The Healthcare Problem We Keep Trying to Automate Instead of Eliminate
“Whether the technology does the work, or whether we hire
people to do the work the technology creates.” — Brian Hudes, MD
Healthcare doesn’t need more AI to manage the mess—it
needs intelligence that prevents the mess from happening.
That may be one of the most important questions in
healthcare right now.
Not because AI is coming.
AI is already here.
The more uncomfortable question is whether we are using it
to remove work or simply to create a faster way of doing work we should
have eliminated years ago.
Brian Hudes, MD, a gastroenterologist and longtime physician
leader, made that argument this week while examining the administrative bloat
inside hospitals. His point is deceptively simple: technology transformed many
industries by making certain administrative jobs unnecessary. Healthcare, in
many cases, used technology differently. Instead of eliminating the work, it
often redistributed it to physicians, nurses and other highly trained
professionals.
And then I read the story of Enoch Machora.
It stopped me.
Because Machora understands healthcare from both sides of
the bed.
He was a caregiver.
Then he became a patient.
Then he became a caregiver again.
And somewhere in between those three identities is a lesson
about the future of healthcare that has almost nothing to do with another shiny
AI dashboard.
It has to do with what we choose to make people do.
The Man Who Saw Healthcare From the Other Side
In July 2025, Enoch Machora arrived at Providence Sacred
Heart Medical Center in Spokane, Washington, after suffering a massive
stroke.
He knew healthcare.
He had worked as an in-home medical caregiver.
He understood what it meant to help another person through
vulnerability, uncertainty and dependence.
But suddenly he was the one in the bed.
Machora lost movement in parts of his body.
He lost his ability to see.
He lost his ability to speak.
He spent 15 days in the hospital, moving in and out of
consciousness and sometimes unable to communicate.
But his hearing remained.
He could hear the nurses talking about his care.
He could hear discussions about medications.
He could hear people around him even when he could not
answer them.
And he remembers one nurse staying at his bedside for 24
hours so he would not be alone.
The nurse's name was not published by Providence, and that
matters too.
Because this is not a story about a celebrity physician or a
famous hospital executive.
It is about ordinary healthcare workers doing
extraordinarily human things.
Machora later said:
“In one of the most vulnerable periods of my life, they
cared for me with dignity.”
That sentence deserves more attention than most healthcare
technology announcements.
Because dignity is not an efficiency metric.
It does not appear neatly in a dashboard.
It does not generate a productivity score.
It is difficult to automate.
And patients know when it is there.
They also know when it isn't.
Then Something Unexpected Happened
During his recovery, Machora says he prayed and asked what
he was supposed to do with the life he had been given.
He believed he received an answer:
Work at Providence.
He eventually transferred to Providence St. Luke's
Rehabilitation Medical Center, where he spent five months recovering.
He regained the use of his limbs.
He relearned how to write.
He relearned how to talk.
By the end of 2025, he enrolled in a certified nursing
assistant program.
He was moving toward a new chapter.
Then healthcare did what healthcare sometimes does.
It complicated the plot.
Machora accepted a job at an assisted-living facility in
Stevens County.
Three days into the job, he became ill again.
He was rushed to Providence St. Joseph's Hospital in
Chewelah.
He learned he was having another stroke.
He was airlifted back to Sacred Heart Medical Center in
Spokane.
And as the helicopter landed, Machora says he again sensed
the same message:
You are supposed to work here.
He spent another month recovering.
Then he applied to work at Sacred Heart.
He was accepted.
Today, Enoch Machora works in the Pediatric Emergency
Department.
He walks through the doors not as the patient he once was,
but as the caregiver he had hoped to become.
He says it feels like home.
That is a remarkable human story.
But there is another story hiding inside it.
And it is the story I think healthcare technology should be
paying attention to.
The Healthcare System Has a Strange Obsession With Adding
Work
Healthcare loves the word innovation.
We have innovative platforms.
Innovative workflows.
Innovative AI.
Innovative revenue-cycle management.
Innovative clinical documentation.
Innovative automation.
Innovative interoperability.
