Healthcare isn’t struggling to adopt AI. In fact, the industry has moved quickly from early experimentation to widespread investment in tools designed to improve clinical decision-making, reduce administrative burden, and drive financial performance. Many health systems and payers now have AI strategies, and many have already demonstrated that these technologies can work in controlled environments.
And yet, a familiar pattern continues to emerge. Pilot programs show promise and early results generate excitement, but when it comes time to scale, impact stalls. Adoption lags, outcomes plateau, and organizations are left wondering why tools that seemed so powerful in theory are not delivering in practice.
This challenge isn’t unique to healthcare. McKinsey research has shown that while AI adoption is accelerating rapidly, many organizations are still early in translating that momentum into scaled operational and financial impact, reinforcing that the gap between experimentation and execution remains a core barrier.
The answer is less about the sophistication of the technology and more about how it fits into the reality of healthcare operations. AI isn’t failing because it lacks capability. It’s failing because it isn’t consistently reaching the moments where decisions are made.
The problem isn’t AI. It’s where it shows up.
AI is exceptionally good at generating insight. It can identify rising-risk patients, flag gaps in care, detect anomalies in claims, and surface meaningful trends across millions of records with remarkable speed. But healthcare does not run on insight alone. It runs on action.
In many cases, the output of AI systems exists outside the environments where that action occurs. Too often, I see AI buried in dashboards, delivered through disconnected tools, or surfaced after the moment when they could have influenced a decision. Workflow integration, not model performance, is where most healthcare AI breaks down.
At its core, AI creates value in two ways: it can automate tasks humans already do, and it can unlock capabilities humans could never do at scale. Yet, most scaled AI use cases today remain concentrated in administrative workflows such as revenue cycle management, ambient documentation, and strengthening payment integrity. These are important wins because they automate manual work and create immediate efficiency. But when we look at the bigger opportunity of improving the total cost of care across populations, we’ve barely scratched the surface.
For example, a care management team may receive a weekly AI-generated list of high-risk patients through a standalone analytics dashboard. The insight itself may be accurate, but because it sits outside of the care management platform and requires manual review and re-entry, it often goes untouched. By the time action is taken, the window for early intervention may have already passed.
This disconnect creates friction. In a system already under immense pressure, friction is often the deciding factor between action and inaction. Clinicians and operators are not resisting AI. They are responding to environmental constraints. When insight requires additional steps, additional systems, or additional time, it simply does not get used.
Studies reinforce this reality, showing that up to 81% of clinicians overlook tools external to their primary EHR workflows. If AI is not embedded into workflows, it’s effectively invisible.
This gap between insight and execution consistently shows up in both research and real-world experience. Across the industry, and in my experience working with payers and providers at scale nationwide, workflow integration consistently emerges as one of the primary barriers to scaling AI beyond initial pilots.
Taken together, the message is clear: healthcare has made meaningful progress in generating intelligence, but it is still early in translating that intelligence into consistent, operational impact.
From AI tools to operational impact
These challenges point to a broader conclusion. The barrier to effective AI adoption in healthcare is not technical. The models are advancing rapidly, data capabilities continue to improve, and new tools are entering the market at an accelerating pace. The real barrier is design — specifically, designing AI to reflect how healthcare actually works.
For AI to move beyond the pilot phase and deliver sustained impact, organizations must shift their focus from deploying tools to operationalizing intelligence. That means embedding insights directly into the systems where decisions are made, prioritizing signals so users can focus on what matters most, tailoring outputs to the needs of specific roles, and ensuring insights are delivered in real time, aligned with key decision moments.
There are also real barriers inside the traditional EHR environment that make activating the “action layer” difficult. Solving this challenge may require thinking differently. Instead of continuing to force AI into existing workflows, build AI around traditional systems so it can act in ways humans never could. That means moving beyond the EHR as the sole center of gravity and designing orchestration layers that manage complexity across the ecosystem.
When these elements come together, AI stops being something users seek out and becomes something that actively supports how they work.
For instance, when AI is embedded directly into the EHR or when the EHR fades to the background and new apps support clinicians, the care team can see a prioritized care gap alert during a patient visit along with a recommended next action. Instead of requiring follow-up work after the encounter, the insight becomes part of the clinical workflow, enabling immediate intervention, improving quality performance, and reducing missed opportunities for care.
This isn’t a 12-month transformation. It’s a massive change-management undertaking that requires organizations to rethink workflows, accountability, and even where decisions should be made. If we’re serious about bending the cost curve and improving outcomes, we must think differently about how AI activates action—not just where it surfaces insight. That’s how our industry moves from automation to transformation.
As healthcare enters a more accountable phase of AI adoption, the conversation is shifting from whether organizations can implement AI to whether they can prove it delivers measurable outcomes. Those who succeed will be defined by their ability to close the gap between insight and action.
Because in healthcare, AI does not create value when it identifies a problem. It creates value when someone acts on it.
Image: Getty Images, erhui1979
Michael Meucci is President and Chief Executive Officer at Arcadia, where he is focused on helping Arcadia customers — and, by extension, the healthcare industry — embrace technology and use data to create happier, healthier days for all. Michael is a recognized leader in the healthcare industry, particularly for his work and innovation in advancing data for better healthcare outcomes. Becker’s Hospital Review named him a Rising Star and Boston Business Journal named him a 40 Under 40.
Michael’s depth of industry and technical knowledge, combined with his strategic mindset, have allowed him to forge strong relationships with leaders at many of the largest and most innovative organizations in healthcare. As a result, he’s earned board positions at the Healthcare AI platform N1 Health and Boston-based federally qualified health center Fenway Health. Additionally, Michael serves as an advisor to Guidehealth, a value-based care enabler focused on patient-centered care solutions for health systems.
This post appears through the MedCity Influencers program. Anyone can publish their perspective on business and innovation in healthcare on MedCity News through MedCity Influencers. Click here to find out how.
