Why the Next Breakthrough in Medical Billing May Have Nothing to Do With Billing
“For too long, fee-for-service payment has rewarded
volume over value.” — Dr.
Mehmet Oz, CMS Administrator
The Healthcare Industry Has Perfected an Expensive Art:
Fixing Problems It Could Have Prevented
Imagine running a restaurant where the chef prepares an
excellent meal.
The customer enjoys it.
The waiter delivers the bill.
Then the restaurant spends three weeks proving that the
customer actually ordered the food.
Welcome to healthcare.
We call it revenue cycle management.
In almost any other industry, we might call it a design
problem.
Here is the uncomfortable question.
What if the medical billing industry has become
exceptionally good at solving problems that should never have existed?
We hire billing specialists to correct claims.
We hire coding experts to interpret documentation.
We hire denial management teams to challenge payers.
We purchase software to track all the work.
Then we purchase AI to make the work faster.
Everyone is busy.
Everyone has a dashboard.
Everyone can explain why the claim was denied.
Meanwhile, the physician is still waiting to get paid.
Something about this deserves a second opinion.
I am a physician and the founder of OnnX, an AI-powered
medical billing initiative focused on independent clinics.
The more I examine revenue cycle management, the more I
question one of its fundamental assumptions.
We treat billing failures as downstream administrative
problems.
But many begin much earlier.
At registration.
During eligibility verification.
When authorization requirements are overlooked.
When clinical documentation is transformed into structured
data.
When information moves between disconnected systems.
The denial is often the symptom. The disease is upstream.
And healthcare has developed an impressive industry around
treating the symptom.
That is not an argument against billing professionals.
Their work is essential.
It is an argument against designing systems that repeatedly
require them to repair preventable mistakes.
A good billing department fixes errors. A better system
prevents the avoidable ones.
The distinction may determine the future of independent
medical practice.
Seven Uncomfortable Truths Every Clinic Owner Should
Consider
Before diving into the details, consider these seven ideas.
- Your
billing department may be doing excellent work inside a poorly designed
system. More effort cannot always compensate for broken information
flow.
- A
paid claim is not necessarily a successful process. If it took five
corrections, three phone calls, and two appeals, payment may have come at
an unnecessarily high administrative cost.
- Your
EHR may be excellent at storing information and terrible at moving the
right information to the right place. Storage is not understanding.
- AI
cannot repair clinical meaning it never received. A confident answer
built on incomplete information remains unreliable.
- A
high denial recovery rate can coexist with an inefficient revenue cycle.
Recovery is valuable. Prevention may be more valuable.
- The
most useful billing dashboard may be the one that explains why problems
happen. Knowing that money is missing is less useful than knowing what
to change.
- The
future of medical billing may belong to companies that make billing less
complicated, not companies that make billing software more complicated.
That last point is worth remembering.
The ultimate purpose of automation is not to create a
more impressive workflow. It is to make unnecessary work disappear.
The Patient Was Treated Successfully. Then the
Administrative Adventure Began.
Consider a fictional independent clinic.
The physician sees a patient named Michael.
Michael is 62 years old and has diabetes, hypertension, and
worsening fatigue.
The physician reviews his medications, evaluates his
symptoms, documents the findings, and develops an appropriate care plan.
The encounter takes 25 minutes.
The administrative consequences may last 25 days.
Here is how.
Monday: Everything Looks Fine
Michael checks in.
His insurance card is scanned.
The front desk confirms his demographic information.
The physician sees him.
The encounter is documented.
The charge is entered.
The claim is submitted.
The clinic moves on.
Nobody notices that Michael recently changed insurance
plans.
His new coverage has different administrative requirements.
The system has information.
Unfortunately, it is yesterday's information.
Outdated data is not harmless simply because it is neatly
organized.
Wednesday: The First Problem Appears
The claim fails an eligibility-related requirement.
The billing specialist begins investigating.
The front desk is contacted.
Michael may need to provide updated information.
The billing system displays an error.
But the error did not begin in the billing system.
It began before the encounter.
The billing system is merely delivering the bad news.
Like a smoke alarm, it is useful.
But nobody would describe a smoke alarm as a fire prevention
strategy.