Innovative dashboards.
At some point, however, we should ask a slightly impolite
question:
What if the innovation is being applied to the wrong
problem?
Imagine a restaurant where the kitchen accidentally sends
every order to the wrong table.
Management responds by purchasing a faster order printer.
Then another printer.
Then an AI printer.
Then a dashboard to monitor printer performance.
Then a consultant recommends hiring someone to reconcile the
printers.
Everyone celebrates the digital transformation.
Meanwhile, the kitchen is still sending the food to the
wrong table.
That is not transformation.
That is faster confusion.
Healthcare has versions of this problem everywhere.
A patient provides information.
The information gets entered into one system.
Someone re-enters it into another.
Another system cannot read the first system correctly.
Someone calls someone else.
A fax appears.
An authorization is requested.
A payer asks for information already available somewhere
else.
A staff member searches for it.
A physician gets involved.
A claim is submitted.
The claim is denied.
Someone researches the denial.
Someone appeals it.
Someone checks the status.
Someone documents the call.
Someone follows up.
Someone sends another message.
Someone waits.
Then somebody asks why the practice needs so many people.
We call this a workflow.
Sometimes it is.
Sometimes it is just a pile of historical decisions nobody
has had the courage to question.
The Administrative Burden Problem Is Bigger Than Burnout
We often describe administrative burden as a physician
wellness issue.
That is true.
But it is incomplete.
Administrative burden is also a data problem, a workflow
problem, a financial problem, a patient-access problem, an operational
problem, and increasingly, an AI architecture problem.
The American Medical Association's 2026 physician survey on
prior authorization makes the scale difficult to ignore.
Physicians reported completing an average of 40 prior
authorizations per week.
The process consumes an average of 13 hours of physician
and staff time every week.
Ninety-five percent said prior authorization delays
necessary care.
Ninety-two percent said it negatively affects clinical
outcomes.
Ninety-four percent said it contributes to burnout.
And 26% reported that prior authorization had contributed to
a serious adverse event, including hospitalization, permanent impairment or
death.
Those are not merely productivity statistics.
They are signals that the system is consuming scarce human
attention.
And attention is a clinical resource.
We treat it as if it were infinite.
It isn't.
Here's the Contrarian Part
I don't think healthcare has an AI shortage.
I think healthcare has a problem-definition shortage.
We are increasingly capable of automating individual tasks.
But we are not nearly as good at asking whether the task
should exist.
That distinction matters.
Because AI can make bad processes dramatically more
efficient.
And that is not necessarily progress.
AI + broken workflow = faster broken workflow.
AI + fragmented data = faster navigation through fragmented
data.
AI + unnecessary authorization = faster authorization
paperwork.
AI + inconsistent documentation = faster processing of
inconsistent documentation.
AI + poor information architecture = a very intelligent
person searching a very messy filing cabinet.
That last one sounds funny.
Until the filing cabinet contains someone's healthcare.
We Keep Starting at the Claim
This is where the medical billing industry gets particularly
interesting.
The claim is visible.
The denial is visible.
The unpaid balance is visible.
The rejected code is visible.
The appeal is visible.
So naturally, we build tools around them.
But visibility is not the same thing as causality.
The claim may be where the problem becomes visible.
It may not be where the problem began.
A denial could be caused by something that happened much
earlier.
A registration issue.
Eligibility information.
A missing authorization.
A referral problem.
A documentation inconsistency.
A charge-capture failure.
A coding mismatch.
A payer-specific requirement.
A data-field problem.
A workflow handoff.
A human assumption.
A software integration.
Or simply information that existed but never made it to the
person who needed it.
Most of the problem starts upstream.
By the time the claim reaches the billing system, the
organization may already be trying to recover from decisions made hours, days
or weeks earlier.
Then we call the downstream correction process "revenue
cycle management."
Sometimes it feels more like revenue-cycle archaeology.
Digging through the ruins to figure out what happened.
The Physician Has Become the API
Here is another uncomfortable thought:
The physician has become the API.
Not intentionally.
Not because anyone designed it that way.
