From AI Pilots to Real Transformation: 7 Keys Healthcare Leaders Need to Get Right

AI is moving at warp speed. Experimenting with AI is easy. Transforming an organization with it is much harder.

Healthcare organizations are experimenting with artificial intelligence at an extraordinary pace.

We are seeing AI used for documentation, coding, revenue cycle management, prior authorization, analytics, compliance monitoring, patient communication, regulatory research, clinical decision support, and countless administrative functions.

But there is a significant difference between using AI and transforming an organization with AI.

A pilot can demonstrate that technology works. Transformation requires proving that it creates value, can be governed, can scale, and can become part of how the organization actually operates.

Recent research reinforces this distinction. Organizations capturing meaningful value from AI are increasingly redesigning workflows and operating models rather than simply placing AI on top of existing processes. Leadership fluency, workforce capability, governance, data readiness, and organizational change are becoming as important as the technology itself.

In my work at the intersection of healthcare compliance and healthcare AI transformation, I see seven capabilities that leaders should be thinking about now.

1. Leadership Mindset Shift: Curiosity Over Authority

AI is changing too quickly for any leader to have all the answers. That can be uncomfortable, particularly in healthcare, where leadership traditionally depends heavily on expertise, established processes, regulation, and clearly defined authority. AI requires something additional: curiosity.

The strongest leaders will be willing to ask:

• What don’t we understand yet?

• What could this technology make possible?

• What risks haven’t we considered?

• What assumptions about our current workflow should we challenge?

• Who else needs to be part of this conversation?

This does not diminish leadership authority. It changes how that authority is exercised. Leaders increasingly need enough AI fluency to ask better questions, evaluate recommendations, challenge vendors, understand limitations, and govern decisions, even if they are not technologists. Recent research found leaders reporting enterprise value from AI at substantially higher rates when leadership teams demonstrate high AI fluency.

Leadership takeaway: You don’t have to become an AI engineer. You do need to become an AI-literate leader.

2. Value-First Execution: Measure Outcomes, Not Activity

One of the easiest traps in AI transformation is confusing activity with progress. Organizations may ask how many AI pilots they have launched, how many employees are using AI, or how many AI tools they have purchased.

Those numbers may be interesting, but they don’t necessarily tell us whether AI is creating value.

Healthcare organizations should start with the business problem and define the desired outcome.

If AI is introduced into compliance monitoring, for example, success might mean:

• Reduced research time

• Faster identification of regulatory changes

• Improved consistency

• Fewer missed updates

• Faster response to emerging risks

• Increased professional capacity without proportional staffing growth

Those are measurable outcomes. “We implemented AI” is not.

This is also why organizations should be willing to discontinue pilots that don’t demonstrate meaningful value. Research on scaling AI increasingly emphasizes concentrating resources on important business problems rather than accumulating experiments.

Leadership takeaway: Start with the outcome, not the technology.

3. Data and Knowledge Liberation: Unlock What Your Organization Already Knows

This may be one of the most overlooked opportunities in healthcare AI. Healthcare organizations possess enormous amounts of valuable knowledge.

The problem is that much of it is fragmented across:

• Policies

• Shared drives

• Emails

• Audit reports

• Regulatory interpretations

• Training materials

• Corrective action plans

• Meeting notes

• Historical decisions

• The minds of experienced employees

That knowledge has tremendous value, but AI cannot effectively leverage information it cannot reliably find, access, understand, or trust. This means organizations need to think beyond traditional data.

They need a knowledge strategy that answers critical questions:

• What information is authoritative?

• Who owns it?

• Where is it stored?

• How is it updated?

• What is outdated?

• What may AI access?

• How does the system know which source should take precedence?

Enterprise-specific institutional and domain knowledge can become a significant source of differentiation when AI systems can access it appropriately.

Leadership takeaway: Before asking what AI knows, ask whether your organization has made its own knowledge usable.

4. Distributed AI Literacy: Empower the Domain Experts

AI transformation cannot belong exclusively to IT. This is particularly important in healthcare.

Technology teams understand systems, but compliance professionals understand regulatory risk. Coders understand coding rules, revenue cycle leaders understand reimbursement workflows, clinicians understand patient care, privacy professionals understand information risk, and legal counsel understands legal exposure.

These domain experts need enough AI literacy to participate meaningfully in AI design and governance.

I believe this will become one of the defining capabilities of successful healthcare organizations.

We should not expect every compliance professional, physician, auditor, or revenue cycle leader to become a data scientist.

