The Missing Link in Healthcare AI
Ask most frontline clinicians what they truly need from AI, and the answer rarely begins with more data. Monitors already sound alarms. Dashboards already flag risks. EHR systems already surface risk scores. Yet, what clinicians need most — and what most AI deployments still fail to deliver — is clear, timely action.
This critical gap between insight and execution emerged as the defining theme at a panel on care model transformation at the Becker’s 16th Annual Meeting. Healthcare executives gathered to discuss what it truly takes to move AI from an advisory tool to an operational one.
From Alerts to Action: Closing the Gap
Why More Intelligence Is Not the Answer
“The gap today is not insight,” said Jason Cohen, MD, Chief Medical Officer of Qventus. “We don’t need more intelligence. As a clinician, please don’t give me another alert — instead, start automating the next workflow step for me.”
Dr. Cohen offered a direct example. A sepsis alert that stops at the notification is, effectively, a half-built solution. What clinicians need instead is an AI system that identifies early sepsis risk, notes a missing lactate order, and asks: Do you want me to order that for you? Furthermore, such a system might also flag that the morning’s chest X-ray suggests a right lower lobe pneumonia — all in one coordinated moment.
The Workflow Execution Problem
This distinction — between surfacing a finding and completing a workflow step — is precisely where most health system AI implementations stall today. Closing that gap, moreover, requires more than a technology upgrade.
“That requires health systems to build the muscle of moving from insights and actions to execution,” Dr. Cohen explained. “That is where both the value is, and where we will see a real reduction in burden for frontline teams. But it demands serious workforce development.”
That muscle has a specific anatomy. Health systems need staff who understand clinical workflows deeply — not just analysts who run models, but actual builders who translate AI output into reliable, point-of-care operational steps.
What Full AI Operationalization Looks Like
A Voice Agent That Handles Scheduling End to End
Michelle Stansbury, Associate Chief Innovation Officer and Vice President of Applications at Houston Methodist, offered one of the panel’s most concrete examples. Her team deployed an AI voice agent that handles inbound patient scheduling calls completely — with no human handoff required.
The rollout required its own change management strategy. “I’ve had to have people listen to that call recording because they don’t believe it,” Ms. Stansbury said. “But once they hear it, they become your change agents going forward.”
Converting skeptics through firsthand experience reflects a broader truth: workflow transformation rarely succeeds through top-down mandates. Instead, it succeeds when frontline staff experience the change as relief rather than imposition.
Change Management Is the Real Challenge
People, Not Platforms, Drive Adoption
Technology alone does not shift culture. Therefore, health systems that invest equally in people and process alongside their AI platforms gain a significant advantage. Staff must feel ownership over the change — not just recipients of it.
Consequently, showing early adopters real, working examples builds the internal momentum that no executive directive can manufacture. When colleagues hear a real patient call handled entirely by AI, belief replaces skepticism. That belief then spreads organically across teams.
SWAT Teams and Structured Governance
Jefferson Health’s Operational Blueprint
At Jefferson Health, Judd Hollander, MD, Senior Vice President of Healthcare Delivery Innovations, has institutionalized this principle through formal governance. His system created eight SWAT teams — an acronym Dr. Hollander coined for Synchronizing Workflows and Technology.
Each SWAT team includes leads from virtual care, nursing, medical informatics, IS&T, and a dedicated project manager. Teams operate on a standing biweekly cadence and complete required checklists along with full workflow documentation before any solution goes live.
“It really is all about the workflows and operations, not the technology,” Dr. Hollander said.
This structure keeps virtual care strategy tightly coupled to IT implementation — a connection that, at many health systems, exists in name only. The result is a system where no AI tool reaches deployment without a clear operational home.
Leadership Fluency Shapes AI Success
Knowing What to Build, Buy, and Wait For
Dr. Cohen identified a second, often overlooked prerequisite for execution: leadership teams with enough technical fluency to make sound decisions. Specifically, leaders must judge what to build, what to buy, and what to wait for.
That means distinguishing which AI capabilities are genuinely differentiating versus which already appear on an EHR vendor’s roadmap. It also means resisting pressure to purchase solutions that address visible problems while quietly shifting complexity elsewhere in the system.
Health systems that develop this discernment avoid vendor dependency and build durable, integrated AI capabilities instead.
The Road Ahead for Healthcare AI
Weaving Intelligence Into Existing Workflows
The future of healthcare AI does not belong to the system with the most models. It belongs to the system that most effectively weaves intelligence into existing workflows and builds solutions alongside the right partners.
“The future is less going to be defined by what we actually build and more by how we build it and who we build it with,” Dr. Cohen said. “Those health systems that turn intelligence into execution will be the ones that succeed.”
In summary, the real AI gap in healthcare is not a lack of intelligence. It is a lack of execution. Health systems that bridge this gap — through workforce development, structured governance, change management, and leadership fluency — will lead the next era of AI-driven care delivery.