But because physicians increasingly sit at the intersection
of disconnected systems.
The EHR needs something.
The payer needs something.
The pharmacy needs something.
The patient needs something.
The authorization team needs something.
The billing system needs something.
The specialist needs something.
The physician becomes the human translator between systems
that cannot reliably communicate.
That is not physician empowerment.
It is an architecture failure disguised as professional
responsibility.
And it has a cost.
Every time a clinician has to stop practicing medicine to
resolve an avoidable information problem, the system has effectively converted
clinical expertise into administrative infrastructure.
That is a terrible use of a scarce resource.
What Enoch Machora's Story Changes
This is where Machora's story becomes more than a feel-good
healthcare story.
He knows what it feels like to be unable to speak.
He knows what it feels like to be unable to explain what is
happening.
He knows what it feels like to depend on another person.
He also knows what it feels like to be the person
responsible for helping someone else.
That creates a perspective technology cannot manufacture.
Machora said:
“I know what it feels like to lie in a hospital bed, unable
to move, speak, or explain what is happening inside you.”
Then he described the value of a gentle voice, a patient
explanation and someone who cares.
That is the test.
Not whether the hospital has an AI strategy.
Not whether the software has 47 integrations.
Not whether the dashboard has real-time analytics.
The question is:
Does the system give the human caregiver more capacity to
care?
If yes, keep going.
If not, what exactly are we optimizing?
The Technology Paradox
Technology was supposed to give healthcare professionals
more time.
Instead, technology has sometimes given them more places
to work.
More inboxes.
More alerts.
More portals.
More passwords.
More fields.
More notifications.
More documentation.
More reconciliation.
More messages.
More clicks.
More "action items."
We digitized the paper.
Then we digitized the fax.
Then we digitized the phone call.
Then we digitized the inbox.
Then we built AI to help us manage the digital mess.
There is a joke hiding in there somewhere.
Actually, the joke is the business model.
We Are Optimizing Around Noise Instead of Removing It
This is one of the biggest mistakes in healthcare
technology.
We see friction.
We build a workflow around the friction.
Then we build automation around the workflow.
Then we build analytics around the automation.
Then we build AI around the analytics.
Eventually we have a very sophisticated system for managing
the original problem.
But the original problem is still there.
We are optimizing around noise instead of removing it.
The better question is:
Why did the noise exist in the first place?
That question changes everything.
The New Healthcare Technology Test
Before buying another AI tool, ask five questions.
1. Does this eliminate a task?
Not "make it easier."
Not "reduce clicks."
Not "summarize it."
Does the task disappear?
2. Does it prevent downstream work?
If a technology makes a downstream correction easier but
does nothing to reduce the number of corrections, its value may be smaller than
advertised.
3. Does it improve the information at the moment of
capture?
Bad information becomes expensive later.
The earlier you improve it, the fewer downstream systems
have to compensate for it.
4. Does it preserve context?
A data point without context is often just another reason
for a human to investigate.
5. Does it give time back to the caregiver?
This is the Enoch Machora test.
If the answer is no, keep asking questions.
The Enoch Machora Test
I would argue that every healthcare technology company
should have a version of this test.
Before deploying a new system, ask:
Will this give the caregiver more capacity to care?
Not more capacity to click.
Not more capacity to process forms.
Not more capacity to answer messages.
Not more capacity to reconcile databases.
Not more capacity to move work from one queue to another.
More capacity to care.
That distinction sounds philosophical.
It isn't.
It is operational.
It affects staffing.
It affects retention.
It affects patient experience.
It affects revenue.
It affects safety.
It affects burnout.
It affects whether clinicians want to stay.
And it affects whether healthcare technology actually
improves healthcare.
Three Experts, Three Warnings
Brian Hudes, MD: Don't Automate the Bureaucracy
Hudes' October 6 essay makes a point healthcare technology
leaders should take seriously: the question is whether technology actually does
the work or whether it creates another layer of work around the technology.
That is an important distinction for AI builders.
A successful implementation should eventually make itself
less noticeable.
