A three-year-old girl's story exposes an uncomfortable truth: healthcare has become remarkably good at recording patients while surprisingly bad at connecting the information that follows them.
“AI won't replace doctors — it will work alongside them.”
— John
Whyte, MD, MPH, CEO of the American Medical Association
That idea is worth sitting with.
Because the future of healthcare should not be about making
medicine less human.
It should be about using technology to remove the work that
never required a human being in the first place.
And that distinction becomes much clearer when we look at
one little girl.
On October 6, 2025, Tabitha Black lost her
three-and-a-half-year-old daughter, Skylar Black, after a
two-and-a-half-year battle with childhood cancer.
Her grandfather, Galen Stewart, described Skylar as “the
light of our world.”
The story was reported this week by WAFF in Huntsville,
Alabama, as Skylar's family turned their grief into advocacy during Childhood
Cancer Awareness Month.
Skylar endured an extraordinary amount of treatment.
Chemotherapy.
Radiation.
Major surgeries.
Repeated sedation.
A stem-cell harvest.
A permanent chest tube.
Oxygen.
More medical appointments than most adults could imagine.
But those aren't the details that make the story stay with
you.
Skylar wore a duck costume to chemotherapy.
She helped nurses access her port.
And when she heard another child crying, she would go
comfort them.
Because she was immunocompromised, ordinary childhood
experiences could become dangerous.
The clinic became part of her world.
The children there became her friends.
Think about that.
A place many adults associate with needles, waiting rooms,
bad news and uncomfortable chairs became part hospital and part childhood for a
little girl.
And that is where Skylar's story becomes bigger than
childhood cancer.
Because healthcare has a peculiar habit.
We are very good at documenting what happened to a
patient while sometimes losing the patient inside the documentation.
Diagnosis.
Procedure.
Medication.
Authorization.
Encounter.
Note.
Code.
Claim.
Payment.
Denial.
Appeal.
Each item may be correct.
And yet the collection can still fail to tell one coherent
story.
Skylar wasn't a diagnosis.
She wasn't a claim.
She wasn't a billing code.
She was a daughter.
A granddaughter.
A friend.
A little girl in a duck costume trying to make another child
feel better.
So here's the question I want physicians and clinic owners
to consider:
When your healthcare system records a patient, does it
actually remember the patient's journey—or does it simply remember the
transactions generated by that journey?
Because those are not the same thing.
And that difference may explain more about healthcare's
operational problems than we have been willing to admit.
The Patient Has One Story. Your Software Has Twelve.
Patients don't experience healthcare in modules.
Nobody wakes up thinking:
“Today I will interact with the eligibility subsystem.”
Nobody says:
“After lunch, I expect to encounter the authorization
workflow.”
And no patient has ever happily announced:
“Excellent. My claim has entered the clearinghouse.”
Patients think:
I need help.
That's it.
Meanwhile, the healthcare organization starts opening tabs.
Scheduling.
Registration.
Eligibility.
Referral.
Authorization.
EHR.
Documentation.
Coding.
Clearinghouse.
Claim.
Payer portal.
Payment.
Denial.
Appeal.
Spreadsheet.
Email.
Phone call.
Another phone call.
And somewhere in the middle is usually a person named Susan
who knows what happened.
Susan is extremely important.
Susan also has vacation days.
This is one of the hidden risks in healthcare operations:
Institutional knowledge often lives inside people instead
of systems.
When Susan leaves, everyone discovers how much software they
supposedly had.
We Didn't Build Healthcare Around the Patient. We Built
It Around Transactions.
That may sound harsh.
But look at the structure.
A patient enters.
Information is captured.
That information is copied.
Then translated.
Then coded.
Then submitted.
Then interpreted.
Then paid.
Or denied.
Then somebody investigates why.
We call this a revenue cycle.
The patient calls it:
Tuesday.
The healthcare industry sees transactions.
The patient experiences a journey.
That distinction matters.
Because every time the journey is broken into disconnected
transactions, somebody eventually has to reconstruct it.
And reconstruction is expensive.
Sometimes financially.
Sometimes operationally.
Sometimes emotionally.
Here's My Contrarian Take
Healthcare doesn't have a data shortage.
It has a data continuity problem.
We have enormous amounts of information.
What we often lack is the ability to answer five simple
questions:
What happened?
When did it happen?
Why did it happen?
Who knew?
What happens next?
