Breaking Down the Silos: Why AI Transformation Requires a More Connected Healthcare Organization

The next generation of healthcare AI will not succeed inside traditional organizational silos. The real opportunity is connecting people, knowledge, data, technology, and decisions around outcomes.

Healthcare organizations have traditionally been built around departments. Compliance, revenue cycle, IT, clinical operations, finance, privacy, quality, legal, and human resources each have defined responsibilities. There are good reasons for these structures: they establish expertise, accountability, and clear lines of responsibility.

But they also create silos. As artificial intelligence becomes embedded across healthcare operations, those silos may become one of the greatest barriers to meaningful AI transformation because AI does not neatly respect our organizational chart.

A single AI-enabled workflow may touch data, technology, compliance, privacy, clinical operations, reimbursement, workforce, and governance—all at the same time. That requires us to rethink not only the technology we use, but also how our organizations work together.

Traditional Hierarchies Were Designed for a Different Era

Most organizations still operate through relatively linear structures. Information moves up, decisions move down, issues pass from one department to another, expertise is concentrated within functional areas, and approvals move through layers. That structure can provide important oversight, but it can also make organizations slow.

Consider implementing an AI-assisted coding workflow. At first glance, it may sound like a coding department initiative. It is not. The technology team needs to understand integration and infrastructure. Revenue cycle must evaluate reimbursement implications, coding professionals need to validate accuracy, and compliance must assess regulatory risk. Privacy and security need to understand the data the system accesses, while legal and procurement evaluate the vendor.

Operational leaders need to redesign the workflow, leadership must understand cost and organizational value, and someone must determine where human judgment remains required. If each group evaluates the project independently and sequentially, implementation can become painfully slow—or critical risks can fall between departmental boundaries. AI requires something more connected.

Organize Around Outcomes, Not Just Departments

One concept I believe healthcare leaders should begin considering is organizing AI initiatives around business outcomes rather than organizational silos. Instead of asking which department owns an AI project, ask: What outcome are we trying to achieve, and whose expertise is required to achieve it responsibly?

That changes the structure of the conversation. If the objective is to reduce claim denials, the work may require coding, revenue cycle, clinical documentation, IT, data analytics, compliance, and payer expertise. If the objective is to improve regulatory intelligence, the team may need compliance professionals, regulatory experts, data specialists, technology teams, operational leaders, and AI governance.

The team forms around the value being created, not simply the department implementing the technology.

AI Makes Cross-Functional Teams More Important

I believe we will increasingly see healthcare organizations create smaller, cross-functional teams around important AI use cases. Think of them as adaptive teams or pods.

A pod may include a domain expert who understands the work, a technical expert who understands the AI and infrastructure, a data expert who understands the information feeding the system, a governance or compliance expert who understands risk and controls, and an operational leader who understands the workflow. Depending on the use case, it may also require clinical, privacy, security, legal, finance, or other expertise.

The composition should change based on the problem. We do not need another permanent committee for every AI use case. We need the right expertise connected to the right problem at the right time.

Knowledge Needs to Move Across Boundaries

This connects directly to another major AI transformation challenge: organizational knowledge. In many healthcare organizations, knowledge remains trapped within departments. Compliance knows something revenue cycle does not. Revenue cycle knows something IT does not. Clinical operations understand workflow nuances the vendor does not, while IT understands system limitations that operational leaders may never see.

AI creates an opportunity to connect that knowledge, but only if we intentionally make it accessible. Organizational knowledge is becoming an increasingly important AI asset. The next step is making that knowledge available across appropriate organizational boundaries while maintaining the necessary controls.

Accessible does not mean unrestricted. It means the right people—and the right AI systems—can access the right information for the right purpose.

AI Agents May Become Part of the Team

This is where the future becomes particularly interesting. As organizations move beyond generative AI toward agentic AI, AI may become embedded directly into cross-functional workflows.