Friday: The Second Problem Appears
While reviewing the claim, the billing specialist notices a
discrepancy between the documented assessment and a diagnosis selected from the
patient's historical problem list.
The physician receives a clarification request.
The physician must reopen the record.
The billing team waits.
The practice manager follows up.
A problem that might have been identified during charge
capture has become another administrative task.
Nobody intended to create the work.
Yet everyone is now participating in it.
Two Weeks Later: The Claim Is Corrected
The information is updated.
The documentation discrepancy is resolved.
The claim is resubmitted.
Eventually, payment arrives.
The billing department records a successful resolution.
And technically, it is.
But what exactly was accomplished?
The practice delivered one clinical service.
Then several employees worked to correct information
associated with that service.
The clinic did not necessarily have a billing problem. It
had an information problem that billing exposed.
This fictional example is a composite of common operational
issues, not a documented OnnX customer case.
Its purpose is to reveal a larger structural weakness.
The claim was not the beginning of the revenue cycle.
It was the final messenger carrying information assembled by
many people and systems.
And as physicians know, treating the messenger rarely cures
the underlying condition.
The Most Expensive Word in Medical Billing May Be
“Rework”
Healthcare loves productivity metrics.
Claims processed per employee.
Days in accounts receivable.
Denial recovery rates.
Average reimbursement.
Collections per physician.
These numbers matter.
But one of the most revealing measurements receives less
attention.
How much work does your clinic perform more than once?
Consider a simple hypothetical example.
A clinic submits 1,000 claims monthly.
One hundred require preventable correction or additional
administrative work.
That is a 10% rework rate.
Assume each affected claim consumes 20 minutes of staff
time.
That produces approximately 33 hours of rework every
month.
At an illustrative fully loaded labor cost of $35 per hour,
the direct expense is approximately $1,167 monthly, or $14,000
annually.
That excludes physician interruptions, payment delays,
vendor costs, and potentially unrecovered revenue.
Now imagine reducing the preventable rework by half.
The clinic could save approximately 17 staff hours monthly.
Not because employees worked harder.
Not because the clinic hired more people.
Not because the physician saw more patients.
Simply because fewer mistakes needed correction.
These are illustrative calculations, not verified savings
from a particular practice.
But the principle is powerful.
The cheapest administrative task is often the one you
never need to perform.
That raises a provocative question.
Why do so many revenue cycle conversations begin with
collections instead of rework?
Perhaps because collections are easy to count.
Prevented mistakes are harder to see.
A recovered $5,000 claim appears on a report.
An eligibility error prevented before the encounter rarely
receives applause.
There is no dramatic recovery story.
No heroic phone call.
No appeal letter.
Just a claim that processes correctly.
How boring.
How valuable.
The best operational improvement often looks like nothing
happened.
Healthcare AI Has a Garbage-In Problem. But the Real
Problem Is More Interesting.
The familiar technology expression is garbage in, garbage
out.
It is useful.
But healthcare deserves a more precise version.
Incomplete context in. Confident mistakes out.
Healthcare information is rarely simple.
A physician's clinical note may contain symptoms,
observations, uncertainty, medical reasoning, patient preferences, and
contextual details.
A billing claim contains a much narrower representation of
the encounter.
Between those two forms of information, something important
happens.
Clinical meaning is translated into structured data.
Diagnoses become codes.
Services become billing entries.
Clinical documentation becomes evidence supporting the
reported services.
Patient information becomes administrative fields.
This transformation is necessary.
But it is not always perfect.
The Translation Problem
Imagine translating a novel into a restaurant menu.
Both contain words.
Both can be structured.
But they serve different purposes.
A menu cannot preserve every detail of a novel.
Similarly, a claim cannot contain every detail of a clinical
encounter.
That is not inherently a flaw.
The problem arises when information essential to an
administrative or clinical decision is lost, misinterpreted, or disconnected
during transformation.
A physician may document a clinically important finding in
free text.
A downstream system may not recognize its relevance.
A diagnosis may be selected from an outdated problem list.
A required authorization may exist but remain inaccessible
to the billing workflow.
A claim may contain technically valid codes without adequate
supporting documentation.
Data can be structured, standardized, and still wrong.
