If you listen to any health system executive team conversation, when the discussion inevitably turns to what to do about artificial intelligence, you might hear an increasingly common refrain: "Let's wait and see what our EHR vendor delivers."
It's an understandable instinct, because it starts from something true. The EHR is the operating system of clinical care. It is where the clinical record lives, where orders are placed, where lab results land, where clinicians spend the bulk of their working hours. It’s where a health system's most sensitive data is managed and governed. Clinicians have spent two decades building workflows, trust, and muscle memory around it. Any AI strategy that ignores that reality, or tries to route around the system of record, is not a serious strategy.
That is exactly why the EHR vendors' recent push into AI has real weight behind it. They are investing heavily, and early tools should prove useful. When the company that runs your system of record promises AI is coming, and delivers tools clinicians actually like, waiting can feel like discipline rather than delay.
The mistake is not in valuing the EHR. It is in concluding that because the EHR is central to clinical care, it should also be the boundary of a health system's AI strategy. That belief rests on a misunderstanding of how both health systems and AI work, and recognizing this can empower leaders to take charge of their AI future.
1. Cross-system connectivity is now a pre-condition for value
A health system runs on hundreds of applications: the EHR, yes, but also workforce systems, payer portals, contract management, revenue cycle systems, imaging archives, call centers, and the platforms running home health, hospice, and long-term care, which often have none of the automation the flagship EHR enjoys. Just as the modern health system manages care in multiple settings and modalities, the technology footprint mirrors that. The most valuable work, and the most manual, happens in the handoffs between all of these systems, where a person re-keys what one application already knows into another that doesn't. Determining whether a patient should be admitted as inpatient or held under observation status means reconciling clinical data with payer-specific rules that live outside the chart. Managing capacity means joining clinical demand with staffing data no EHR contains. Prior authorization remains one of medicine's great time sinks precisely because payer portals lack even basic connectivity.
An AI tool that lives inside one system, even the biggest one, sees only one room of the house, and the next wave of AI makes that constraint decisive. The field is moving from question-answering tools to agents that complete multi-step work that will cross systems almost by definition: an agent resolving a denied claim must read the chart, the payer contract, and the correspondence history, and only one of those lives consistently in the EHR. The agentic era turns cross-system connectivity from a futuristic possibility into the precondition to drive value.
2. Frontier models move years faster than EHR vendors
A second constraint is pace. EHR vendors will deliver useful tools; the problem is pace. Frontier models from OpenAI, Anthropic, and Google now improve meaningfully every few months, on their builders' calendars, while testing, integrating, and shipping enterprise software takes years; whatever arrives in the EHR will be at least a generation behind on the day it lands.
Meanwhile, the losses AI could address keep accruing: denials written off, capacity stranded, patients who cannot get scheduled, clinicians drowning in documentation. Waiting caps you at one roadmap during the fastest capability growth health IT has ever faced.
3. Health systems must control their durable assets: governance, data, and implementation
If the alternative to waiting for your EHR vendor was to sign contracts with dozens of AI point solutions, waiting might genuinely be attractive. Health system leaders already exhausted by vendor sprawl, are rightly alarmed by its successor, AI sprawl: models and tools multiplying across the enterprise faster than anyone can track what they touch or whether they work.
Health systems seeking to maximize the value of their AI investments must build a set of durable assets that move the system towards coherent, reliable implementation. The durable assets in our health systems are the connective tissue: the data infrastructure, the governance, and the organizational implementation capability that let a health system absorb each new wave of technology as it arrives, from its EHR vendor or anyone else.
Governance: Every AI deployment, whether it came from your EHR vendor, a partner, or your own staff, should be evaluated, monitored, and measured the same way. Value should count only when the health system's own finance and clinical teams say it does, not when a vendor's dashboard does.
Data: Broad, well-governed data access across the EHR, workforce systems, payer portals, and document repositories is what makes the tenth use case fast, and it is exactly the investment no single application vendor, including your EHR vendor, will make on your behalf.
Implementation Capability: The uncomfortable truth of 2026 is that technology is no longer the primary barrier to AI value in healthcare; change management, workflow redesign, and leadership alignment are, and no vendor roadmap solves those, because they are yours. Organizations reporting durable gains build trusted, reliable implementation support capacity that can scale with the AI use cases.
4. Waiting for the EHR means competing on the same tools as everyone else
There is a fourth cost to waiting that has nothing to do with speed or connectivity: it puts your AI strategy on the same footing as every other health system running the same EHR, the lowest common denominator. When Epic or Oracle ships a feature, they ship it to the entire install base at once, on the same release schedule, with the same defaults. That is precisely the point of the product, and it is exactly why it cannot be a source of advantage. A capability every competitor receives on the same day, configured in similar ways, is not a strategic advantage; it rapidly becomes table stakes.
Health systems that compete well on cost, quality, or access have always done so by building something distinctive in how they find and retain patients, in their staffing model, or in their markets. AI does not change that logic; if anything it sharpens it, because the tools that create real separation are the ones built or configured around your specific populations, your specific set of payer relationships, and your specific operational bottlenecks, none of which an EHR vendor is optimizing for. A health system that treats the EHR's AI roadmap as its AI strategy is choosing to be as slow and standard as every other health system on that EHR, indefinitely.
Leverage the EHR but don’t wait on it
None of this is an argument against EHRs. They will remain the clinical systems of record, and they should. Adopt their AI tools where they fit; tools embedded in the clinician's native workflow have advantages. The most sophisticated systems already run a simple triage rule: take the native tool when it exists on a credible near-term roadmap, and act independently the moment a needed capability is years out and the return is clear. Applied consistently across the enterprise, that rule will distribute the AI roadmap not to hundreds of vendors but to a handful of platforms including the EHR.
Healthcare has spent decades as the economy's great productivity paradox: ever more technology, ever less time for the work that matters. AI is the first technology in a generation with a credible shot at changing that, but only for organizations that treat it as an enterprise capability rather than an EHR feature. The EHR is the foundation of clinical care, not the ceiling of what AI can do for a health system. The governance, the data infrastructure, and the implementation muscle are yours to build, and they will compound long after today's features are table stakes.
Ready to build an AI strategy that extends beyond the EHR roadmap? Start a conversation with our team.
.png)
.png)
.png)