Imagine a regulatory intelligence workflow. An Assistant Agent retrieves and summarizes new regulatory information. An Analyst Agent identifies changes and potential implications. Specialized agents examine reimbursement, documentation, operational, or compliance implications. An Orchestrator Agent coordinates the workflow and determines what happens next, while a Guardian Agent checks source traceability, required approvals, and defined governance rules.

Human experts then review the consequential conclusions and determine the appropriate response. The AI agents are not replacing the team. They are becoming part of how the team works. That is a significant change in organizational design.

AI Should Not Become Another Silo

There is a risk here that healthcare organizations should recognize early: we could easily create an AI department that becomes another organizational silo. AI specialists sit together, develop tools, choose technology, create models, and then deliver solutions to the rest of the organization.

That approach may work technically, but it can fail operationally because the people designing the AI may not understand the work deeply enough. This is why I continue to emphasize the importance of domain experts—the coder, compliance officer, auditor, physician, nurse, revenue cycle professional, privacy professional, and person actually performing the workflow.

They need a meaningful role in AI design. The people who know the work must help design how AI participates in the work.

Decisions Should Move Closer to the Expertise

Traditional organizations often move decisions upward. AI-enabled organizations may need to move some decisions closer to the expertise and evidence. That does not mean eliminating leadership oversight. It means defining decision authority more intelligently.

Organizations need to determine what a cross-functional team may decide, what an AI agent may recommend or execute, and which matters require compliance, executive, or legal approval. They must also define which decisions always remain human-owned.

Those boundaries should be designed deliberately. This is where governance and organizational design begin to intersect.

Human-in-Governance™ Becomes Even More Important

As teams become more fluid and AI agents participate in workflows, accountability cannot become fluid. Someone still needs to own the decision. That is why I believe Human-in-Governance™ becomes even more important in adaptive organizations.

AI can retrieve, analyze, compare, monitor, prioritize, route, coordinate, and—within defined boundaries—execute. But humans establish those boundaries. Humans determine the acceptable level of risk, decide which matters require professional judgment, approve consequential actions, and remain accountable for the operating model.

The organizational structure may become more adaptive. Accountability should become clearer—not weaker.

The Goal Is Not to Tear Down Every Silo

There is an important distinction here. Silos are not inherently bad. Healthcare requires specialized expertise. Privacy professionals should understand privacy, coders should understand coding, cybersecurity professionals should understand cybersecurity, and compliance professionals should maintain independence. Legal privilege needs appropriate protection, and certain information should absolutely remain restricted.

The goal is not to eliminate specialization. The goal is to prevent specialization from becoming isolation. We want deep expertise and better connection between experts.

Measure Whether the New Model Is Actually Better

Organizational redesign should also be measurable. If we create cross-functional AI teams, we should determine whether decisions happen faster, fewer issues are lost during handoffs, and implementation time decreases. We should also assess whether decision quality improves, compliance concerns are identified earlier, and employees spend less time searching for information.

Leaders need to know whether AI initiatives produce measurable business outcomes, teams learn faster, and customers or patients experience better results. We should not reorganize simply because “adaptive organizations” sound innovative. The structure should improve the outcome.

The ProCode Perspective

Healthcare AI transformation is forcing us to reconsider something much larger than technology: How should work itself be organized? The traditional model has often been department, hierarchy, handoff, approval, and then the next department. AI gives us the opportunity to create something more connected: People + Knowledge + Data + AI + Governance → Business Outcome.

The organizations that succeed will still need deep functional expertise, but they will become much better at connecting that expertise across organizational boundaries. They will bring domain experts closer to technology teams, make knowledge more accessible, organize teams around outcomes, and embed AI into workflows rather than simply purchasing AI tools.

They will move appropriate decisions closer to the people with the knowledge to make them while maintaining clear human accountability throughout the process. The future of healthcare AI is not simply about building smarter technology. It is about building smarter organizations around it.

Sometimes the biggest barrier AI needs to overcome is not technological at all. It is the walls we built between ourselves.