This is where the conversation about healthcare AI needs
more sophistication.
The question is not simply whether an AI model can read
clinical documentation.
The question is whether it can preserve the meaning required
for a particular decision.
And whether it can recognize when that meaning is missing.
The Role of FHIR
Fast Healthcare Interoperability Resources, or FHIR,
provides a standardized framework for exchanging healthcare information.
It can support more consistent access to clinical data.
But interoperability is not the same as clinical
understanding.
A successful transfer does not prove that the information is
complete.
A valid code does not prove that it is appropriate.
A structured field does not prove that the original clinical
meaning survived.
And an AI model receiving the data does not automatically
know what was excluded.
Moving bad information faster is not interoperability
success. It is efficient confusion.
That is why data validation should occur throughout the
information lifecycle.
Not only after the claim fails.
We Have Been Asking AI the Wrong Question
Most discussions about AI-powered medical billing begin with
capabilities.
Can AI code encounters?
Can it identify denials?
Can it generate appeals?
Can it predict payment delays?
Can it automate follow-up?
These are reasonable questions.
But I would begin somewhere else.
What administrative work should not exist in the first
place?
This reframes the problem.
Instead of asking how AI can accelerate denial management,
ask how AI can help identify preventable errors before submission.
Instead of asking how AI can summarize unpaid claims, ask
why those claims became unpaid.
Instead of asking how AI can increase billing productivity,
ask whether the workflow contains unnecessary tasks.
This does not mean downstream automation lacks value.
It does.
Payer disputes are real.
Coverage policies are complex.
Some denials are inappropriate.
Appeals remain necessary.
But a system focused entirely on downstream recovery risks
becoming very good at managing recurring failure.
Consider the difference.
Reactive AI: This claim was denied because the
authorization number was missing.
Preventive AI: This service may require
authorization, and the required information has not been confirmed.
The first identifies a problem.
The second may help prevent it.
Both are useful.
But they operate at different points in the process.
One repairs.
The other protects.
The most intelligent billing system may be the one that
knows when not to submit a claim.
That does not mean delaying valid claims unnecessarily.
It means identifying material, resolvable problems before
they become avoidable rejections or denials.
Three Medical Experts, Three Lessons the Billing Industry
Should Hear
The future of medical billing is not only a technology
question.
It is a systems-design question.
A clinical workflow question.
A policy question.
And ultimately, a human question.
Three prominent physicians offer useful perspectives.
Their ideas are interpreted here through their established
work and public policy positions. These are not direct interviews or
endorsements of OnnX.
Dr. Donald Berwick: Stop Blaming People for Problems
Created by Systems
Dr. Donald Berwick, a pediatrician and influential
healthcare quality leader, has spent much of his career advancing systems-based
improvement.
His work challenges a common management reflex.
When something goes wrong, find the person responsible.
In medical billing, that often sounds like this.
Who entered the wrong insurance number?
Who missed the authorization?
Who selected the incorrect code?
Who forgot to follow up?
These questions may identify the immediate action.
But they often miss the conditions that made the error
likely.
Perhaps the front desk was switching between three
applications.
Perhaps the authorization requirements were not visible
during scheduling.
Perhaps the clinical note and charge entry systems did not
share information effectively.
Perhaps nobody owned the final validation step.
The better question is:
Why did the system allow this mistake to happen?
This does not eliminate individual accountability.
It places accountability within a better-designed process.
A clinic should make correct actions easier and predictable
mistakes harder.
When the same mistake happens repeatedly, stop calling it
a mistake. Start investigating the system.
That is the first expert lesson.
Dr. Robert Wachter: A Brilliant Algorithm Can Still Be a
Terrible Coworker
Dr. Robert Wachter, an internist and leading voice in
digital healthcare, has examined the complicated relationship between
technology and clinical practice.
His work offers an important warning.
Technology does not operate in a vacuum.
It operates around real people.
People who are interrupted.
People who are busy.
People who have incomplete information.
People who must make decisions under pressure.
An AI system may demonstrate impressive coding accuracy
during testing.
But what happens inside a clinic?
Does it generate excessive alerts?
Does it require physicians to review trivial discrepancies?
Does it interrupt documentation?
Does it explain why a recommendation was made?