If the organization needs an entire new bureaucracy to
supervise the AI bureaucracy, something went wrong.
Bobby Mukkamala, MD: Trust Is Now an Operational Metric
AMA President Bobby Mukkamala, MD, recently described
physician trust in voluntary insurer promises as deeply eroded after years of
unfulfilled commitments.
The AMA's survey found that only 33% of physicians
believed the latest insurer prior-authorization pledge would make a meaningful
difference.
That is not simply a political problem.
It is a technology adoption problem.
Healthcare workers have seen too many "solutions"
arrive as additional work.
If we want clinicians to trust AI, we cannot simply tell
them it will save time.
We have to demonstrate it.
The Doximity Physician Perspective: AI Can Help—If We Use
It Carefully
Doximity's 2026 State of AI in Medicine report found
substantial physician optimism about AI's potential.
Three-quarters of physician AI users surveyed reported
reduced administrative burden and improved job satisfaction.
Sixty-nine percent reported better patient care and
outcomes.
And 91% of physicians surveyed believed AI could reduce
administrative workload and create more time for patient care.
That is the good news.
The caution is equally important.
Physicians still cite concerns about accuracy and
reliability, and responsible implementation requires clinical oversight and
governance.
In other words:
Doctors are not anti-AI.
They are anti-useless AI.
There is a difference.
The Statistics Tell a Strange Story
Consider the contradiction.
AI adoption is increasing.
Physicians increasingly believe AI can reduce administrative
work.
Yet administrative burden remains enormous.
Prior authorization alone consumes an average of 13 hours of
physician and staff time each week.
Physicians complete approximately 40 prior authorizations
per week.
Ninety-four percent say the process contributes to burnout.
And 74% say denials have increased over the past five years.
So perhaps the question is not:
"How much AI are we using?"
Perhaps the better question is:
"How much unnecessary work still exists after we use
AI?"
That is a much harder metric.
It is also a much more useful one.
A Better Definition of Productivity
Healthcare productivity is often measured by volume.
Patients seen.
Claims submitted.
Notes completed.
Messages answered.
Authorizations processed.
Collections generated.
But there is another metric that deserves attention:
Avoided work.
How many calls never had to happen?
How many denials never occurred?
How many authorization requests never needed to be
submitted?
How many duplicate entries disappeared?
How many staff escalations were prevented?
How many physician interruptions never happened?
How many patient calls were unnecessary because the
information was already available and correct?
Zero is a very efficient number.
We should measure it more often.
The Hidden Revenue Leakage Nobody Puts on the Dashboard
Healthcare organizations understand financial leakage.
They track:
- denied
claims
- unpaid
claims
- underpayments
- missed
charges
- coding
errors
- delayed
collections
- authorization
failures
But there is another form of leakage.
Attention leakage.
Every avoidable task consumes human attention.
A staff member spends 20 minutes looking for information.
A physician spends 10 minutes responding to an avoidable
question.
A biller spends 15 minutes researching why something failed.
A patient spends another 20 minutes calling the office.
Multiply that across a practice.
Then multiply it across thousands of practices.
The financial value of that lost attention becomes enormous.
And unlike a denied claim, attention leakage often never
appears as a line item.
This Is a Data Structure Problem, Not Just a Workflow
Problem
This distinction is critical.
A workflow tells us what happens next.
Data structure determines what information is available
when it happens.
If the underlying information is fragmented, incomplete,
inconsistent or trapped in another system, the workflow will compensate.
People become the glue.
That works until it doesn't.
And healthcare has become remarkably dependent on human
glue.
The system is compensating for variability that shouldn't
exist.
What OnnX Is Trying to Do Differently
This is also where my own work with OnnX comes in.
The premise is simple:
Don't begin with the claim. Begin with the conditions
that eventually create the claim.
OnnX is being developed around an upstream intelligence
model for medical practices.
The objective isn't to replace the EHR.
It isn't to replace the billing system.
It isn't to replace the biller.
It isn't to tell physicians how to practice medicine.
The goal is to understand the practice's context well enough
to help prevent avoidable downstream administrative and billing noise.
