From AI Strategy to Execution: Healthcare Cannot Afford to Stand Still

AI is moving at warp speed. The question for healthcare leaders is no longer whether to engage with AI, but whether their organizations can build the leadership, governance, and operating models needed to keep pace responsibly.

Artificial intelligence is advancing at a speed unlike most technology transformations healthcare has experienced. New models, platforms, agents, capabilities, and use cases are emerging continuously. What seemed experimental a year ago is rapidly becoming operational.

Generative AI introduced us to systems that could create, summarize, analyze, and retrieve information. Agentic AI is pushing us further—toward systems that can monitor, reason across information, coordinate tasks, route work, interact with other systems, and take defined actions toward a goal. AI is moving at warp speed, and healthcare organizations do not have the luxury of ignoring that pace.

That does not mean adopting every new AI tool that enters the market. Doing so may create more risk than value. But organizations that are not actively building their understanding of AI, evaluating where it belongs in their operations, developing governance, and preparing their workforce will increasingly find themselves trying to catch up.

The challenge for healthcare leadership is therefore twofold: Move fast enough to remain relevant—but deliberately enough to remain responsible.

We Have Been Here Before—but This Time Is Different

Healthcare has undergone enormous technological transformation over the past several decades. We moved from paper medical records to electronic health records. HIPAA fundamentally changed how we approached health information. Electronic claims transformed reimbursement, value-based care changed how we think about performance, and data analytics reshaped payment integrity and compliance monitoring.

Each transformation required organizations to adapt. AI is different in one important respect: traditional technology generally waited for a human to tell it what to do. Increasingly, AI can participate in the work itself.

An AI system can retrieve information, analyze it, identify patterns, generate recommendations, prioritize work, coordinate tasks, and—in an agentic environment—potentially initiate actions. That changes the leadership conversation. We are no longer simply asking what technology to buy. We need to ask: What role should AI be allowed to play in our organization? That is an operating-model question.

An AI Tool Is Not an AI Strategy

One mistake organizations can make right now is equating AI adoption with AI strategy. Providing employees access to an AI platform, implementing an AI-enabled application, or running a pilot does not by itself constitute a strategy. Those may all be components of one.

A meaningful AI strategy connects technology to the organization’s objectives, workflows, people, risks, governance, and measurable outcomes. Healthcare leaders should understand which business problems they are trying to solve, which workflows are appropriate for AI augmentation, and where human judgment remains essential.

They also need to know which data AI will access, what actions it may take, which risks it introduces, which controls are required, and how the organization will measure whether AI actually improves the outcome. Without those answers, organizations risk accumulating AI tools without developing an AI operating model.

Start With the Workflow, Not the Technology

This has become one of the most important lessons in my own work with AI transformation. Do not begin by asking, “Where can we use AI?” Begin by asking where the work is difficult, repetitive, inconsistent, slow, or unnecessarily dependent on manual effort—and which outcome the organization is trying to improve.

Then map the workflow. Identify who performs the work, which information is required, where it comes from, where decisions are made, and where handoffs occur. Leaders should understand where professional judgment matters, where errors happen, and where compliance risks exist. Only then should they determine where AI belongs.

This distinction matters because there is a significant difference between automating a task and redesigning a workflow around AI and human collaboration.

Automation and Agentic AI Are Not the Same Thing

If a process is repetitive, predictable, and rules-based, traditional automation may be all that is necessary. Agentic AI becomes more interesting when work requires gathering information from multiple sources, analyzing context, determining what should happen next, coordinating activities, managing exceptions, and supporting human decision-making.

Consider regulatory intelligence in healthcare. A traditional process may require a compliance professional to visit multiple regulatory websites, review email alerts, identify new guidance, determine what changed, assess its significance, identify affected providers, and decide whether action is required.

An agentic workflow could continuously monitor approved authoritative sources, identify changes, retrieve the underlying guidance, summarize the update, compare it with prior requirements, classify its potential significance, identify affected areas, and prepare the information for expert review. But that does not mean AI should independently determine what a healthcare organization must do. That distinction is critical.

AI Needs Defined Roles

As organizations move toward agentic workflows, it becomes helpful to stop thinking about “the AI” as one undifferentiated capability. Different AI functions can play different roles.