That sounds almost embarrassingly simple.
Which is probably why it gets overlooked.
Healthcare loves complexity.
Complexity sounds sophisticated.
Sometimes complexity is simply poor organization wearing a
suit.
The Denial Isn't Always the Problem
This is where I disagree with a lot of traditional
revenue-cycle thinking.
We tend to focus on the denial.
How many?
How fast can we resolve them?
How much can we recover?
What is the denial rate?
Important questions.
But they can also lead us into a trap.
Because a denial is often where the problem becomes
visible—not where it began.
Maybe eligibility was wrong.
Maybe authorization was missed.
Maybe a referral wasn't complete.
Maybe documentation didn't support the eventual claim.
Maybe information changed.
Maybe the payer made a questionable determination.
Maybe the practice made an error.
Maybe nobody knows.
Those are completely different situations.
Yet they can all end up as:
DENIED.
That's like diagnosing a patient with “fever.”
Technically accurate.
Clinically inadequate.
Stop Celebrating Faster Failure Recovery
This may be unpopular.
But I don't think the ultimate goal of revenue-cycle
technology should be:
“Let's become really fast at fixing things after they
break.”
Imagine a restaurant where 20% of the meals are routinely
burned.
Management proudly announces:
“We hired a faster waiter to handle customer complaints.”
The waiter deserves a raise.
The kitchen needs an investigation.
Healthcare sometimes does the opposite.
We build:
A denial queue.
An escalation queue.
A work queue.
A dashboard.
A tracking spreadsheet.
A task list.
A reminder.
A second reminder.
Then we measure how quickly people move through the queues.
Wonderful.
We have become extremely efficient at cleaning up messes.
But who is studying the kitchen?
The Most Expensive Word in Healthcare May Be “Later”
Eligibility problem?
We'll find out later.
Authorization issue?
We'll deal with it later.
Documentation problem?
Billing will catch it later.
Claim problem?
The payer will tell us later.
Denial?
We'll work it later.
Appeal?
Later.
Follow-up?
Later.
Eventually, later becomes expensive.
The earlier a problem is detected, the cheaper it usually is
to correct.
That is not uniquely a healthcare principle.
It is common sense.
Yet healthcare frequently discovers problems at the most
expensive point in the workflow.
Why?
Because that's where the system finally has enough
information to notice them.
That's backwards.
What If the System Knew Sooner?
Imagine a practice where the system could connect:
Patient.
Insurance.
Eligibility.
Referral.
Authorization.
Encounter.
Documentation.
Coding.
Claim.
Payer.
Payment.
Exception.
Resolution.
Not merely store them.
Connect them.
Now imagine the system noticing:
“This authorization requirement may create a problem.”
Before the appointment.
Not after the claim.
Or:
“This information conflicts with what was previously
captured.”
Before submission.
Not after denial.
Or:
“This claim resembles a previous exception.”
Before somebody spends 45 minutes researching it.
That's a different philosophy.
It's not:
How quickly can we clean up the mess?
It's:
How early can we see the mess forming?
That Is Where AI Gets Interesting
Healthcare AI has become obsessed with intelligence.
Can it write?
Can it summarize?
Can it code?
Can it predict?
Can it answer?
Those capabilities matter.
But I think we are asking the wrong question.
The better question is:
Can AI understand what needs to happen next?
Healthcare isn't just an information business.
It is a coordination business.
Someone has to do something.
At the right time.
With the right information.
For the right patient.
And somebody needs to know whether it happened.
That's workflow.
That's state.
That's ownership.
That's accountability.
And that's where AI becomes much more interesting than a
chatbot.
The Future of Healthcare AI Might Be Boring
I actually hope it is.
The most valuable AI in a clinic might never generate a
viral demo.
It might simply prevent:
One missed authorization.
One eligibility surprise.
One unnecessary phone call.
One duplicate entry.
One preventable denial.
One forgotten follow-up.
One hour of physician administrative work.
Nobody will clap.
There will be no dramatic music.
No robot walking into the exam room.
No humanoid announcing:
“Doctor, I have analyzed your dashboard.”
The claim will simply get paid.
And honestly?
That's pretty impressive.
Physicians Don't Need Another Dashboard
This is another hill I'm willing to stand on.
Healthcare has enough dashboards.
We have dashboards about dashboards.
If another screen appears asking physicians to monitor the
dashboard that monitors the dashboard, someone should probably call a meeting.