That means understanding the practice's clinicians.
Its workflows.
Its systems.
Its billing behavior.
Its payer environment.
Its historical patterns.
Its preferences.
Its recurring exceptions.
Its outcomes.
The technology should adapt to the practice—not force the
practice to become a perfect specimen of someone else's workflow diagram.
Capture → Structure → Preserve → Propagate → Act → Learn
That is the basic philosophy.
Capture
Get the right information at the point where it is created.
Structure
Make the information usable rather than merely storing it.
Preserve
Keep context attached to the information.
Propagate
Move the relevant context to the next person or system that
needs it.
Act
Trigger the appropriate action before the problem becomes
expensive.
Learn
Use outcomes to improve the practice-specific intelligence
over time.
The objective is not more automation for its own sake.
The objective is less avoidable work.
The Local Practice Intelligence Idea
Healthcare practices are not interchangeable.
Two clinics may use the same EHR and have completely
different operational realities.
Different physicians.
Different payer mixes.
Different referral patterns.
Different staffing.
Different specialties.
Different documentation habits.
Different patient populations.
Different workflows.
Different recurring failure points.
A generic AI model may know a tremendous amount about
healthcare.
That does not mean it knows your practice.
The interesting layer is the one that learns the local
environment.
Over time, a practice-specific intelligence layer should
understand what normal looks like.
Then it can identify what is unusual.
That changes AI from:
"Here is an answer."
to:
"Something is different from how this practice
normally operates, and here is why it may matter."
That is a much more interesting form of intelligence.
The Biggest Pitfall: Automating the Wrong Thing
Here are the traps I would watch closely.
Pitfall 1: AI as a Band-Aid
If the underlying process is broken, AI may simply hide the
symptoms.
Pitfall 2: Measuring clicks instead of outcomes
Reducing clicks is nice.
Preventing a denial is better.
Preventing the need for the denial is better still.
Pitfall 3: Building another dashboard
Healthcare does not need another place to look.
It needs fewer reasons to look.
Pitfall 4: Ignoring upstream causes
If your system only analyzes the claim after it fails,
you're already late.
Pitfall 5: Treating every practice as identical
A generic workflow can be useful.
A generic reality is dangerous.
Pitfall 6: Confusing automation with autonomy
Automating a task does not necessarily mean the system
understands the context surrounding the task.
Pitfall 7: Forgetting the human
Healthcare is not a logistics company with patients
attached.
The human relationship is the product.
Technology should protect it.
What Healthcare Founders Should Learn From Enoch Machora
If I were building healthcare technology today, I would ask
a different set of questions.
Not:
What task can AI perform?
But:
What work should no longer exist?
Not:
How do we automate the workflow?
But:
Why does this workflow exist?
Not:
How do we process more claims?
But:
How do we create fewer claims that require intervention?
Not:
How do we make staff more productive?
But:
How much unnecessary work can we remove from their day?
And perhaps most importantly:
What happens to the human capacity we create when the
unnecessary work disappears?
That final question is where healthcare technology gets
interesting.
Because the point of efficiency isn't to make everyone run
faster.
The point is to give people something back.
Time.
Attention.
Judgment.
Presence.
Energy.
Human connection.
A Practical Six-Step Approach for Medical Practices
If you're a physician, practice administrator or healthcare
operator, you don't need to wait for the perfect AI platform.
Start here.
Step 1: Track recurring friction
For 30 days, record the administrative problems that
repeatedly interrupt your practice.
Don't categorize them yet.
Just capture them.
Step 2: Identify where each problem begins
Don't start with the denial.
Ask:
Where did the information first become incomplete,
inconsistent or unavailable?
Step 3: Calculate the human cost
Estimate staff time.
Physician time.
Patient time.
Management time.
Then calculate the frequency.
Step 4: Separate preventable from unavoidable
Not every administrative task is bad.
Some are required.
Some are clinical.
Some are regulatory.
Some exist because healthcare is complex.
Others exist because nobody has redesigned the process.
Find the fourth category.