An Assistant may retrieve and summarize information. An Analyst may compare information, identify patterns, or classify risk. A Tasker may perform a bounded, approved action. An Orchestrator may coordinate a multi-step workflow and determine where work should be routed. A Guardian may monitor whether defined governance, quality, evidence, and approval requirements have been satisfied.

Then there is the human. The human should not become an afterthought in this architecture. Human authority must be designed into it.

From Human-in-the-Loop to Human-in-Governance™

We hear frequently about keeping a “human in the loop.” That is necessary, but I believe healthcare organizations need to think more broadly. I call this Human-in-Governance™.

The important question is not simply where a human reviews the AI output. It is: Where does human authority reside throughout the system? Humans should define which decisions AI may support, which decisions remain exclusively human, what evidence is required, what risk thresholds trigger escalation, when AI must stop, and who remains accountable for the final outcome.

An AI system may appropriately identify a significant CMS policy change and classify it as high risk. But the decision that the policy requires a client to change its billing practices should remain with a qualified professional. AI performs the work it is authorized to perform. Humans govern the boundaries.

Speed Without Governance Is Not Transformation

This is where healthcare faces an important tension. We need to move quickly, but moving quickly cannot mean abandoning governance. Privacy, security, regulatory compliance, accuracy, transparency, bias, source reliability, vendor risk, auditability, and accountability all need to be considered as AI becomes embedded in healthcare operations.

The answer, however, cannot simply be to slow everything down. Good governance should enable responsible speed. When organizations clearly establish approved and prohibited uses, decision rights, escalation criteria, data boundaries, validation requirements, and human approval points, employees and leaders gain a framework within which they can innovate.

Governance should not simply tell people what they cannot do. It should help them understand how they can use AI responsibly.

The Cost of Waiting Is Growing

Healthcare leaders also need to consider the risk of doing nothing. Organizations that wait several years to develop AI capabilities may discover that competitors have already redesigned workflows, developed AI-literate workforces, improved turnaround times, built stronger knowledge systems, and learned through multiple generations of implementation.

The knowledge gap itself can become a competitive disadvantage. Healthcare leaders do not need to know how to build a large language model, but they increasingly need enough AI literacy to ask informed questions, evaluate proposals, understand risk, challenge vendors, govern implementation, and make strategic decisions.

Compliance professionals need to understand how AI affects their responsibilities. Boards need to understand the organization’s AI exposure. Employees need clarity about what is permitted and expected. Organizations must begin identifying which capabilities they will need tomorrow—not merely which tools they want today.

Measure Outcomes, Not AI Activity

Another important shift is how we define success. The goal should never be simply to “use more AI.” A successful AI transformation should produce measurable improvements.

Leaders should ask whether regulatory research time decreased, audit preparation became more efficient, turnaround times improved, and error rates declined. They need to know whether employees gained capacity for higher-value work, consistency improved, and compliance risk decreased—or increased. They should also measure how frequently humans corrected the AI and whether the organization achieved meaningful financial or operational value. AI utilization is not an outcome. Value is.

Healthcare Leaders Need to Start Now

Organizations do not need to have every answer today. None of us do. AI is evolving too rapidly for any governance framework, technology strategy, or operating model to remain static. That is precisely why organizations need to begin building the capability to adapt.

Start with a meaningful workflow. Understand it, identify the desired outcome, determine where AI can add value, and define what remains human-owned. Establish governance, test the workflow, measure results, learn, and then scale.

The organizations that succeed with AI will not necessarily be those that moved first or bought the most technology. I believe they will be the organizations that learned how to move responsibly at the speed of change.

The ProCode Perspective

At ProCode Compliance Solutions, we are approaching AI transformation through the same lens we have applied to healthcare compliance for decades: understand the risk, establish accountability, build appropriate controls, monitor performance, and continuously improve.

But we are also recognizing something equally important: AI is not waiting for healthcare to catch up. The pace of change is extraordinary, and healthcare leaders need to be actively developing their own knowledge and organizational capabilities now.

The choice should not be between innovation and governance. We need both. Strategy gives us direction. Governance gives us boundaries. Human expertise gives us judgment. AI gives us new capabilities. Bringing those elements together is where real transformation begins.

AI is moving at warp speed. We do not need to chase every new development, but we cannot afford to stand still.