Actually, don't.
That's how we got here.
The goal shouldn't be more visibility for its own sake.
The goal should be:
Useful visibility at the moment a decision matters.
A physician doesn't need to know everything.
A practice manager doesn't need to know everything.
A biller doesn't need to know everything.
They need to know what matters now.
Physician Time Is Not Free
Recent AMA data reinforces something physicians already know
from experience: the workday doesn't end when the patient schedule ends.
Physicians reported substantial weekly hours devoted to indirect patient care
and administrative work, alongside continued burnout concerns.
And prior authorization is an especially obvious example.
Recent physician survey data reported by healthcare
organizations based on AMA research shows that practices handle roughly 39–40
prior authorization requests per physician per week, with about 13 hours
of physician and staff time devoted to the process.
Let's translate that.
Thirteen hours isn't “administrative overhead.”
It's thirteen hours.
It is payroll.
It is physician attention.
It is staff capacity.
It is delayed work.
It is patient frustration.
It is opportunity cost.
And it is time that cannot be recovered by telling a
physician to “work smarter.”
At some point, the system needs to stop asking the human to
absorb the inefficiency.
Don't Blame the Biller
This is important.
When a process is broken, the person closest to the problem
often becomes the person blamed for the problem.
That's backwards.
Your biller didn't create the payer ecosystem.
Your biller didn't design the EHR.
Your biller didn't create fragmented data.
Your biller didn't decide that five systems should each
contain a different version of the same patient.
And your biller shouldn't have to become a human search
engine.
Experienced billing professionals are valuable precisely
because they know how to navigate complexity.
The opportunity is to stop wasting that expertise on work a
system should handle.
Don't Fire Your Biller. Change the Job.
Good automation should move humans toward higher-value work.
Less:
Searching.
Copying.
Re-entering.
Reconciling.
Chasing.
Remembering.
More:
Judgment.
Exceptions.
Analysis.
Patient communication.
Payer strategy.
Quality control.
Process improvement.
Training.
The goal isn't:
Human versus AI.
The goal is:
Human judgment plus machine consistency.
A More Useful Definition of Automation
Here's my definition:
Automation is successful when the organization no longer
needs a human to repeatedly perform a low-value step.
Not:
“An AI generated something.”
Not:
“We added a chatbot.”
Not:
“We have a new dashboard.”
The question is:
Did the work disappear?
That's the test.
The Skylar Test
Now return to Skylar.
Imagine trying to reconstruct her healthcare journey.
Could you answer:
What happened?
When?
Why?
Who knew?
What was waiting?
What was missing?
Who owned the next step?
What changed?
What was communicated?
What happened afterward?
If the answer is:
“Probably. We'd have to check the EHR, billing system,
email, payer portal, authorization system and ask a few people.”
Then you don't have one connected patient journey.
You have fragments.
And fragments create work.
Skylar's story reminds us why that matters.
Because behind every fragmented record is a person whose
life is not fragmented.
The Data Is Not the Patient
This distinction sounds philosophical.
It isn't.
A diagnosis is data.
A claim is data.
An authorization is data.
A denial is data.
A clinical note is data.
But none of those things is the patient.
They are representations of the patient's experience.
The danger comes when the representation becomes more
visible than the person.
Healthcare starts optimizing:
The claim.
The code.
The metric.
The queue.
The dashboard.
The productivity number.
And eventually someone asks:
How did we improve the metric while making the experience
worse?
That's not a technology failure.
That's a measurement failure.
The Metric Problem
Healthcare loves measurable things.
That's understandable.
But measurable doesn't automatically mean meaningful.
Claims processed?
Easy.
Tasks completed?
Easy.
Messages answered?
Easy.
Denials resolved?
Easy.
But ask:
How much unnecessary work did we prevent?
Suddenly things become more interesting.
I would rather know:
How many exceptions never happened?
That is a harder metric.
It is also potentially more valuable.
The Metric I Want More Practices to Track
Preventable Exception Rate
Take the problems that required human intervention.
Then ask:
How many could reasonably have been prevented upstream?
Classify them.
Preventable.
Probably preventable.
Payer-driven.
Unknown.
Then track the percentage.
Because there is a huge difference between:
“We resolved 500 problems.”
and:
“We prevented 200 problems from being created.”
The first measures recovery.
The second measures system improvement.
Healthcare's Favorite Workaround
You know the one.
