In the age of AI, interoperability becomes core operating infrastructure

Healthcare interoperability is materially better than it was even three to five years ago, even if it is still far from seamless. The industry has made real structural progress. Hospitals are far more capable of sending, receiving, finding and integrating outside information than they were a decade ago. API-based patient access is now mainstream. Trusted Exchange Framework and Common Agreement has moved from being a policy concept to live exchange infrastructure.

That matters because it means the industry is no longer debating whether data can move at all, Zack Tisch, partner, portfolio services at Pivot Point Consulting, said. In many settings, it clearly can.

"At the same time, I would not describe the current state as true plug-and-play interoperability," Tisch said. "Once you move beyond basic record exchange and patient access and into real operational workflows – such as prior authorization, referral management, remote patient monitoring, quality reporting, digital front-door capabilities and third-party application integration – the experience becomes much more uneven.

"Health systems still rely on a mix of proprietary APIs, custom interfaces, point-to-point connections and manual workarounds to make important workflows function reliably," he added. 

"So, my high-level view is this: The foundation is real, the momentum is real and the direction is right, but the market is still in a transition phase. We are better connected than before, yet interoperability is not consistently easy, inexpensive or scalable across the enterprise."

Healthcare IT News spoke with Tisch to discuss the current state of interoperability, the challenges hospitals and health systems face – and how they can work to overcome them.

Q. Why do health systems still struggle to scale interoperability despite modern APIs and standards?

A. The short answer is standards solve part of the problem, not the entire problem. FHIR gives the industry a common technical language, which is a major step forward, but it does not automatically create a common operating model.

It does not solve governance, identity, workflow design, testing, security review, release management, data stewardship or contracting. It also does not guarantee that every organization exposes the same data in the same way or supports the same degree of write-back, eventing or workflow integration. That is why an organization can look modern on paper and still feel highly customized in practice.

What I see in provider environments is that scale usually breaks down at the handoff between the standard and the real-world operating model. To move an integration from pilot to enterprise capability, a health system has to align its EHR team, third-party vendor, security team, clinicians, operational owners, data governance stakeholders, and legal or procurement functions.

That is a lot of coordination, and every one of those groups has legitimate concerns. Even when the API itself is standards-based, the surrounding work can still feel bespoke. That is where the industry continues to run into special effort.

And the digital health side of the market still feels that friction. Companies continue to report barriers such as high fees, limited access to the data elements they actually need, incomplete testing environments, and uneven availability of standards-based APIs across products and use cases.

So, for me, the issue is no longer whether standards exist. The issue is how much customization and organizational effort it still takes to get those standards to produce dependable workflow value at scale.

Q. Why do financial pressures, regulatory alignment and implementation challenges continue to slow progress?

A. Because interoperability is being built in one of the most demanding operating environments health systems have faced in years. Hospital leaders are trying to modernize data exchange while also managing labor pressure, cyber-risk, denied claims, access challenges, enterprise AI priorities and tight margin expectations.

When hospital expenses rise faster than hospital prices, and organizations are spending billions nationally just trying to collect payment for care they have already delivered, every major initiative has to compete for limited dollars and limited attention. In that environment, interoperability is strategically important, but it is not the only urgent priority on the table.

The policy landscape also adds complexity. Directionally, I think the federal push has been very positive. HTI-1 raised the certification baseline to USCDI v3 as of Jan. 1, 2026. CMS's prior authorization rule is pushing payer API requirements largely to Jan. 1, 2027. TEFCA continues to mature, and information-blocking expectations remain part of the compliance backdrop.

More recently, HTI-4 added new provider-side certification capabilities tied to electronic prior authorization and payer interaction. The challenge is that these policies do not arrive as one clean, synchronized program. They touch different stakeholders, require different implementation paths and land on overlapping timelines.

Then there is the implementation reality. Even when the strategic case is clear, execution is cross-enterprise work. Health systems have to coordinate vendor readiness, governance, testing, workflow redesign, payer participation, change management and internal accountability.

None of that is trivial. So, when progress slows, I do not read that as resistance to interoperability as a concept. I read it as a reflection of the fact that organizations are trying to sequence multiple mission-critical initiatives at once in an environment where both capital and implementation talent are constrained.

Q. What is the growing influence of large EHR vendors as they expand into AI, CRM and other platforms? And what does this mean for innovation across the healthcare technology landscape?

A. I think the large EHR vendors are no longer just systems of record. They are increasingly becoming enterprise platform companies. Epic is embedding AI across clinician, patient and operational workflows, and it has also built out its own healthcare CRM capability.

Oracle Health is positioning a unified ecosystem that combines AI-enabled cloud infrastructure, interoperability, analytics and intelligent data platforms. When vendors at that scale expand beyond the core chart and into patient engagement, revenue cycle, CRM, analytics and AI, the center of gravity in health IT starts to shift.

That matters because Epic and Oracle together represent well over half of the inpatient EHR market, so their product direction has an outsized impact on what the rest of the industry prioritizes. When a dominant platform standardizes a capability – whether that is AI documentation, data-sharing functionality or consumer engagement tools – adoption can move much faster because health systems trust the governance model, understand the workflow implications, and can buy it through an existing platform relationship.

You can already see that platform effect in AI adoption patterns, where hospitals using the market-leading EHR vendor report materially higher use of predictive AI than hospitals using other vendors.

The upside of that influence is speed, scale and operational consistency. The tradeoff is that innovation increasingly has to happen inside those ecosystems or very deliberately alongside them. Startups still matter enormously, and in many cases, they continue to lead with the most innovative ideas.

But more of them have to integrate on platform terms, work within established governance and commercial models, and prove they can fit into existing enterprise workflows. For health systems, the real strategic question is not platform versus innovation. It is how to use the platform for scale and standardization without giving up the flexibility to bring in best-of-breed capabilities when they create differentiated value.

Q. In conclusion, what is the state of interoperability in the age of AI?

A. In the age of AI, interoperability stops being a compliance project and becomes core operating infrastructure. AI can only be as strong as the data environment around it. If data are incomplete, delayed, poorly normalized, missing provenance or disconnected from workflow, AI scales noise just as efficiently as it scales insight.

That is why the bar is getting higher. It is no longer enough for data to move eventually. For AI, data have to be timely, trustworthy, longitudinal and computable in a way that supports real decision-making.

We are already seeing that shift. Predictive AI is moving into mainstream hospital operations, and much of it is being deployed inside or directly alongside the EHR. At the same time, federal policy is tying interoperability more closely to standardized data, transparency and trust.

That changes the conversation. The question is no longer just, "Can data move?" The more important question is, "Can data move in a way that is complete, usable, governed and credible enough to support safe AI and measurable operational value?"

My bottom line is this: Interoperability is improving, and the direction is absolutely right, but AI is exposing the difference between exchanging data and making data usable. The organizations that will lead in this next phase will treat interoperability as enterprise data strategy, not just interface work.

That means standards, governance, architecture, workflow and trust all have to work together. In practical terms, the winners will not simply have more connections. They will have cleaner data, stronger accountability, better workflow alignment and a much clearer foundation for responsible AI.

Follow Bill's health IT coverage on LinkedIn: Bill Siwicki
Email him: bsiwicki@himss.org
Healthcare IT News is a HIMSS Media publication.