Daniel Cham MD

Why We Keep Fixing Healthcare Problems After They Happen

“It risks undercutting the benefits of primary care.” — Dr. Ishani Ganguli, Harvard Medical School

Sara Sandoval’s cancer changed everything. Healthcare’s administrative problems don’t have to. Why do we keep fixing what we could prevent?

Dr. Ishani Ganguli, an associate professor of medicine at Harvard Medical School and practicing primary care physician, made that observation this week while discussing the growing practice of charging patients for clinician email interactions. Her argument is broader than email: when healthcare turns every interaction into another transaction, it can undermine the relationships and preventive work that make healthcare valuable in the first place.

That tension is hiding in plain sight.

Healthcare is full of transactions.

Visits.

Codes.

Claims.

Authorizations.

Messages.

Denials.

Appeals.

Forms.

Portals.

Passwords.

Fax machines that somehow survived the internet.

And somewhere in the middle of all of it is a human being.

Sometimes that human being is a physician.

Sometimes a nurse.

Sometimes a practice manager.

Sometimes a billing specialist.

And sometimes it is a 35-year-old woman who thought she was getting ready to start her life.

Her name is Sara Sandoval.

Her story should make us think differently about what we mean when we say we are “improving healthcare.”

Because perhaps the healthcare problem we keep trying to solve is not the one we should have started with.

Sara Was Getting Ready to Start Her Life

In 2024, Sara Sandoval was 35 years old.

She was preparing for her October wedding to Ramiro.

She had begun proactively preparing to start a family.

She went to the doctor for extensive blood work because she wanted to understand where she stood with her health.

The results were normal.

Everything looked fine.

Then something didn't.

Sara had noticed a strange sensation in her upper-left armpit.

She assumed she had pulled a muscle.

She was exercising.

She was preparing for her wedding.

She was trying to lose weight.

There was nothing about the situation that screamed stage 4 cancer.

Then she noticed a small lump.

A mammogram followed.

Then an ultrasound.

Then a radiologist.

And then came the sentence nobody expects to hear.

Breast cancer.

About three weeks later, a full-body CT scan showed that the cancer had spread to her liver.

Sara had gone from no obvious warning signs to a stage 4 metastatic breast cancer diagnosis in roughly eight weeks.

The diagnosis came 40 days before her wedding.

It also happened to fall on Ramiro's birthday.

Her mother was with her when the doctor delivered the diagnosis.

Sara later described the experience in words that are difficult to forget:

“I was just getting ready to start my life.”

There is something important in that sentence.

Not because it is dramatic.

Because it is ordinary.

That is what makes it powerful.

She wasn't preparing for a medical crisis.

She was preparing for a wedding.

A marriage.

A family.

A future.

Cancer interrupted that plan.

But something else happened.

Her healthcare team recognized that Sara's life had not stopped being a life simply because treatment had started.

Her oncology team pushed one treatment back by a week so she could attend her wedding.

She shaved her head on the night of her rehearsal dinner after her hair became matted.

She wore a wig.

She got married.

She danced.

She celebrated.

She and Ramiro partied.

They made a memory in the middle of an experience that could easily have consumed everything else.

Today, Sara is receiving immunotherapy, hormone therapy and targeted therapy. PEOPLE reports that she currently has no signs of active cancer cells. She also works as a marketing executive in Chicago, runs Cancer Unfiltered, participates in support communities and advocates for people living with metastatic breast cancer.

She and Ramiro have also received the green light to begin the adoption process after a year with no evidence of active disease.

Sara has been candid about uncertainty.

She does not know exactly what the future holds.

But she is planning for one.

That distinction matters.

Healthcare Is Not a Workflow

Here is where I want to make a deliberately uncomfortable argument.

Healthcare has spent decades becoming better at managing workflows while sometimes forgetting that the workflow is not the point.

The workflow exists because a human being needs something.

A diagnosis.

A treatment.

A prescription.

A procedure.

A conversation.

An answer.

A second opinion.

A chance to live long enough to attend a wedding.

Yet much of healthcare technology is designed as though the ultimate goal is to move information from one box to another.

Patient enters.

Data captured.

Code assigned.