A spreadsheet.
Every organization has one.
Somewhere there is an Excel file named:
FINAL.xlsx
Then:
FINAL2.xlsx
Then:
FINAL_NEW.xlsx
Then:
FINAL_NEW_USE_THIS_ONE.xlsx
Then someone emails:
“Please don't edit the original.”
That spreadsheet is not the problem.
The spreadsheet is evidence.
It is a tiny protest against a system that doesn't quite
work.
Employees create workarounds because they are trying to get
the job done.
Instead of asking:
“Why is Susan using a spreadsheet?”
Ask:
“What unmet need is Susan solving with that spreadsheet?”
That question can reveal your real product requirement.
Start With the Work, Not the Software
If you're a physician-owner, don't begin with:
“What AI should we buy?”
Start with:
Where are we losing time?
Then:
Where are we losing information?
Then:
Where are we discovering problems too late?
Then:
Where does one person have to remember something the
system should remember?
Then:
Where are we doing the same thing twice?
Those questions are far more valuable than starting with a
vendor demo.
A 30-Day Practice Experiment
Days 1–5: Follow One Patient
Pick a common patient journey.
Follow it from scheduling through payment.
Watch the actual process.
Don't rely on the policy manual.
Reality is usually more creative.
Document every handoff.
Days 6–10: Count the Systems
How many systems does the patient journey touch?
EHR.
Scheduling.
Eligibility.
Payer portal.
Authorization.
Clearinghouse.
Billing.
Payment.
Email.
Spreadsheet.
Count them.
Then ask:
Why?
Days 11–15: Follow 25 Problems Backward
Take 25 denials, delays or exceptions.
Don't start at the denial.
Start at the beginning.
Trace backward.
Where did the problem actually begin?
You may discover the billing department is where the problem
became visible—not where it originated.
Days 16–20: Identify Preventable Work
Label every problem:
Preventable.
Possibly preventable.
Payer-driven.
Unknown.
This alone can expose patterns.
Days 21–25: Assign Ownership
Every important exception should have:
An owner.
A next action.
A deadline.
A dependency.
An escalation path.
If nobody owns the next step, the workflow owns nobody.
That's not a workflow.
That's a hope.
Days 26–30: Measure Again
Track:
Clean-claim rate
First-pass acceptance
Denial rate
Days in A/R
Manual touches
Exception-resolution time
Preventable exception rate
Staff hours spent on rework
Don't create another 40-metric dashboard.
Nobody needs that.
Pick the metrics that explain the economics and the patient
experience.
What OnnX Is Trying to Change
This is the thinking behind OnnX.
The core belief is simple:
Healthcare billing is often treated as a downstream
billing problem when much of the opportunity exists upstream.
The objective isn't another dashboard.
It isn't another chatbot.
It isn't another system that gives staff one more place to
look.
The larger opportunity is to connect the events that already
happen:
Patient → Insurance → Eligibility → Referral →
Authorization → Encounter → Documentation → Coding → Claim → Payer → Payment →
Exception → Resolution → Feedback
The patient experiences one journey.
The organization should be able to understand one journey.
That's the idea.
Not magic.
Not perfect prediction.
Connected context.
The OnnX Thesis
Most of the problem starts upstream.
If information is incomplete upstream, somebody downstream
eventually pays for it.
Sometimes the payer.
Sometimes the practice.
Sometimes the physician.
Sometimes the staff.
Sometimes the patient.
And sometimes everybody.
So instead of asking:
“How do we fix the denial?”
Ask:
“How could we have known sooner?”
That is a much more interesting question.
It shifts the conversation from recovery to prevention.
From transactions to relationships.
From documentation to context.
From reactive work to intelligent orchestration.
From:
Data → Work → Problem
toward:
Data → Context → Decision → Action → Outcome
The AI Test for Healthcare
Before buying another AI product, ask seven questions.
1. What does it actually know?
Not what the marketing page says.
What data does it really receive?
2. Does it understand sequence?
Healthcare events have order.
Context without sequence can be misleading.
3. Can it explain itself?
If the system recommends something important, can people
understand why?
4. What happens when it is wrong?
Every system will be wrong sometimes.
The question is whether failure is visible and manageable.
5. Who owns the decision?
AI should not become an accountability escape hatch.
6. Can the organization reconstruct what happened?
Auditability matters.
Especially when money, compliance or patient care is
involved.