Does it recognize exceptions?
Does it reduce work?
Or does it simply create a more technologically advanced way
to annoy everyone?
A brilliant algorithm can still be a terrible coworker.
The true test of healthcare AI is not how impressive it
looks in a demonstration. It is how much unnecessary work disappears on Monday
morning.
That is the second expert lesson.
Dr. Mehmet Oz: Payment Policy Is Changing. Administrative
Systems Must Keep Up.
Dr. Mehmet Oz, CMS Administrator, has emphasized changes in
Medicare payment policy intended to better align reimbursement with value,
prevention, and accountability.
The November 4, 2026 Medicare Physician Fee Schedule
announcement illustrates the connection between payment policy and practice
operations.
Policy changes affect more than reimbursement calculations.
They can influence documentation, coding, billing, staffing,
and workflow design.
A large healthcare organization may assign teams to
interpret and implement new rules.
An independent clinic may rely on one practice manager.
Perhaps that manager is also handling payroll, scheduling,
vendor contracts, and an employee who called in sick.
This is the reality of small-practice management.
A new payment rule may look like a paragraph in a federal
document.
Inside a clinic, it can become several operational changes.
A regulation does not implement itself. Someone in the
clinic has to translate it into work.
The third expert lesson is that technology should help
practices adapt to changing requirements without creating another layer of
complexity.
This Week's News: Medicare Changes the Rules. Again.
On November 4, 2026, CMS announced the final 2027
Medicare Physician Fee Schedule.
The announcement included payment adjustments, coding
changes, and policies affecting several aspects of physician practice.
For independent clinics, three developments deserve
attention.
First: Payment Pressure Is Not Disappearing
CMS reported 2027 conversion factors of approximately $32.95
for qualifying alternative payment model participants and $32.63 for
nonqualifying participants.
Those represent projected decreases of 1.83% and 2.31%,
respectively, from the corresponding 2026 conversion factors.
These percentages are not estimates of every physician's
overall revenue change. Actual effects vary by specialty, service mix,
geographic adjustments, and other policies.
Still, the operational implication is obvious.
When reimbursement is under pressure, preventable
administrative waste becomes harder to ignore.
A clinic cannot control every payment rule. It can
control more of the information it submits.
Second: Coding Rules Continue to Evolve
CMS finalized changes involving separately identifiable
office and outpatient evaluation and management services performed on the same
day as certain procedures.
The final policy applies to specified circumstances
involving 0-day global procedures, with the lower-valued separately
payable service paid at 80% rather than the initially proposed 50%.
The agency also finalized changes transitioning the G2211
office and outpatient visit complexity add-on to a modifier-based approach.
These details matter because a billing system configured for
yesterday's rules can become tomorrow's source of errors.
Third: Technology-Enabled Care Still Requires
Administrative Discipline
CMS also finalized remote monitoring policy changes,
including delaying a restriction involving contracted clinical staff until
2029.
The broader lesson extends beyond remote monitoring.
A clinical service may be appropriate.
A technology may function properly.
A physician may deliver excellent care.
But reimbursement still depends on meeting applicable
administrative and documentation requirements.
Clinical innovation moves quickly. Payment rules move
differently. Independent clinics live in the gap.
That gap is where better information management can create
practical value.
The Statistics Physicians Should Actually Care About
Healthcare loves big numbers.
Billions lost.
Millions of claims.
Thousands of administrative hours.
These statistics attract attention.
But national estimates rarely tell a clinic owner what to
fix on Tuesday.
A practice needs its own measurements.
Here are the numbers I would prioritize.
First-Pass Claim Acceptance
How many claims are accepted for processing on the first
submission?
A clinic might establish 95% or higher as an internal
target, depending on its baseline and definition.
Acceptance is not payment.
But repeated submission failures reveal unnecessary work.
Initial Denial Rate
How often do payers initially deny claims?
Measure consistently by claim count, claim line, or dollars.
Separate preventable administrative failures from disputed
payer decisions.
A denial rate without a root-cause breakdown is a symptom
report, not a diagnosis.
Accounts Receivable Aging
How much collectible revenue remains outstanding beyond 30,
60, 90, and 120 days?