Claim generated.

Claim submitted.

Claim denied.

Denial worked.

Claim appealed.

Payment posted.

A/R followed.

Revenue collected.

Congratulations.

The machine has successfully processed another human experience.

Except the patient wasn't a claim.

And the physician wasn't a claims processor.

And the nurse wasn't born to become a prior-authorization specialist.

And the practice manager probably did not go into healthcare because she dreamed of spending Thursday afternoon resetting a payer portal password.

Yet here we are.

The Great Healthcare Paradox

We have more healthcare technology than any generation before us.

We also have more healthcare administrative complexity.

That should make us pause.

Because technology is supposed to reduce unnecessary work.

Sometimes it does.

Sometimes it merely moves the work around.

We digitized forms.

Then we created portals.

We created portals for the portals.

We created logins for the portals.

We created multifactor authentication for the logins.

We created software that integrates with the portals.

Then we created software to reconcile what the first software thought happened.

Then we hired someone to check whether the second software was correct.

At some point, we should probably ask a rude question:

What exactly did we automate?

This is not an argument against technology.

It is an argument against confusing digitization with simplification.

A paper form becoming a digital form is not necessarily progress.

A manual workflow becoming an automated manual workflow is not necessarily progress.

An AI system that helps a person process a bad workflow faster is not necessarily prevention.

Sometimes it is just a faster way to run in circles.

The AI Trap Nobody Likes Talking About

The healthcare AI conversation often begins with an assumption:

There is a problem. AI can automate it.

Fair enough.

But what if the problem exists because the underlying system creates bad inputs?

Then AI has a choice.

It can help repair the bad input.

Or it can help prevent the bad input.

Those are very different businesses.

Imagine a practice where information is captured inconsistently.

A patient's insurance information is incomplete.

A referral requirement is missed.

An authorization is initiated too late.

A clinical detail needed for a claim is buried somewhere in documentation.

A handoff between departments loses context.

A payer receives something different from what the practice believed it submitted.

The claim eventually gets denied.

Now everyone is very busy.

The billing team investigates.

Someone opens the chart.

Someone checks the payer portal.

Someone calls the payer.

Someone sends a message.

Someone requests documentation.

Someone corrects something.

Someone resubmits.

Someone waits.

Someone checks again.

Someone appeals.

Someone follows up.

Someone finally gets paid.

Then everyone congratulates themselves for recovering revenue.

But let's ask the uncomfortable question.

Why did the revenue need rescuing in the first place?

That is the question I think healthcare technology should ask more often.

The Problem May Not Be the Billing Department

This is one of the most important ideas behind OnnX:

Where you discover a problem is not necessarily where the problem began.

A denial is discovered in billing.

But the cause may have started much earlier.

At scheduling.

At registration.

During eligibility verification.

During clinical documentation.

During order entry.

During authorization.

During referral processing.

During a handoff.

During data transmission.

During an interface.

During a workflow decision nobody remembers making.

By the time the billing team sees the problem, the original mistake may be weeks old.

And now the organization is paying people to investigate history.

That is expensive.

Not just financially.

It consumes attention.

Attention Is Healthcare's Hidden Currency

We talk constantly about dollars.

We talk about labor.

We talk about productivity.

We talk about reimbursement.

We talk about utilization.

But one of the most valuable resources in healthcare is much harder to measure.

Human attention.

A physician's attention.

A nurse's attention.

A medical assistant's attention.

A practice manager's attention.

A biller's attention.

A patient's attention.

And attention is finite.

Every unnecessary administrative task takes a small piece of it.

One phone call.

One portal.

One clarification.

One missing field.

One authorization.

One denial.

One correction.

One follow-up.

Individually, each seems manageable.

Multiply them by thousands of encounters and suddenly the healthcare system has created an enormous attention tax.

And then we wonder why people are exhausted.

The AMA's 2025 physician survey found that physicians complete an average of 40 prior authorizations per week, consuming about 13 hours of physician and staff time each week. Ninety-five percent said prior authorization delays access to necessary care, while 94% said it contributes to burnout.

That is not a small workflow annoyance.

That is an operating model.

Forty Prior Authorizations a Week Is Not a Feature

Let's put the AMA number in human terms.

