Healthcare Already Adopted AI. Governance Just Never Caught Up.
- Zentara Group
- May 6
- 6 min read
By Allison Muhl, Founder of Zentara Group
If you run or lead an independent healthcare practice, this is written for you.
Let’s be honest about how a lot of this actually happens.
Before I say another word I want to acknowledge something.
I know and respect a lot of healthcare vendors. I came from that world. I sold pharmaceuticals and devices earlier in my career and I learned quickly that most sales conversations only scratched the surface of what was actually happening operationally inside a practice. I started thinking through every workflow like the physician using the product, then stepping back and thinking like the patient moving through the experience.
That perspective taught me one thing most vendors tend to take too lightly at times. A product can look incredible in a demo and still create operational confusion, workflow drift, or risk once it hits a real healthcare environment.
Which brings me to AI.
A physician hears about a new device, platform, or workflow tool at a conference, through a colleague, or during a vendor presentation.
The pitch sounds great.
Improved efficiency. Faster turnaround. Reduced administrative burden. Better patient experience. Staffing relief. New service capabilities. Maybe even a new billing code attached to it.
Now leadership is interested.
The device gets purchased. The workflow changes. Staff starts using it. Patients start interacting with it.
And somewhere in the middle of all of that, very few people stop and ask:
What exactly is the AI doing?
Where is the data going?
What role does human review still play?
Who is monitoring outputs?
What happens if the system is wrong?
What boundaries exist around use?
What have we actually approved versus what staff is improvising on their own?
That is not a criticism.
That is the reality of modern healthcare operations.
Most organizations are moving so fast trying to survive staffing shortages, operational pressure, reimbursement changes, and patient demand that governance becomes reactive instead of intentional.
But this is exactly why healthcare leaders should care.
Because AI is no longer sitting in a future conversation.
It is already influencing workflows, operational decisions, patient communication, diagnostics, and revenue generation inside real organizations right now.
And once AI becomes operational, leadership accountability becomes operational too.
Recently, the FDA’s Artificial Intelligence-Enabled Medical Devices list crossed more than 1,400 authorized devices in the United States. That number alone should force healthcare leaders to pause for a second.
Because if you still think AI in healthcare is “coming soon,” you are already behind the conversation.
AI Is Already More Embedded Than Most Organizations Realize
When most people hear “AI,” they still think of tools like ChatGPT, AI scribes, or documentation assistants.
But AI in healthcare now extends far beyond generative tools.
It is already embedded in mammography support systems, cardiac monitoring algorithms, stroke detection tools, retinal imaging analysis, predictive analytics, workflow prioritization systems, patient communication platforms, referral optimization tools, billing support workflows, and vendor software integrations that many organizations may already be using every day.
Some of these systems generate language or summaries. Others analyze patterns, identify abnormalities, prioritize workflows, predict outcomes, or support operational decision-making behind the scenes.
The problem is not that healthcare is using AI.
The problem is that many organizations still do not fully understand where AI already exists inside their operations.
That matters.
Because organizations cannot govern what they do not fully recognize.
How AI Actually Enters a Practice
This is the part most articles skip.
AI rarely enters a healthcare organization through a dramatic “we are implementing AI” moment.
Usually, it enters quietly.
A vendor schedules a demo. Leadership hears promises of improved efficiency, workflow support, faster turnaround, staffing relief, and operational optimization.
The AI component may be heavily marketed.
Or barely mentioned at all.
Then the tool gets approved.
Integrated.
Added to workflows.
Now AI is operational inside the business.
At that point, staff begin interacting with automated recommendations, prioritization systems, predictive workflows, communication support tools, and AI-assisted outputs that slowly become part of everyday operations.
And this is where the real problem starts.
Because many organizations still have not clearly defined who governs use, what the boundaries are, what should never enter a tool, who reviews outputs, who monitors risk, or who has authority to intervene when something goes wrong.
A vendor demo is not governance.