An internal goal of fewer than 45 days in A/R may be
useful for some practices, but specialty and payer mix matter.
The trend is often more informative than a single target.
Staff Rework Hours
How much time is spent correcting information?
How often are physicians interrupted for documentation
clarification?
How many claims require multiple submissions?
These questions expose the hidden labor cost of poor
information quality.
Preventable Denials
Which denials could reasonably have been avoided through
earlier validation?
This is where a clinic begins moving from recovery toward
prevention.
Cost to Collect
How much does the practice spend collecting revenue?
Include relevant staffing, software, vendor, and
administrative costs.
A low cost to collect is not inherently good if collections
or compliance suffer.
The goal is accurate, timely, compliant reimbursement with
reasonable administrative effort.
The best metric is not the one that makes the dashboard
look good. It is the one that tells you what to fix.
The Great Billing Paradox: Success Can Hide Failure
Imagine two clinics.
Clinic A has an excellent denial recovery team.
It successfully overturns many denials.
The team works hard.
The practice eventually collects much of the money.
Clinic B has fewer preventable denials.
Its staff spends less time correcting claims.
It may need fewer appeals.
Which clinic has the better revenue cycle?
The answer depends on the complete financial and operational
results.
But this comparison exposes a blind spot.
A high recovery rate can be evidence of strong
performance after a failure, not evidence that the overall system is well
designed.
Think about hospital quality.
We would not judge surgical safety solely by how effectively
a hospital treats postoperative complications.
We would also ask how many complications could have been
prevented.
Why should billing be different?
Recovery matters.
Prevention matters too.
The industry needs both.
But the emphasis should shift toward eliminating avoidable
work wherever possible.
A system should not receive all the credit for solving a
problem it repeatedly allows to occur.
The Ten Pitfalls That Make AI Billing Projects Fail
Technology can improve medical billing.
It can also create expensive disappointment.
Here are ten pitfalls worth avoiding.
1. Automating before understanding the workflow. If
the process is broken, automation may simply accelerate failure.
2. Treating AI output as authoritative. A confident
coding suggestion is not proof that the code is supported.
3. Ignoring clinical context. Structured data may
omit important information.
4. Confusing acceptance with payment. A claim can
pass initial validation and still be denied.
5. Measuring productivity without measuring rework. More
processed claims do not automatically mean better performance.
6. Applying generic payer rules. Coverage and billing
requirements vary across payers, contracts, and services.
7. Overloading physicians with alerts. A tool that
interrupts clinicians constantly may create more burden than it removes.
8. Ignoring privacy and security. Billing automation
frequently involves protected health information.
9. Buying features instead of outcomes. A long
feature list is not evidence of value.
10. Forgetting that some denials are payer failures.
Practices need effective appeal and escalation processes, not merely internal
correction tools.
The common theme is simple.
Technology cannot substitute for understanding the
problem.
A Practical 90-Day Plan: Fix One Problem Before Buying
Another Platform
Independent practices rarely have the resources for endless
transformation projects.
Fortunately, meaningful improvement can begin with a focused
process.
Days 1–15: Find the Friction
Review recent claims.
Identify recurring denials, rejections, payment delays, and
staff rework.
Interview the people closest to the process.
Ask the front desk what repeatedly goes wrong.
Ask billers which problems consume the most time.
Ask physicians what administrative questions interrupt their
work.
Do not assume management already knows the answers.
The people doing the work often understand the friction
better than the people reviewing the reports.
Days 16–30: Follow the Information
Choose one recurring problem.
Trace it backward.
Where did the information originate?
Who entered it?
Who verified it?
Which system received it?
Where did it become incomplete or incorrect?
Was the problem detectable earlier?
This is the most important step.
Do not begin by asking who made the mistake. Begin by
asking where the information stopped being trustworthy.
Days 31–45: Redesign the Checkpoint
Create a simple prevention mechanism.
For example, verify eligibility before the encounter.
Clarify authorization responsibilities.
Standardize charge capture.
Establish a documentation query process.
Improve communication between clinical and billing staff.
The intervention should be specific enough to measure.
Days 46–60: Introduce Appropriate Automation
Determine whether technology can make the improved process
more reliable.
Sometimes a simple rule is sufficient.