Forty prior authorizations.

Every week.

Thirteen hours.

Every week.

That's roughly an entire working day and a half disappearing into authorization activity.

And the remarkable thing is how quickly we normalize it.

“That's just healthcare.”

No.

That is a healthcare design decision that has become culturally invisible.

There is nothing natural about spending 13 hours a week pushing information through an administrative obstacle course.

We have simply become accustomed to it.

The AMA survey found that 26% of physicians reported prior authorization had led to a serious adverse event for a patient, including hospitalization, permanent impairment or death. Seventy-nine percent said patients abandon treatment because of authorization challenges, and 92% said prior authorization negatively affects clinical outcomes.

Those numbers should change the conversation.

The question is no longer merely:

“How do we process prior authorization faster?”

It should also be:

“Why does the system require so much friction before medically necessary care can move forward?”

That is a different question.

And different questions produce different products.

We Are Optimizing Around Noise

Here is my contrarian take:

Healthcare has become extremely sophisticated at optimizing around problems it has not eliminated.

We measure denials.

We measure turnaround time.

We measure A/R days.

We measure collections.

We measure touches.

We measure productivity.

We measure staff utilization.

We build dashboards.

Then we build dashboards for the dashboards.

And now we are adding AI.

AI can predict denials.

AI can summarize charts.

AI can draft appeals.

AI can identify missing documentation.

AI can route work.

AI can prioritize queues.

All useful.

But there is a dangerous possibility.

We could build a spectacularly intelligent machine for managing a fundamentally unnecessary mess.

That would be impressive.

It would also be a little ridiculous.

Imagine a city with 47 potholes.

Instead of fixing the road, we build an AI system that predicts which pothole your car will hit next.

Then another AI tells you how to steer around it.

Then another AI summarizes your pothole history.

Then a dashboard gives the city manager a pothole risk score.

At some point, someone should probably say:

“Can we just fix the road?”

Healthcare needs more conversations like that.

Automation Is Not Prevention

This distinction matters.

Automation says:

“How can we do this task faster?”

Prevention asks:

“Why does this task exist?”

Automation optimizes the workflow.

Prevention questions the workflow.

Automation can reduce labor.

Prevention can reduce the need for labor.

Both matter.

But they are not the same.

If a practice has 1,000 preventable billing exceptions each month, making the exception team 20% faster is useful.

Reducing the 1,000 exceptions to 200 is a different level of improvement.

One improves the repair shop.

The other reduces the number of cars that need repair.

Healthcare needs both.

But it has historically spent enormous energy building better repair shops.

Sara's Story Shows Why This Matters

None of this caused Sara's cancer.

It would be irresponsible to suggest otherwise.

Her story is not a billing story.

It is not an RCM case study.

It is not an argument that administrative technology would have changed her diagnosis.

It is something more fundamental.

It reminds us what healthcare is supposed to protect.

A person.

A life.

A relationship.

A future.

Sara's clinicians were dealing with an extraordinarily serious medical problem.

Yet they also recognized something simple:

The patient still had a wedding.

That sounds obvious.

It should be.

But healthcare systems can make obvious things strangely difficult.

The more fragmented the system becomes, the easier it is for everyone to focus on the transaction immediately in front of them.

The authorization.

The form.

The claim.

The code.

The queue.

The inbox.

The dashboard.

And the human being can become the thing that gets lost between them.

The Patient Is Not the Diagnosis

This may be the most important sentence in the entire article:

The patient is not the diagnosis.

The diagnosis is part of the patient's story.

The claim is part of the patient's story.

The authorization is part of the patient's story.

The medical record is part of the patient's story.

But none of those things is the patient.

Sara was not “stage IV metastatic breast cancer.”

She was Sara.

A wife.

A daughter.

A future mother.

A professional.

An advocate.

A person who wanted to dance at her wedding.

A person who still makes plans.

That distinction matters enormously when we design technology.

Because software tends to reduce reality into fields.

First name.

Last name.

DOB.

Insurance.

Diagnosis.

Procedure.

Provider.

Payer.

Status.

Claim number.

But life does not arrive in fields.

