By James Henderson, SVP & GM, U.S., Orion Health.

Insights following the 2026 ASTP Annual Meeting 

Prior to the 2026 ASTP Annual Meeting, we published a perspective on the evolving state of U.S. health IT interoperability and the shift from regulatory alignment to operational execution. After my time in D.C., this shift is no longer theoretical. It is now the central challenge facing health systems, states, and technology partners. 

Across sessions and conversations, leaders kept coming back to the same issue: how to operationalize interoperability, data governance, and emerging AI expectations at scale within real funding, workforce, and accountability. 

The policy scaffolding is largely in place. The pressure now is on execution. 

Policy to Practice Gaps: Examples 

If there was one phrase that kept surfacing throughout the meeting, it was this: compliance doesn’t equal usability.  

Interoperability, price transparency, and AI acceleration are not new concepts. What’s changed is scrutiny. Regulators, providers, and patients are asking a simpler question: does it actually work in everyday care? 

One example that stuck with me: patient-facing apps may technically connect via APIs, but response times can stretch to 20 minutes. From a standards perspective, that may count as interoperability.  From a patient perspective, it’s a failure. If data isn’t available when it’s needed, it might as well not exist. 

Another example discussed was the lack of clean workflows for basic care coordination. An urgent care visit still does not reliably notify a patient’s primary care physician automatically. Parents who want to share a controlled and permissioned summary of a child’s allergy history with a caregiver encounter friction. These are not rare edge cases. They’re everyday workflow realities. 

We have electronic systems and standards. What we don’t consistently have is performance, workflow alignment, and accountability for real-world usability. That’s the gap the industry now needs to close. 

Governance in the Real World: Models and Risk Realities 

AI was a central theme, but not in the way many expect. The debate wasn’t about model capability. The constraint lies in system readiness. 

Panelists repeatedly highlighted four blockers for clinical AI adoption. These included: 

  • Fragmented and incomplete data access. 
  • Regulatory uncertainty for patient-facing or clinical-decision AI.
  • Reimbursement models that don’t reward prevention and avoided events.
  • Trust across patients, providers, and institutions.

Several panels moved beyond one-time model validation toward lifecycle monitoring that detects bias, drift, and real-world performance issues after deployment. Risk-based oversight was emphasized, with a focus on regulating where clinical risk is real while avoiding unnecessary slowdowns in use cases where it isn’t. 

One concept introduced was a “supervisory agent”, an additional AI layer designed to continuously audit and monitor clinical AI models. Whether that specific construct becomes a reality remains to be seen. The barrier won’t be the model’s capability, but the system’s inability to create durable trust at scale. 

Where Momentum Is Real and Where It Isn’t 

Not everything discussed was theoretical. There are areas where operational traction is visible. 

TEFCA was described as moving from aspiration to backbone, with exchange volumes accelerating. The expectation from leadership is shifting as well: interoperability must be fast and patient-centered, not just technically compliant. As TEFCA participation expands, performance and transparency expectations will increase. 

CMS’s voluntary accelerator model reflects an iterative approach: test, learn, and refine in the open. That approach differs from the more structured governance model of TEFCA, but the two were presented as complementary levers within the same ecosystem. 

At the same time, foundational gaps remain. 

Provider directories are still inconsistent, expensive to maintain, and often unreliable. API performance expectations, like latency, are not yet consistently enforced. Reimbursement structures continue to lag in prevention-oriented innovation. 

The direction of travel is clearer than it has been in years. Execution at the operational layer remains uneven. 

What This Means for States and Ecosystem Partners 

For organizations that have already built statewide exchange infrastructure or scaled national interoperability programs, the next phase is less about adding new interfaces and more about ensuring that existing infrastructure preforms reliably at scale.  

As participation grows, expectations around latency, governance, and transparency increase. Sustaining trust depends on whether those systems continue to operate consistently under real-world pressure. From my perspective, three implications stand out: 

First, participation shapes outcomes.  

It was emphasized that engaging early through pilots, workgroups, and comment processes is materially influencing how these frameworks get operationalized. If you’re not at the table, you’re accepting someone else’s definition of “good enough.” 

Second, interoperability must evolve from connectivity to performance.  

Exchange volume alone is no longer the success metric. Latency, onboarding burden, pricing fairness, and operational transparency are becoming part of how interoperability is defined. It’s more demanding, but it’s the right one. 

Third, AI scalability requires incentive alignment.  

We can demonstrate predictive accuracy all day long. If reimbursement models do not reward avoided events or upstream prevention, those tools remain stuck in pilots. Technology advancement and payment reform have to move together. 

Across all of this, one reality persists: aligning policy goals with real-world implementation requires coordination across states, vendors, regulators, and payers. No single entity can operationalize this shift alone. 

Moving From Direction to Delivery 

What stood out most at ASTP wasn’t a single announcement or rule change. It was the consistency of the message across panels and discussions.  

We are past the digitization phase. The next phase is operationalization. 

Policy has created direction. Standards have created structure. But unless those expectations translate into fast APIs, reliable directories, transparent governance, aligned incentives, and workflows that don’t add friction, clinicians and patients will continue bridging system gaps manually. 

The destination is broadly agreed upon. The harder question, and the one now squarely in front of us, is whether we can make it work at scale in real systems, for real people, in real time. 

That’s the shift underway. And it’s where real work begins. 

James Henderson
SVP & General Manager, U.S. at Orion Health
Read more from James Henderson