Sometimes integration is necessary.
Sometimes AI can help interpret unstructured information.
Choose the least complicated solution that addresses the
problem effectively.
Do not use a neural network to solve a problem a checkbox
can prevent.
Days 61–75: Measure the Results
Compare performance against the baseline.
Did the error rate fall?
Did rework hours decrease?
Did first-pass acceptance improve?
Were there unintended consequences?
Did staff actually use the new process?
Days 76–90: Decide What to Keep
Retain what works.
Modify what does not.
Document the results.
Then identify the next priority.
A successful improvement program does not need to fix every
problem immediately.
It needs to make the next improvement easier.
Small improvements compound. Repeated mistakes compound
too. Choose which one your clinic wants to accumulate.
What Physicians Should Demand From AI Billing Vendors
If you are evaluating an AI-powered medical billing
platform, ask questions that go beyond the demonstration.
Do not begin with:
How many features does it have?
Begin with:
Which specific problem does it prevent?
Then ask:
- Can
the system identify missing information before submission?
- Does
it explain the reason for its recommendations?
- Can
staff review and override appropriate suggestions?
- How
does it handle uncertain or conflicting information?
- Does
it integrate with existing EHR and practice management workflows?
- How
are payer and coding updates maintained?
- What
security and privacy safeguards are provided?
- Can it
demonstrate measurable reductions in rework?
- What
happens when the system makes a mistake?
- Who
remains accountable for the final claim?
These questions separate impressive technology from
practical technology.
A vendor may show you a beautiful dashboard.
Ask what the dashboard changes.
A vendor may demonstrate an intelligent coding assistant.
Ask how unsupported recommendations are prevented.
A vendor may promise greater efficiency.
Ask how efficiency will be measured.
A demonstration shows what software can do. Evidence
shows what it actually improves.
Legal Implications: AI Does Not Come With a Liability
Eraser
Medical billing is regulated.
AI does not change that.
Practices must still comply with applicable documentation,
coding, reimbursement, privacy, and contractual requirements.
Several issues deserve attention.
Billing Integrity
Claims must accurately reflect services provided and meet
applicable requirements.
AI should not recommend unsupported diagnoses, unnecessary
services, or inappropriate billing strategies.
Improper claims can create serious compliance consequences,
including potential exposure under federal and state laws.
HIPAA Compliance
AI vendors handling protected health information may require
appropriate business associate agreements and security safeguards.
Practices should understand how data is accessed, stored,
processed, and protected.
Human Oversight
Automation should not eliminate required professional
review.
Clinical judgment, medical necessity, documentation
accuracy, and complex coding decisions may require qualified personnel.
Auditability
A practice should be able to reconstruct important
decisions.
What information was used?
What recommendation was generated?
Who reviewed it?
What changed?
Why?
An AI system that cannot explain a consequential
recommendation may create a new compliance problem while trying to solve an old
billing problem.
Ethical Considerations: The Patient Is Not a Billing Code
It is easy to discuss revenue cycle management entirely in
financial terms.
Claims.
Collections.
Margins.
Denials.
Productivity.
But every claim begins with a patient.
And every patient encounter contains more meaning than a
reimbursement transaction.
This creates ethical responsibilities.
Clinical accuracy must come before financial
optimization.
AI should not encourage documentation designed to justify
services that were not provided.
It should not recommend diagnoses merely because they may
increase reimbursement.
It should not conceal uncertainty.
It should not make decisions that unfairly disadvantage
patients with complex or unusual conditions.
And it should not compromise patient privacy in pursuit of
efficiency.
The objective is not to maximize reimbursement at any cost.
It is to support appropriate payment for legitimate,
properly documented care.
The right goal is not getting paid more for the same
care. It is getting paid correctly without unnecessary administrative friction.
That distinction matters.
Especially when technology companies promise to
revolutionize healthcare.
Revolution is a large word.
Accuracy is a more useful starting point.
Myth Busters: Eight Beliefs That Deserve a Second Opinion
Myth 1: More AI means fewer billing problems.
Reality: More AI can also mean more automated
mistakes. The outcome depends on data quality, workflow design, and oversight.
Myth 2: A paid claim means the process was efficient.
