Healthcare has never really had a technology problem. The greater challenge has been turning technology into better ways of working.
Over the past decade, healthcare organizations have invested heavily in electronic health records, digital services, automation, analytics, and virtual care. Now, artificial intelligence is moving rapidly into the same environment. AI can summarize information, identify patterns, support decision-making, and take on repetitive tasks that consume valuable staff time.
The potential is significant. But before introducing another AI capability, healthcare leaders should ask a more fundamental question:
Is the workflow ready for it?
It is easy to focus on what a new technology can do. It is much harder, and often more important, to understand how work is actually being done today.
A fragmented workflow does not automatically become a better workflow when AI is added. Sometimes, it simply becomes a more complicated digital version of the same problem.
Technology Does Not Fix a Workflow by Itself
Healthcare processes rarely operate as neatly as they appear on a process map.
A patient may move through registration, assessment, diagnostics, treatment planning, and follow-up, with different people responsible for different parts of the journey. Information may be entered more than once. A decision may depend on information that exists somewhere in the organization but is not available when it is needed.
The problem may not be a lack of technology at all. It may be the way the work has been organized.
In healthcare operations, some of the most persistent problems are created not by a lack of capability, but by the way a process has evolved over time—one handoff, one workaround, and one additional requirement at a time.
Consider a simple patient journey. A referral arrives, an appointment is scheduled, the patient is assessed, information is gathered, a decision is made, and follow-up is arranged. On paper, this sounds straightforward.
In practice, a delay at any point can affect everything that follows. Missing information may lead to another phone call. A handoff between teams may create uncertainty about responsibility. A clinician may spend valuable time searching for information that should have been available earlier.
Adding AI to one step does not necessarily solve the problem.
Someone still needs to receive the information, understand it, determine what to do with it, and take responsibility for the next step.
That is why the World Health Organization emphasizes human autonomy, safety, transparency, responsibility, and accountability in the governance of AI for health. [1] Technology cannot be separated from the environment in which it operates.
Before Automating, Understand How Work Actually Happens
One of the most important lessons in healthcare transformation is that the documented process is not always the real process.
People develop workarounds. Experienced staff learn how to deal with exceptions. Teams create informal ways of communicating when the formal process does not work efficiently.
These workarounds may not appear in a technology assessment, but they can be essential to keeping operations moving.
Before introducing AI, leaders should therefore look closely at the current workflow.
Where does the patient wait unnecessarily?
Where do staff repeat the same work?
Where are handoffs creating delays?
Which decisions require information that is difficult to access?
Where are clinicians spending time looking for information instead of using it?
Which activities genuinely require professional judgment, and which are repetitive enough to be supported by automation?
These questions can reveal opportunities that are more valuable than simply adding another digital capability.
Sometimes the answer will be AI.
Sometimes the answer will be removing an unnecessary step.
Sometimes it will be clarifying who owns a decision.
That distinction is part of operational maturity.
AI Should Enter the Workflow, Not Sit Beside It
Another common mistake is treating AI as something that can simply be placed next to an existing process.
A better approach is to first ask how the workflow should operate and then determine where AI can contribute.
Take clinical documentation as an example. The value of an AI-enabled documentation process is not simply that it can produce text. The real question is whether it makes documentation easier for the clinician, produces information that is accurate and useful, and reduces overall administrative effort.
The same principle applies to administrative work.
An organization might want to use AI to prioritize referrals or requests. But if the underlying referral process is unclear, responsibilities are fragmented, or essential information is missing at the point of review, automating prioritization may simply move the bottleneck somewhere else.
This is why workflow redesign should come before automation.
The objective is not to automate everything that can be automated.
The objective is to create a better way of working and then use technology where it can genuinely strengthen that process.
Frontline Staff Are Part of the Solution
Healthcare transformation can sometimes focus heavily on leadership, technology, and implementation plans while overlooking the people who experience the workflow every day.
Frontline professionals see problems that are often invisible from a strategic or technical perspective.
They know which alerts are routinely ignored. They know which steps create unnecessary work during a busy shift. They know where patients become confused and where information gets lost between teams.
That knowledge is valuable.
Involving frontline staff early can also improve the quality of the final solution. Instead of asking people to adapt to technology that has already been selected, organizations can design the future workflow around the realities of care delivery.
This does not mean that every request from frontline staff should determine a transformation strategy. It means that operational experience should be considered alongside technical capability and strategic objectives.
This becomes particularly important with AI.
As these systems become more capable, the question should not simply be whether they can replace human involvement wherever possible. The more useful question is where AI can reduce low-value work while allowing professionals to focus on judgment, communication, empathy, and accountability.
From Digital Projects to Operational Intelligence
Healthcare organizations have become increasingly capable of launching digital projects.
The harder challenge is turning those projects into sustainable operational improvement.
A successful pilot does not necessarily mean an organization is ready to scale. Before expanding an AI-enabled workflow, leaders need to consider whether the underlying process is consistent, whether the necessary data are available and reliable, whether responsibilities are clear, and whether outcomes can be measured.
There also needs to be room for adjustment.
Healthcare operations are not static. Staffing changes. Patient volumes change. Clinical practices evolve. Regulations change. New technologies appear.
A workflow that works well today may need to be redesigned tomorrow.
This is where operational intelligence becomes more important than technology adoption alone.
Operational intelligence means understanding what is happening across the operation, recognizing where friction exists, understanding why it exists, and using that knowledge to make better decisions about how work should be organized.
AI can support that capability.
It cannot replace it.
Recent WHO guidance on large multi-modal models emphasizes the importance of governance, risk management, human oversight, and ongoing evaluation as these systems are introduced into health settings. [2]
The goal, therefore, should not be to create more technology around existing processes. It should be to create greater visibility into how work is performed and where technology can meaningfully improve it.
The Question Leaders Should Be Asking
The next phase of healthcare transformation should not be measured simply by how many AI tools an organization has implemented.
It should be measured by what has actually improved.
Has unnecessary work been removed?
Has staff time been redirected toward higher-value activities?
Has the patient journey become easier to navigate?
Are decisions being made with better information at the right time?
Are clinicians spending less time dealing with administrative friction?
Can the organization demonstrate that the technology is producing the outcome it was introduced to achieve?
These are operational questions, but they are also the questions that determine whether digital investment creates real value.
Healthcare should not resist AI. There is too much potential in these technologies to do that.
But neither should healthcare organizations assume that every problem is an AI problem.
If a workflow is broken, redesign it before automating it.
If information is difficult to access, improve the information flow.
If responsibilities are unclear, clarify them.
If frontline professionals are struggling with a process, understand why.
And when AI is introduced, place it within a workflow that has a clear purpose, clear ownership, appropriate human oversight, and a measurable outcome.
The future of healthcare transformation will not necessarily belong to the organizations that adopt AI the fastest.
It will belong to the organizations that understand their operations well enough to know where AI belongs, where it does not, and how people, processes, and technology need to work together to create better care.
That is the shift from digital transformation to operational intelligence.
And it may be one of the most important shifts healthcare leaders make in the years ahead.
Author Bio