Step 5: Fix upstream information
Before buying automation, ask whether better information at
the beginning would prevent the downstream work.
Step 6: Measure avoided work
Track:
- denied
claims prevented
- authorization
requests avoided
- duplicate
entries eliminated
- staff
hours recovered
- physician
interruptions prevented
- patient
callbacks reduced
- unresolved
tasks reduced
- days
to resolution
- downstream
corrections avoided
Then ask the most important question:
What work disappeared?
The Metrics That Actually Matter
Healthcare technology companies love utilization metrics.
Logins.
Users.
Clicks.
API calls.
Messages.
Tasks completed.
Those metrics have their place.
But healthcare operators should also track friction
metrics.
Administrative Work Created
How many tasks does the system generate?
Administrative Work Avoided
How many tasks disappeared?
Time to Resolution
How long does it take to resolve an issue?
First-Pass Success
How often does information move through the system correctly
the first time?
Escalation Rate
How often does a routine issue require human intervention?
Rework Rate
How often must someone redo something?
Exception Rate
How often does the workflow leave the expected path?
Physician Interruption Rate
How frequently is clinical staff pulled into administrative
work?
Patient Friction
How many unnecessary calls, delays or repeated requests
reach the patient?
Revenue Leakage
How much financial impact results from preventable upstream
failures?
These metrics tell a different story.
They measure whether the system is becoming simpler.
Legal and Ethical Considerations
There is an important boundary here.
Reducing administrative work does not mean removing
accountability.
Healthcare AI operates inside a highly regulated environment
involving patient privacy, clinical responsibility, payer requirements,
documentation standards, fraud-and-abuse laws, billing rules and professional
obligations.
Automation cannot become an excuse for poor documentation.
AI cannot manufacture clinical facts.
A system should not silently alter information simply
because the change improves reimbursement.
And "the algorithm did it" is not a defense.
The more consequential the decision, the more important it
becomes to understand:
What information did the system use?
What did it infer?
What did it change?
Who approved the action?
Can the decision be explained later?
That is why responsible healthcare AI needs traceability,
governance and appropriate human oversight.
The goal isn't to remove humans from every decision.
The goal is to remove humans from the decisions and tasks
that never should have required their attention in the first place.
The AI Paradox We Need to Resolve
There are two possible futures.
In the first, every administrative process gets an AI
assistant.
An AI assistant for prior authorization.
An AI assistant for coding.
An AI assistant for claims.
An AI assistant for denials.
An AI assistant for scheduling.
An AI assistant for documentation.
An AI assistant for inboxes.
An AI assistant for the AI assistants.
Congratulations.
We have automated the bureaucracy.
In the second future, AI operates further upstream.
It recognizes patterns.
It preserves context.
It anticipates missing information.
It identifies exceptions early.
It learns the practice.
It helps information travel correctly.
It prevents avoidable work.
And eventually, people barely notice it.
That is the future I find more compelling.
The Best Technology May Be the Work You No Longer Notice
We have become obsessed with visible technology.
The dashboard.
The chatbot.
The agent.
The notification.
The interface.
The AI button.
But the most valuable technology may eventually be
invisible.
The patient does not care that an intelligent system
prevented an authorization problem.
They care that their treatment wasn't delayed.
The physician does not care that a model reconciled
information across three systems.
They care that they didn't have to spend another 20 minutes
doing it.
The biller does not care that an algorithm detected an
upstream inconsistency.
They care that they didn't have to chase it three days
later.
The caregiver does not care about the architecture diagram.
They care that they have enough attention left for the
person in front of them.
That is the real product.
Healthcare Needs Fewer Heroes
This may sound strange coming from a story about someone
like Enoch Machora.
But I think healthcare needs fewer heroic workarounds.
We celebrate the nurse who stays at the bedside.
The physician who works late.
The biller who rescues the claim.
The administrator who knows the payer's secret rules.
The staff member who remembers how to fix the EHR problem
nobody documented.
These people are invaluable.
But a system should not depend on extraordinary human effort
to compensate for ordinary system failure.