But they should increasingly understand concepts such as LLMs, agents, hallucinations, retrieval, orchestration, human oversight, data provenance, confidence, escalation, and governance.

Why? Because the people who understand the work must help determine how AI participates in the work.

Leadership takeaway: AI literacy should not be concentrated in a small technology team. It needs to spread across the organization.

5. Adaptive Organization: Break Down the Silos

AI does not respect organizational charts. A single AI workflow may touch compliance, IT, privacy, cybersecurity, legal, finance, revenue cycle, operations, clinical leadership, and data governance.

That means traditional organizational silos can quickly become barriers.

Consider an AI-enabled coding workflow. That workflow raises important ownership and oversight questions:

• Does ownership sit with coding, revenue cycle, compliance, IT, or the vendor?

• Who validates the model?

• Who monitors accuracy?

• Who determines acceptable risk?

• Who investigates an error?

• Who has authority to suspend it?

These questions demonstrate why AI transformation requires cross-functional governance and ownership. The future organization may increasingly rely on multidisciplinary teams assembled around workflows and outcomes rather than traditional departmental boundaries. That also means decision rights must be clear, because collaboration without accountability creates another problem.

Leadership takeaway: AI transformation is enterprise transformation. It cannot be delegated to one department.

6. Will Through Action: Governance Must Operate, Not Merely Exist

Organizations can develop excellent AI principles covering responsible AI, transparency, accountability, fairness, human oversight, privacy, and security. Those principles matter, but principles alone do not govern AI.

They must become operational. Organizations need to answer:

• Who approves an AI use case?

• What information may the system access?

• Which sources may it rely upon?

• Which decisions require human review?

• What happens when the AI is uncertain?

• What happens when authoritative sources conflict?

• What prevents an unapproved action?

• Who investigates an exception?

• Which controls monitor behavior, and which controls actually stop the workflow?

This is where governance moves from intention to execution. It also connects directly to what I describe as Human-in-Governance™. The human should not merely appear at the end of the workflow to approve what AI has already done. Humans should establish the boundaries within which AI operates, determine which decisions remain human-owned, define escalation criteria, and retain accountability for consequential outcomes. Governance should not exist to prevent innovation. Good governance creates the conditions for responsible speed.

Leadership takeaway: If your governance cannot influence what actually happens inside the workflow, it isn’t enough.

7. Momentum Over Perfection: Launch, Learn, Improve

This may be one of the most important principles right now. AI is moving too quickly to wait for perfection. That doesn’t mean healthcare organizations should recklessly deploy AI. It means we need a disciplined way to experiment.

A disciplined experiment should:

1. Start with a meaningful but manageable workflow.

2. Establish boundaries and define the desired outcome.

3. Identify the risks and build the controls.

4. Pilot and measure the workflow.

5. Learn from the results and improve it.

6. Decide whether it is ready to scale.

Pilots should not simply prove that the technology functions. They should test whether the organization is ready, whether the workflow works, the controls work, employees trust it, humans understand their responsibilities, and measurable value is being created.

Research on AI transformation similarly emphasizes using pilots as stepping-stones to scalable operating change rather than endpoints in themselves. There will never be a moment when AI stops evolving long enough for healthcare organizations to develop the “perfect” AI strategy. We have to build the capability to learn while moving.

Leadership takeaway: Responsible experimentation is not the opposite of governance. It should be part of governance.

The Bigger Picture: AI Transformation Is Organizational Transformation

When you look at these seven capabilities together, something becomes very clear.

Only part of AI transformation is about AI. The rest is about leadership, value, knowledge, people, organizational design, governance, and change. That is why purchasing better technology alone will never produce true AI transformation. The technology may be extraordinary, but organizations still have to redesign how work gets done around it.

The ProCode Perspective

Healthcare has been transforming for decades, but AI presents a different kind of challenge because the technology is advancing faster than many organizations’ ability to establish the structures around it.

That gap concerns me more than the technology itself.

AI is moving at warp speed. Organizations do not need to chase every new model, platform, or agent that enters the market, but they do need to keep learning.

They need leaders who understand enough to lead, domain experts who understand enough to participate, knowledge that AI can reliably access, governance that operates inside the workflow, measurable outcomes, and the willingness to start, learn, adjust, and move forward. Because the divide I expect to see in healthcare will not simply be between organizations that use AI and those that don’t.

It will be between organizations that experimented with AI, and organizations that learned how to transform with it. And that transformation starts long before the technology goes live