Life arrives messy.

Unexpected.

Emotional.

Complicated.

Human.

The purpose of healthcare technology should not be to make humans fit more neatly into databases.

It should be to make the databases work better for humans.

The Myth: “More AI Will Fix Healthcare”

No.

Not automatically.

AI is a capability.

It is not a philosophy.

AI can make a good system better.

It can make a bad system faster.

And it can make a bad system operate at extraordinary scale.

That is why the question should not be:

“Where can we put AI?”

The better question is:

“Where is the system generating unnecessary complexity, and could better information prevent it?”

Then ask whether AI is useful there.

That sequence matters.

Otherwise, AI becomes the answer searching for a problem.

Healthcare already has enough of those.

Three Questions Every Healthcare AI Buyer Should Ask

Before buying another AI platform, I would ask three questions.

1. Does this prevent work or merely accelerate work?

If the answer is “accelerate,” that can still be valuable.

But call it what it is.

Do not confuse faster repair with prevention.

2. What upstream condition creates the downstream problem?

If the vendor cannot explain where the problem begins, be careful.

A denial is an event.

A missing piece of information may be the cause.

Those are not the same thing.

3. What happens to human attention?

If the software saves money but creates more exceptions, more monitoring, more reconciliation and more oversight, the organization may simply be moving work from one person to another.

That is not necessarily transformation.

It may be redistribution.

The New RCM Question

Revenue cycle management traditionally asks:

Did we get paid?

Important question.

But I think the next generation of RCM needs to ask something earlier:

Why did getting paid become difficult?

That changes everything.

Because if payment is delayed because of a denial, we should investigate the denial.

But if the denial happened because the wrong information entered the process three weeks earlier, fixing the denial is not the deepest intervention.

The deeper intervention is upstream.

This is where OnnX's philosophy begins.

Not with:

“How do we automate billing?”

But with:

“How do we structure information early enough that billing has less to repair later?”

That is a very different proposition.

Healthcare Has a Data Problem Wearing a Workflow Costume

We often describe healthcare problems as workflow problems.

Sometimes they are.

But many workflow problems are symptoms of inconsistent information.

One person enters something.

Another person interprets it.

A third person copies it.

A fourth person corrects it.

A fifth person sends it somewhere else.

Then a sixth person asks:

“Wait. Which one is correct?”

Congratulations.

You have created a workflow.

And then you need software to manage the workflow.

This is how complexity compounds.

Bad structure creates variability.

Variability creates exceptions.

Exceptions create work.

Work creates queues.

Queues create delays.

Delays create follow-up.

Follow-up creates cost.

Cost creates pressure.

Pressure creates more automation.

And sometimes the automation simply manages the original variability.

That is the cycle.

Structure breaks the cycle.

The Human Cost of “Just One More Step”

Healthcare loves the phrase:

“It's just one more step.”

One more field.

One more verification.

One more authorization.

One more signature.

One more document.

One more portal.

One more phone call.

One more review.

One more confirmation.

One more checkbox.

One more reconciliation.

No individual step looks catastrophic.

But healthcare does not operate on individual steps.

It operates on millions of them.

The problem is not always the size of the step.

It is the number of times the step happens.

And the number of people it touches.

A two-minute task repeated 10,000 times is not a two-minute task.

It is 333 hours.

That's more than eight full workweeks.

Tiny friction becomes massive friction at scale.

The “Human-in-the-Loop” Problem

AI companies frequently say:

“Don't worry. There's a human in the loop.”

Good.

There should be.

But let's ask a better question:

Why is the human in the loop?

If the human is there because the AI needs oversight for legitimate judgment, excellent.

If the human is there because the system routinely produces ambiguous, incomplete or unreliable outputs, that's different.

Human oversight is not automatically a virtue.

Sometimes it is evidence that the system is doing exactly what it was designed to do.

Poorly.

The goal should not be:

Human in the loop.

The goal should be:

Human where human judgment adds value.

Those are very different architectures.

The Goal Is Not Fewer Humans

This deserves emphasis.

I am not arguing for eliminating healthcare workers.

Quite the opposite.

The goal is better human work.

If a physician spends less time fighting administrative friction, that is good.