7. Does it remove work?
This is the big one.
If AI creates another queue, another dashboard and another
login, congratulations.
You may have automated the creation of more work.
The Legal Question Nobody Should Skip
AI doesn't magically transfer responsibility.
If a system touches protected health information, healthcare
organizations still need appropriate privacy, security and contractual
safeguards.
Depending on the application, that can involve:
HIPAA.
Business associate agreements.
Access controls.
Audit logs.
Data minimization.
Security monitoring.
Vendor oversight.
Retention policies.
Incident response.
And if AI participates in a consequential workflow,
organizations should know:
What information it uses.
What it is allowed to do.
What it cannot do.
How confidence is handled.
When a human reviews the result.
How an error is corrected.
“Because the AI said so” is not a governance model.
The Ethical Question Is Bigger
Healthcare leaders should also ask:
What are we optimizing for?
Speed?
Revenue?
Staff productivity?
Patient access?
Clinical quality?
Convenience?
Sometimes those objectives align.
Sometimes they don't.
An efficient system can still produce a bad experience.
A fast authorization can still be wrong.
A perfectly optimized claim can still represent poor care.
Technology should reduce unnecessary friction without
reducing human judgment.
Efficiency is not the same thing as humanity.
The Real AI Debate Isn't Human Versus Machine
That debate is already becoming boring.
The better question is:
What should humans stop doing?
Physicians should spend more time exercising judgment.
Nurses should spend more time caring.
Staff should spend more time solving real problems.
Billers should spend more time on complex exceptions.
Practice owners should spend more time making decisions.
Machines are very good at repetitive pattern recognition.
Machines are very good at monitoring.
Machines are very good at routing.
Machines are very good at remembering.
Humans are very good at context, judgment, relationships and
responsibility.
The opportunity is obvious.
Automate the friction surrounding humanity.
Don't automate humanity itself.
What Healthcare Leaders Should Stop Saying
“That's just how insurance works.”
Maybe.
But which part?
Payer policy?
Practice workflow?
Bad information?
Missing documentation?
Unclear ownership?
Don't use “insurance” as a universal explanation.
“Our biller catches it.”
Maybe.
But why does the biller have to catch it?
The question isn't whether Susan can fix it.
The question is whether Susan should have to.
“We have all the data.”
Great.
Now answer:
Can you connect it?
“We just need better AI.”
Maybe.
Or maybe you need a better workflow.
AI cannot fix a process nobody understands.
The Healthcare Innovation Trap
There is a strange phenomenon in healthcare technology.
We build software to solve a problem.
The software creates a new workflow.
The workflow creates new tasks.
The tasks create new notifications.
The notifications create alert fatigue.
Then someone builds AI to summarize the alerts.
Eventually we need AI to explain the AI.
At some point, perhaps we should stop and ask:
What if we simply removed the original problem?
That is not anti-technology.
It is pro-design.
The Best Technology May Be the Technology Nobody Notices
Imagine a practice where:
An authorization problem is identified before the
appointment.
An eligibility conflict is caught before the claim.
A missing piece of information is surfaced before
submission.
A likely exception is routed to the right person.
A staff member knows what needs attention without searching
five systems.
A physician doesn't discover an administrative problem three
days later.
A patient doesn't have to explain the same situation four
times.
Nobody calls it revolutionary.
That's okay.
Healthcare doesn't need more impressive technology.
It needs more technology that quietly works.
Why Skylar Belongs at the Center of This Conversation
Skylar's story is not a metaphor for billing.
It shouldn't be turned into one.
Her story belongs here for a different reason.
It reminds us of the thing every healthcare system is
supposed to protect:
the person behind the information.
Skylar's medical record could contain thousands of pieces of
information.
But none of them alone explains who she was.
Her mother, Tabitha Black, knew her differently.
Her grandfather, Galen Stewart, knew her differently.
The nurses knew her differently.
The children around her knew her differently.
They knew the little girl.
The systems knew the data.
Healthcare needs both.
But the data must serve the person.
Never the other way around.
The Most Important Question Isn't “Can AI Do It?”
That's the question everybody asks.
Can AI code?
Can AI document?
Can AI authorize?
Can AI predict?
Can AI answer?
I think there is a better question:
Should a human have to do this at all?
If the answer is no, automate it.
If the answer is yes, make the human's job easier.
If the answer is unclear, investigate.
That's a much healthier AI strategy.