A signed contract is not governance.
And “our staff knows not to do anything inappropriate” is definitely not governance.
I have had conversations with physicians who genuinely believed their EHR vendor or software company was “handling the AI part.” That mindset is exactly where organizations get into trouble. Vendors may provide the technology, but they are not sitting inside your workflows watching how your staff uses it, what information gets entered into it, how outputs are being interpreted, or where operational shortcuts start forming under pressure.
The moment AI enters your workflow, leadership responsibility enters with it.
FDA Authorization Is Not the Same Thing as Operational Oversight
This distinction matters more than people realize.
The FDA’s AI-enabled device list identifies devices that have gone through an applicable FDA marketing pathway for a specific intended use.
That does not mean unrestricted implementation, unlimited use, automatic compliance, or that the technology no longer requires human accountability and operational oversight.
The FDA can authorize a device for marketing.
It cannot govern how every organization uses that technology inside real-world workflows.
That responsibility still belongs to leadership.
What Governance Should Actually Look Like
Governance is not anti-AI.
It is what keeps convenience from quietly becoming operational exposure.
At a minimum, organizations should have clearly defined approved use, role-based boundaries, staff training, vendor oversight, escalation processes, leadership accountability, and ongoing review of where AI is influencing workflows.
Because AI does not eliminate accountability.
It redistributes it across people, systems, workflows, vendors, and leadership.
And right now, too many healthcare organizations still have dangerously blurry lines around who owns what.
What Happens Without Governance?
Without governance, organizations often cannot fully account for what tools their staff is actually using, what information is being entered into those tools, where patient data is flowing after it leaves the practice, how vendors are retaining or using what gets submitted, or who is actually responsible for reviewing AI outputs before they influence a decision.
That is not a theoretical risk.
That is the current operational reality for a significant number of practices right now.
Staff may begin improvising with tools because they are trying to move faster.
Vendors may layer AI into platforms without leadership fully understanding the operational implications.
Different teams may begin using different tools with different standards and different levels of oversight.
And eventually, organizations end up with something healthcare should never be comfortable with:
Critical workflows being influenced by systems nobody is fully governing.
That is not modernization.
That is drift.
So Where Does Healthcare Even Start?
This is the part where many healthcare leaders freeze.
Because once you realize how deeply AI is already layered into modern healthcare operations, the next question becomes:
Now what?
And honestly, that is a fair question.
Most independent practices and even larger organizations were never handed a roadmap for this. AI entered healthcare faster than most operational governance models evolved to manage it.
The honest answer is that most practices do not need a complete overhaul. They need a clear starting point and someone who understands both the operational reality of running an independent practice and the compliance landscape well enough to tell them where to focus first.
That is exactly why I created the AI Safety Guide for Independent Healthcare Practices.
The guide was built to give practices a practical starting point for understanding where AI is already showing up, where exposure can quietly develop, and what operational guardrails should begin existing now, not after a problem surfaces.
Because this space is moving fast.
And it is only accelerating.
Volume 2 of the guide, coming soon, will go even deeper into AI-enabled medical devices, vendor oversight, workflow governance, operational accountability, and the growing gap between implementation and leadership readiness.
Healthcare does not need more panic around AI.
It needs smarter operational leadership around it.
And right now, most organizations are much earlier in that journey than they think.
Healthcare already adopted AI.
Leadership now has to decide whether governance is going to catch up before operational drift becomes operational damage.
If your practice is earlier in this journey than you realized, the AI Safety Guide for Independent Healthcare Practices is your starting point. Role-by-role rules, a ready-to-use policy template, an approved tools list, and a breach response protocol. Everything your team needs to start governing AI correctly this week.
Get Zentara’s AI Safety Guide
Practical, healthcare-specific guidance designed to help practices navigate AI governance, reduce avoidable risk, and build safer operational foundations.
Not ready yet?
Download the Free AI Safety Checklist to assess where your office currently stands.
Comments