Heroism should be available for emergencies—not required
for Tuesday afternoon.
That is an important distinction.
The Future of Healthcare May Be Boring
And that would be wonderful.
Imagine a future where:
The right information is available.
The authorization is already resolved.
The documentation follows the patient.
The charge is captured correctly.
The claim makes sense.
The payer receives what it needs.
The patient doesn't repeat the same information four times.
The physician doesn't become the translator between
incompatible systems.
The biller doesn't spend the afternoon investigating
something that should have been correct at the beginning.
Nobody writes a LinkedIn post about it.
Nothing went viral.
No one calls it revolutionary.
It just worked.
That may be the real definition of healthcare innovation.
Recent News Is Pointing in the Same Direction
The healthcare industry is simultaneously investing heavily
in AI and struggling with the administrative complexity that AI is supposed to
reduce.
The AMA's latest prior-authorization survey shows how
persistent the problem remains.
CMS is advancing electronic prior authorization initiatives
ahead of 2027 requirements, with healthcare organizations, EHR developers,
physician practices and digital-health companies participating in efforts to
address technical and workflow barriers.
Doximity's 2026 physician AI report shows that clinicians
increasingly see real potential for AI to reduce administrative burden and
create more time for patient care.
And this week's commentary from physicians such as Brian
Hudes is pushing the conversation toward a more fundamental question: does
technology actually eliminate work, or does it merely reorganize it?
That question is going to become increasingly important.
Because the next phase of healthcare AI will not be judged
merely by what it can generate.
It will be judged by what it can prevent.
Myth-Busting: Five Things We Need to Stop Saying
Myth #1: “AI will eliminate administrative burden.”
Not automatically.
AI can reduce burden.
It can also increase it if it is layered onto poor
workflows, unreliable data or unnecessary processes.
Myth #2: “More automation means better healthcare.”
Only if the automation improves the outcome that matters.
Automating useless work creates highly efficient useless
work.
Myth #3: “The claim is where the billing problem starts.”
Often it is where the problem becomes visible.
The root cause may be much earlier.
Myth #4: “Physicians don't want technology.”
Many physicians want technology desperately.
What they don't want is another system that requires them to
do more work in order to save work later.
Myth #5: “The future is autonomous healthcare.”
The better goal may be augmented humans with fewer
unnecessary tasks.
There is a difference between replacing judgment and
protecting it.
Tactical Advice for Practice Leaders
If you manage a medical practice, pick one recurring
administrative problem this month.
Just one.
Then ask your team:
Where does this problem actually begin?
Not where do we notice it.
Not who fixes it.
Not which department owns it.
Where does it begin?
Then ask:
What information was missing, wrong, delayed or trapped
when it began?
That question often produces more useful insight than
another software demonstration.
And before you buy anything, calculate the annual cost of
the problem.
If a task consumes 20 minutes.
Three times a day.
Five days a week.
Fifty weeks a year.
That's 250 hours.
One small recurring friction point can quietly become more
than six workweeks of labor.
And that is before considering physician time, patient
frustration, delayed care or lost revenue.
The math gets interesting very quickly.
What Should We Stop Doing?
This may be the most important exercise of all.
Every practice should periodically create a "work
that should not exist" list.
Not a wish list.
Not a software wishlist.
A work-elimination list.
Examples:
"Why are we entering this information twice?"
"Why does the physician have to answer this?"
"Why are we asking the patient for information we
already have?"
"Why does this authorization require a manual phone
call?"
"Why are we checking this status manually?"
"Why does the same denial keep happening?"
"Why are we fixing this downstream every week?"
"Why does someone have to remember this?"
"Why is this information not traveling with the
patient?"
These questions sound small.
They aren't.
Systems are often defined by the small things everyone has
stopped questioning.
A Better Definition of AI in Healthcare
Maybe AI should not be defined primarily by intelligence.
Maybe we should define it by friction removed.
A useful AI system should make the environment around the
clinician calmer.
More predictable.
More contextual.
More connected.
More transparent.
Less repetitive.
Less fragmented.
Less reactive.
