represent lost opportunity. While Epic builds out its roadmap, specialized vendors are often already offering similar tools. Further, health systems report difficulty determining whether Epic’s upcoming features will initially launch as limited pilots or full organizational rollouts, and whether they’ll require additional payments. As one Chief Digital Health Officer told us, "Waiting for Epic? I just don't think we can do that."
Specialized institutions may need specialized solutions. Academic medical centers and specialty hospitals, in particular, report that Epic's AI solutions aren't optimized for their distinct needs — a limitation that Epic itself generally acknowledges. “We are seeing not as much utility [with Epic's AI tools],” one CMO at an academic medical center explained. “It is having a hard time creating a useful response for my patient population when it's a single model shared between specialties.”
Many specialized vendors prioritize ROI tracking, while Epic largely doesn't. This difference becomes especially important when health system leaders need to demonstrate clear value to justify AI investments: Epic's reticence to measure ROI can undermine your ability to build a business case. As one AI Catalyst member observed: “When you can get [and measure] hundreds of thousands of dollars in a 45-minute timeframe, that is a tangible ROI that's hard to argue against.”
Post-pilot scaling often requires capabilities beyond what Epic provides. Organizations that successfully scale beyond pilots report that they typically build their own implementation capabilities rather than relying solely on Epic. This pattern recalls challenges many systems experienced with previous Epic rollouts, where staff often felt they carried much of the implementation burden.
Which should you choose: Epic-first or best-in-class?
While there's no one-size-fits-all answer, here’s the decision-making framework we’d recommend. Consider taking an “Epic-first” approach if:
Your health system has clinical and administrative needs that are standardized across sites and relatively similar to those of other Epic customers.
You have low tolerance for AI vendor risk or limited IT integration capabilities.
You simply can't afford to pay other vendors in addition to Epic.
In that case:
Review Epic's roadmap frequently and prepare for rapid adoption. Epic regularly releases new AI features, so make sure you’re ready to evaluate and deploy promising tools quickly.
Build your own AI governance and ROI measurement frameworks. Bear in mind: Epic will provide access to AI tools, but generally not the structures to govern them or measure their impact.
Consider providing "sandbox" tools where Epic doesn't yet have solutions. While off-the-shelf tools like Microsoft Copilot can’t solve every healthcare AI use case, they can offer a relatively controlled environment for staff to experiment with some use cases not yet supported by Epic.
On the other hand, consider taking a “best-in-class” approach if:
You have specialized needs or unique patient populations that require tailored solutions.
You simply can't wait for Epic's development timeline.
You have strong technical integration capabilities and vendor management processes.
You can identify vendors whose tools will pay for themselves in the next few years – before Epic’s solutions are fully mature.
In that case:
Negotiate flexible contracts with clear exit strategies. Include opt-out clauses or shorter contract terms (12-24 months) that allow you to pivot if Epic eventually delivers a better solution.
Look beyond surface-level “Epic integration” claims. Not all vendors that advertise Epic integration offer the same depth of workflow incorporation.
Bear in mind: Even if you don’t choose Epic now, you may pivot later. Ensure your vendor contracts address data ownership and portability, so you can migrate content and configurations to Epic solutions if you later switch.
So how are real-world health systems navigating this dilemma? When we posed the "Epic-first vs. best-in-class" dilemma to health system CEOs and board members at The Health Management Academy's recent Trustee Summit, 69% favored the "best-in-class" approach.
That attitude differs from what we've heard from other C-suite leaders: CIOs have told us they'd rather utilize Epic tools wherever possible to avoid integration nightmares, while CFOs have made clear they’d like to avoid an endless parade of new vendor contracts.
But it appears your system's top executives may be aiming higher: They might want you to find the best AI solution – even if it's painful or expensive to launch.
Questions to consider:
What framework will you use to determine whether the urgency of adopting a particular AI use case
now outweighs the benefits of waiting for Epic's own offerings?
In what areas do you feel Epic is least likely to develop effective AI solutions for your institution's specific needs?
How can you structure vendor contracts to maintain flexibility as Epic's AI roadmap evolves?