The Future of Healthcare Will Be Won Upstream
I believe the next major opportunity in healthcare
operations won't come from building a better cleanup crew.
It will come from preventing the mess.
Better information at capture.
Better connections between events.
Better understanding of workflow state.
Better ownership.
Better timing.
Better feedback.
The goal is not merely:
Fix the denial.
It is:
Understand why the denial happened.
Then:
Prevent the next one.
That's how systems learn.
Five Questions Every Practice Owner Should Ask This Month
1. Where are we repeatedly doing the same work?
Repetition is a signal.
2. Where do we discover problems too late?
Late discovery is expensive.
3. Where does one employee carry knowledge that the
system should carry?
That is institutional risk.
4. Which exceptions could have been prevented?
That's where improvement begins.
5. What does the patient experience while all of this is
happening?
Because operational efficiency that creates patient
frustration isn't really efficiency.
It's just shifting the cost.
What If Healthcare's Biggest Problem Is Not
Documentation?
Here's another uncomfortable thought.
We often blame documentation.
Too much documentation.
Too little documentation.
Wrong documentation.
Incomplete documentation.
Poor documentation.
And yes, documentation matters.
But perhaps documentation is sometimes just the most visible
symptom.
The deeper problem may be that healthcare is not
operating as a real-time connected system.
The patient has an encounter today.
Documentation may happen later.
Coding may happen later.
Claims may be submitted later.
The payer may respond later.
The denial may arrive weeks later.
Then someone tries to reconstruct what happened.
That's not a data problem alone.
It's a feedback-loop problem.
The system learns too slowly.
Healthcare Should Learn While the Work Is Happening
Imagine a system that doesn't wait for the denial to teach
you.
It learns from:
Eligibility.
Authorization.
Documentation.
Coding.
Claims.
Payments.
Denials.
Appeals.
Resolutions.
Then feeds those lessons upstream.
That creates a loop.
Capture → Detect → Decide → Act → Measure → Learn →
Improve
That's much closer to how modern intelligent systems should
work.
Not:
Work → Wait → Denial → Panic → Spreadsheet
Although, to be fair, the second system has had an
impressive run.
Proof Doesn't Have to Mean a Perfect AI Model
Healthcare leaders should also rethink what “AI accuracy”
means.
A model can be statistically accurate and operationally
useless.
What matters is whether the organization can understand:
What the system saw.
What it inferred.
What it recommended.
What happened next.
Whether the recommendation was correct.
What changed.
That is reconstructability.
And I think reconstructability will become increasingly
important as AI moves deeper into healthcare operations.
Because when something goes wrong, “the model predicted it”
isn't an explanation.
A Better Healthcare AI Scorecard
Don't ask only:
How accurate is it?
Ask:
How much work did it eliminate?
How early did it detect the problem?
How often did humans override it?
Why did humans override it?
How many preventable exceptions disappeared?
Did staff trust it?
Could the organization audit it?
Did the patient experience improve?
Those questions measure whether AI is actually improving the
system.
Final Thoughts: Stop Building Faster Band-Aids
Skylar Black's story began with something no healthcare
dashboard can fully represent.
A little girl.
A mother.
A grandfather.
A family.
A community.
A child who, despite everything she was going through, still
found the instinct to comfort another child.
Her family is now turning grief into purpose and asking
people not to look away from children still fighting cancer.
That message applies beyond childhood cancer.
Healthcare leaders should not look away from the things
their organizations have quietly normalized.
The spreadsheet.
The duplicate entry.
The forgotten authorization.
The recurring denial.
The mysterious payer portal.
The physician doing administrative work at night.
The employee who knows everything because the software knows
almost nothing.
Those are not just inconveniences.
They are clues.
They tell us where the system is compensating for itself.
And perhaps that is the real opportunity.
Not to build another layer on top of healthcare's
complexity.
But to remove some of the complexity underneath it.
The Question I Want You to Answer
Think about your practice.
What is the administrative problem everyone has accepted as
normal?
Not the biggest problem.
Not the most expensive problem.
The one everyone simply shrugs at and says:
“That's just how we do it.”
Maybe it's a spreadsheet.
Maybe it's a payer portal.
Maybe it's an authorization.
Maybe it's a recurring denial.
Maybe it's a staff member who has become the unofficial
operating system.
Maybe it's you.
Now ask:
What would happen if we stopped accepting it?