The technology should understand enough of the environment
to know when to stay out of the way.
That is a surprisingly difficult capability.
It may also be one of the most valuable.
The Enoch Machora Lesson
Enoch Machora entered Providence as someone who needed care.
He left with a different relationship to caregiving.
He returned to the same institution not merely as an
employee, but as someone who had personally experienced the vulnerability his
patients experience.
His story reminds us that healthcare is not ultimately about
transactions.
Not claims.
Not codes.
Not authorizations.
Not workflows.
Not software.
Those things matter.
But they are means.
The end is the human being.
And the people caring for that human being need something
too.
They need time.
They need attention.
They need reliable information.
They need systems that support rather than interrupt them.
They need technology that knows when to help and when to get
out of the way.
Final Thoughts
Stop asking how much more work healthcare can automate.
Start asking how much unnecessary work healthcare can eliminate.
Build systems that return time, attention and judgment to
the people who actually care for patients.
The future of healthcare should not be a world where
machines do more and humans work harder; it should be a world where better
systems make human care easier to deliver.
Get Involved
Here is the question I would genuinely like to hear from
physicians, practice administrators, billers, nurses and healthcare operators:
What is one administrative task your practice performs
every week that you believe should not exist at all?
Not the task you wish were easier.
The task you believe should disappear.
Tell me in the comments.
If someone in your organization is fighting the same
unnecessary work, share or repost this article with them.
Because perhaps the next great healthcare innovation isn't
another tool.
Perhaps it is finally having the courage to say:
Why are we doing this at all?
Have your say.
Share your experience.
Challenge the assumptions.
Help shape the future of healthcare around the work that
actually matters.
About the Author
Dr. Daniel Cham is a physician, medical consultant
and entrepreneur working at the intersection of medicine, healthcare
technology, medical practice management and medical billing. His work focuses
on practical insights that help healthcare professionals navigate the increasingly
complex relationship between clinical care, technology, operations and the
business of medicine.
He is the founder of OnnX, an emerging healthcare
technology company focused on upstream intelligence for medical
practices—addressing the information and operational problems that often create
downstream administrative and billing friction.
Dr. Cham writes about the practical realities of modern
medical practice, healthcare operations, innovation, medical technology and the
future of medicine.
Connect with Dr.
Daniel Cham on LinkedIn
The views expressed in this article are for educational
and informational purposes and are not legal, medical, financial or regulatory
advice. Healthcare organizations should consult appropriately qualified
professionals when evaluating clinical, compliance, privacy, billing or
technology decisions.
Continue the Conversation
The conversation doesn't end with one article.
Explore more perspectives, interviews and practical ideas
across the platforms below:
Listen to the
podcast on Spotify
Subscribe and
watch on YouTube
Knowledge creates momentum. Keep exploring, keep
questioning and keep learning.
And if you want to go deeper, check the Featured
section of my LinkedIn profile for a free resource. No signup is required here.
Start there.
Learn something useful.
Then share it with someone who needs it.
Three References Worth Reading
1. Providence — “Stroke Patient Returns as Providence
Caregiver”
The original October 6, 2026 story of Enoch Machora's journey from caregiver to
stroke patient and ultimately back to caregiving at Sacred Heart Medical
Center.
Read
Enoch Machora’s story at Providence
2. American Medical Association — 2026 Prior
Authorization Physician Survey
The latest AMA data documenting the continuing clinical, operational and human
cost of prior authorization.
3. Brian Hudes, MD — “Administrative bloat in hospitals
is a design choice”
A timely physician perspective on whether technology eliminates work or simply
creates another layer of work around it.
Read
Brian Hudes, MD’s October 6 essay
#Healthcare #HealthcareAI #MedicalAI #HealthTech
#PhysicianLeadership #MedicalBilling #RevenueCycleManagement
#HealthcareInnovation #AdministrativeBurden #PriorAuthorization
#HealthcareTechnology #PhysicianBurnout #DigitalHealth #MedicalPractice
#AIinHealthcare #HealthcareOperations #OnnX
No comments:
Post a Comment