If a nurse spends less time tracking paperwork, that is good.

If a practice manager spends less time reconciling contradictory information, that is good.

If a biller spends less time repairing preventable errors, that is good.

The human being should move up the value chain.

Away from repetitive repair.

Toward judgment.

Communication.

Relationships.

Problem-solving.

Clinical care.

Leadership.

Empathy.

The technology should absorb unnecessary friction.

The human should retain meaningful responsibility.

The Prevent–Detect–Repair Framework

One useful way to think about healthcare operations is through three layers.

PREVENT

Stop the problem before it enters the workflow.

Examples:

Better information capture
Eligibility verification
Structured documentation
Clear authorization requirements
Consistent patient data
Standardized handoffs
Early exception identification

DETECT

If prevention fails, identify the problem quickly.

Examples:

Missing information
Inconsistent data
Documentation gaps
Payer mismatches
Workflow exceptions
Authorization problems

REPAIR

If the problem survives both layers, fix it.

Examples:

Correcting claims
Appeals
Rework
Follow-up
Denial management
Payment reconciliation

Most organizations have invested heavily in Layer 3.

Many are now investing heavily in Layer 2.

The strategic opportunity is increasingly Layer 1.

Prevention.

Measure the Problem You Actually Have

If I were sitting with a physician-owned practice trying to identify its biggest operational problem, I would not begin with:

“What AI should we buy?”

I would begin with:

“Where does your staff repeatedly have to stop what they are doing and fix something?”

Then measure it.

Track:

Frequency

How often does it happen?

Touches

How many people interact with it?

Time

How much staff time does it consume?

Recurrence

Does the same issue keep happening?

Preventability

Could the issue have been prevented earlier?

Impact

Does it delay payment, care, scheduling or patient communication?

Exception rate

How often does the normal workflow fail?

Those measurements reveal something more useful than an AI demo.

They reveal where the system is leaking attention.

A Simple Exercise for Practice Owners

Pick your last 50 billing exceptions.

Do not start by looking at dollar value.

Instead, ask:

What happened before the exception?

Create five buckets.

Missing information.
Incorrect information.
Timing problem.
Documentation problem.
External requirement.

Then ask:

Which bucket appears most often?

Now ask:

Where did that information originate?

This is where the real investigation starts.

You may discover that your “billing problem” is actually a registration problem.

Or an authorization problem.

Or a documentation problem.

Or an interface problem.

Or a training problem.

Or a system-design problem wearing a technology costume.

That discovery is valuable.

The Dashboard Might Be Lying to You

Not literally.

But dashboards can create a dangerous illusion.

A practice sees:

Days in A/R: 38.

Denial rate: 7%.

Collection rate: 96%.

Clean claim rate: 92%.

Everything looks measurable.

Therefore everything looks manageable.

But the dashboard may not tell you how many times a nurse interrupted patient care to answer an administrative question.

It may not tell you how many times a physician had to respond to an authorization issue.

It may not tell you how many messages were exchanged because two systems disagreed.

It may not tell you how much cognitive energy was spent reconstructing information that should have been available the first time.

Not every important metric is financial.

Some of the most important metrics are attention metrics.

How many interruptions?

How many handoffs?

How many rework events?

How many exceptions?

How many human touches?

How much preventable work?

Healthcare should measure those too.

The Hidden ROI of Giving Attention Back

Suppose a technology saves a practice $100,000.

Excellent.

Now suppose another system saves 2,000 hours of administrative work.

Which is more valuable?

There is no universal answer.

But the second number creates a question that the first may not:

What could the organization do with those 2,000 hours?

More patient access?

Better follow-up?

Faster communication?

Less burnout?

Better documentation?

More time for physicians to think?

More time for nurses to educate?

More time for practice leaders to improve operations?

The economic value of technology is not only what it removes.

It is what it allows humans to do instead.

That is why “time saved” is not enough.

We should ask:

Whose time?

And:

What becomes possible with it?

A Little Humor, Because Healthcare Needs It

Healthcare technology meetings sometimes sound like this:

“We have an interoperability challenge.”

Translation:

“Our systems don't talk to each other.”

“We have a workflow optimization opportunity.”