Not:
“What software should we buy?”
Not:
“Can AI fix it?”
First ask:
Why does this problem exist?
That's where the interesting work begins.
Three Things I Hope You Remember
The patient has one story. Your systems should be able to
connect it.
The denial is often where the problem becomes visible—not
where it began.
The best healthcare technology doesn't replace human
connection. It creates more room for it.
Get Involved
I want to hear from physicians, clinic owners, practice
managers, billers and healthcare operators.
What is the strangest workaround your practice has
accepted as normal?
Tell me in the comments.
If you've discovered a way to eliminate one of these
problems, share it.
If your practice is still fighting one, share that too.
Because there is a good chance your “unique” problem isn't
unique at all.
Comment with the workaround.
Challenge the assumption.
Repost this for someone who needs to see it.
Healthcare improves when people stop quietly compensating
for broken systems and start asking better questions.
Continue the Conversation
Healthcare innovation should be practical.
It should make work clearer.
It should reduce unnecessary friction.
And it should give physicians, staff and patients more room
to focus on what matters.
For more perspectives on healthcare operations, medical
billing, medical technology, practice management and intelligent automation,
continue the conversation through Dr. Cham's online channels.
Knowledge creates possibility. Applied knowledge creates
progress.
Start with one question.
Challenge one assumption.
Improve one workflow.
Then share what you learn.
Free Resource for Physicians and Clinic Owners
Want a practical place to start?
Visit the Featured section of my LinkedIn profile for
a free resource.
No signup required.
Read it.
Question it.
Use it with your team.
And tell me what you discover.
About the Author
Dr. Daniel Cham is a physician, medical consultant
and healthcare technology entrepreneur with experience spanning medical
practice, healthcare management, medical technology and medical billing.
His work focuses on the practical intersection of healthcare
operations, medical billing, data, workflow and intelligent automation.
As founder of OnnX, he explores how healthcare
organizations can move from fragmented, reactive processes toward more
connected and predictable systems.
His perspective comes from seeing healthcare from multiple
sides: as a physician, as a practice operator and as a technology entrepreneur.
The question behind much of his work is simple:
Can healthcare technology make the system easier for
humans to operate without making healthcare less human?
Connect with Dr. Cham on LinkedIn:
Dr.
Daniel Cham on LinkedIn
Disclaimer
This article is provided for general educational and
informational purposes only. It does not constitute medical, legal, financial,
compliance or other professional advice.
Healthcare organizations should consult appropriately
qualified professionals regarding their individual clinical, legal, regulatory,
privacy, security, payer and operational circumstances.
References and Further Reading
The September 14, 2026 report by Sarah Grace Kennedy tells
the story of Skylar Black, her mother Tabitha Black, grandfather Galen
Stewart, and the family's effort to turn grief into advocacy for children
with cancer.
2. STAT — AMA CEO
John Whyte on AI and the future of medicine
A current September 2026 perspective from AMA CEO John
Whyte, MD, MPH, examining the distinction between automating medical tasks
and automating medicine itself.
3. American
Medical Association — Physician workload, technology and administrative burden
AMA research documents the continuing amount of physician
time devoted to indirect clinical and administrative work and the need to make
technology an asset rather than another burden.
4. American
Medical Association / 2026 prior-authorization research
Recent 2026 reporting on physician prior-authorization
burden highlights the significant amount of time practices spend managing
authorization requirements and the continuing effect on physician and staff
workload.
One Last Thought
Skylar's family is asking people not to look away.
Maybe healthcare leaders should take that seriously.
Don't look away from the patient behind the data.
Don't look away from the employee who has created a
workaround because the system failed to.
Don't look away from the physician who is doing
administrative work at 9 p.m.
Don't look away from the recurring denial everyone has
decided is inevitable.
And don't look away from the possibility that the problem
isn't the person trying to fix the system.
Maybe the system is asking too much of the person.
That is the opportunity.
Not to build a machine that replaces the human.
But to build a system that finally remembers what the human
is there to do.
Care.
Think.
Decide.
Connect.
And, occasionally, go home on time.
#Healthcare #Physicians #MedicalPractice #MedicalBilling
#RevenueCycleManagement #HealthcareLeadership #HealthcareInnovation
#HealthcareAI #HealthTech #PracticeManagement #PatientExperience
#PhysicianEntrepreneur #IndependentPractice #HealthcareTransformation #OnnX