Translation:

“Everyone is doing the same thing six different ways.”

“We need a human-in-the-loop model.”

Translation:

“The robot still needs supervision.”

“We are leveraging AI-enabled automation.”

Translation:

“We taught software to move the paperwork faster.”

And my favorite:

“We're creating a seamless experience.”

That is usually when I start looking for the seams.

Because the patient certainly knows where they are.

What Sara's Story Has to Do With Billing

At first glance, nothing.

That is precisely why it matters.

Sara's story is about cancer.

But the larger lesson is about what healthcare should protect.

Her oncology team did not merely see a treatment schedule.

They saw a person with a wedding.

They did not erase the medical reality.

They made room for the human reality.

That is the standard technology should support.

Technology should not make healthcare less human because humans are inconvenient.

It should make healthcare more human by removing the unnecessary work that keeps people from being human.

That includes billing.

Especially billing.

Because billing sits at the end of an extraordinarily long chain of information.

If the information entering that chain is poor, someone downstream eventually pays for it.

Usually in time.

Sometimes in money.

Sometimes in frustration.

Sometimes in delayed care.

The Future of Medical Billing May Begin Before Billing

This is where I believe the RCM conversation needs to move.

We have spent years asking:

How do we code better?

How do we submit cleaner claims?

How do we manage denials?

How do we improve collections?

How do we accelerate A/R?

Important questions.

But the next question should be:

How do we make the downstream billing process less necessary in the first place?

That means moving upstream.

Capturing structured information earlier.

Reducing ambiguity.

Identifying exceptions before they become claims problems.

Connecting clinical and operational information.

Making data usable before it becomes a billing artifact.

That is not simply a better billing workflow.

It is a different way of thinking about the revenue cycle.

The Revenue Cycle Is Not a Cycle

Here is another contrarian idea.

Maybe the phrase “revenue cycle” hides something important.

A cycle implies repetition.

Something happens.

It comes around again.

And again.

But a healthy system should not repeatedly generate the same avoidable problem.

If a practice repeatedly sees the same denial, the answer should not always be:

“Get better at working denials.”

Sometimes the answer should be:

“Stop creating the denial.”

That is the difference between managing a cycle and redesigning a system.

The Biggest Opportunity May Be Boring

Healthcare founders often want to build something exciting.

Generative AI.

Ambient intelligence.

Autonomous agents.

Predictive analytics.

Digital twins.

Copilots.

All fascinating.

But some of the most valuable healthcare innovation may be remarkably boring.

Correct information.

At the right time.

In the right structure.

Available to the right person.

Without another phone call.

Without another portal.

Without another spreadsheet.

Without another human having to remember what the previous human already knew.

That does not make for the flashiest demo.

But it may make for a better healthcare system.

Three Expert Perspectives Worth Paying Attention To

1. Dr. Ishani Ganguli: Don't Turn Every Interaction Into Another Transaction

Dr. Ganguli's October 1, 2026 commentary focuses on charging patients for clinician email interactions. Her concern is not that physician time lacks value. It is that payment models can unintentionally undermine the relational and preventive work of primary care.

That principle extends beyond email.

If every piece of healthcare is treated as an isolated transaction, the system can lose sight of the relationship connecting those transactions.

Sara's story illustrates the opposite.

Her care team understood that treatment existed inside a life.

2. Dr. Shanthi Sivendran: Metastatic Cancer Is a Disease Patients Live With

Dr. Shanthi Sivendran, interim chief patient officer at the American Cancer Society, explained that metastatic breast cancer occurs when breast cancer spreads beyond the breast, potentially involving areas such as the liver, lungs or bones. She noted that treatment can involve endocrine therapy, chemotherapy, targeted therapy and immunotherapy.

That clinical reality matters because the word “cancer” can collapse an entire person into a diagnosis.

Sara's advocacy pushes against that reduction.

3. The AMA Physician Survey: Administrative Friction Has a Measurable Cost

The AMA's 2025 survey of 1,000 practicing physicians found that 95% reported prior authorization delays necessary care, 79% reported patients abandoning treatment because of authorization challenges, and 94% said prior authorization contributes to burnout. Physicians reported an average of 40 prior authorizations per week and approximately 13 hours of physician and staff time devoted to them.

The lesson is not that every administrative process is unnecessary.

It is that administrative friction has become large enough to measure.

And what can be measured can finally be questioned.

Three Myths We Should Stop Repeating

Myth 1: “The problem is that staff aren't working efficiently enough.”

Sometimes staff performance matters.

But if excellent people repeatedly encounter poorly structured information, the organization may have a system problem rather than a people problem.

Myth 2: “AI will eliminate the administrative burden.”

AI can reduce certain burdens.

But if the underlying data is inconsistent, AI may simply automate interpretation, correction and exception handling.

Myth 3: “Billing owns the revenue problem.”

Billing owns the downstream process.

It does not necessarily own the upstream causes.

A registration error can become a billing problem.

A documentation gap can become a billing problem.

An authorization problem can become a billing problem.

A data mismatch can become a billing problem.

The department where the problem appears is not always the department where the problem should be solved.

The OnnX Principle

This is the philosophy behind OnnX:

Most of the problem starts upstream.

Healthcare has spent enormous energy optimizing downstream workflows.

OnnX approaches the problem differently.

Instead of asking only:

“How can we process this claim?”

Ask:

“What information should have been structured earlier?”

Instead of:

“How do we work this denial?”

Ask:

“What created the conditions for this denial?”

Instead of:

“How do we automate billing?”

Ask:

“How do we reduce the amount of billing repair required in the first place?”

That shift sounds subtle.

It is not.

It moves the center of gravity from repair to prevention.

We Don't Need More Noise-Management

Healthcare does not have a shortage of noise-management tools.

It has EHRs.

PM systems.

Billing systems.

Clearinghouses.

Payer portals.

Authorization platforms.

Analytics.

Dashboards.

Work queues.

Messaging systems.

AI assistants.

And now AI assistants for the AI assistants.

What we need is less noise.

Structure beats noise.

When information is structured properly, fewer people have to interpret it.

When fewer people have to interpret it, fewer exceptions are created.

When fewer exceptions are created, fewer downstream tasks exist.

When fewer downstream tasks exist, humans get time back.

That is the chain.

The Physician Attention Test

Here is a simple test for any healthcare technology.

Ask:

Does this give a physician, nurse or practice employee meaningful attention back?

Not theoretical time.

Actual attention.

If a physician used to spend 30 minutes fixing something and now spends 15 minutes supervising software that fixes it, there is improvement.

But there may be a better opportunity.

What if the problem disappeared?

What if the physician spent zero minutes on it?

That is the north star.

Not faster friction.

Less friction.

What Healthcare Founders Should Be Building

If you are building healthcare technology, here is my challenge:

Don't begin with the technology.

Begin with the recurring human frustration.

Find the task people perform repeatedly.

Follow it backward.

Ask what created it.

Keep following the chain until you reach the earliest point where the problem could have been prevented.

Then ask:

Can technology change that point?

Maybe the answer is AI.

Maybe it is better data architecture.

Maybe it is workflow redesign.

Maybe it is an integration.

Maybe it is a better interface.

Maybe it is nothing technological at all.

That last possibility is important.

Sometimes the best healthcare innovation is a process that no longer needs software.

The Future May Belong to Systems That Know When Not to Intervene

There is another opportunity hiding here.

Good technology does not merely automate.

It knows when to stay out of the way.

If everything becomes an alert, nothing is an alert.

If everything becomes an exception, nothing is exceptional.

If everything requires human review, humans become the queue.

The best system should resolve routine situations quietly and surface only the things that genuinely require judgment.

That means:

Less noise.

Fewer interruptions.

Better exceptions.

Clearer information.

More meaningful human involvement.

That is intelligent automation.

Not simply more automation.

What “Better” Should Mean

Healthcare technology often defines improvement in terms of:

More claims processed.

More tasks completed.

Daniel Cham MD

Friday, October 2, 2026

Sara Sandoval Was Planning Her Wedding. Then Came Stage 4 Cancer.

Sara Sandoval Was Planning Her Wedding. Then Came Stage 4 Cancer.

Report Abuse