Reality: A claim may be paid only after substantial
administrative rework.
Myth 3: Medical billing begins after the patient visit.
Reality: Revenue cycle processes begin before care,
including registration, eligibility, and authorization.
Myth 4: Structured data is trustworthy data.
Reality: Information can be consistently formatted
and consistently wrong.
Myth 5: Outsourcing billing eliminates administrative
responsibility.
Reality: Outsourcing changes who performs the work,
not the practice's need for oversight.
Myth 6: The best billing company recovers the most
denials.
Reality: Recovery performance matters, but so do
preventable denials, cost to collect, and overall financial outcomes.
Myth 7: AI will replace all medical billing
professionals.
Reality: Automation may change roles, but complex
exceptions, compliance, payer disputes, and clinical questions still require
expertise.
Myth 8: Independent clinics need enterprise-level
technology to improve performance.
Reality: Many improvements begin with better
processes, clearer responsibilities, and reliable measurement.
The larger lesson?
Sophisticated technology cannot compensate indefinitely
for unsophisticated processes.
Why I Am Building OnnX Around a Different Question
As a physician-entrepreneur, I have become increasingly
interested in the gap between clinical work and administrative outcomes.
Physicians deliver care.
Then the information describing that care enters a
complicated system of codes, rules, software, vendors, and payer decisions.
Some complexity is necessary.
Much of it deserves scrutiny.
When thinking about OnnX, I could begin with the familiar
technology question:
How can AI make medical billing faster?
But I find another question more interesting.
Why does so much billing work need to happen at all?
That question leads upstream.
Toward eligibility.
Toward demographics.
Toward authorization.
Toward documentation quality.
Toward charge capture.
Toward reliable information exchange.
And toward systems that identify uncertainty before it
becomes an expensive problem.
OnnX is being developed around the premise that revenue
cycle management is partly an information-quality problem disguised as an
administrative workflow problem.
The ambition is to help independent clinics reduce
unnecessary intermediaries and administrative friction while improving
visibility into the revenue cycle.
That is a product thesis, not proof that every problem has
been solved.
The practical value must be demonstrated through real-world
workflows and measurable outcomes.
I am especially interested in understanding which upstream
interventions produce the clearest improvements.
Fewer preventable denials?
Less staff rework?
More reliable first-pass claim acceptance?
Better visibility into outstanding receivables?
More predictable cash flow?
Those are the outcomes that matter.
Not the number of AI features.
Not the complexity of the algorithm.
Not the sophistication of the sales presentation.
Physicians do not need more artificial intelligence to
admire. They need fewer administrative problems to manage.
And perhaps that is the most useful way to judge healthcare
innovation.
Not by how much technology we add.
But by how much unnecessary work we remove.
Future Outlook: The Next Revenue Cycle Revolution May Be
Invisible
Healthcare technology is moving toward greater automation.
AI models are becoming more capable.
Interoperability tools are improving.
Administrative workflows are becoming increasingly digital.
But the future may not look like a giant command center
filled with dashboards.
It may look surprisingly ordinary.
A patient arrives.
Insurance information is current.
Authorization requirements have been checked.
The physician documents care.
The relevant billing information is accurately captured.
The claim is validated.
Exceptions are identified.
Appropriate human review occurs.
The claim is submitted.
Payment is reconciled.
The clinic learns from problems that arise.
Nobody spends the afternoon searching for an insurance
number.
Nobody asks the physician to clarify information already
available in the record.
Nobody celebrates recovering a preventable denial.
The process simply works more reliably.
That may sound less exciting than autonomous AI agents.
But for an independent clinic, boring can be beautiful.
The highest form of administrative intelligence may be
making administration almost invisible.
Of course, payer disputes, regulatory complexity, and
unavoidable exceptions will remain.
The goal is not perfection.
The goal is to make preventable problems less common and
necessary work easier to perform.
That is a future worth pursuing.
Frequently Asked Questions
What is AI-powered medical billing?
AI-powered medical billing uses artificial
intelligence to assist with administrative and revenue cycle tasks, including
documentation review, coding suggestions, claim validation, denial analysis,
and workflow prioritization.
Its effectiveness depends on the underlying information,
implementation, and appropriate oversight.
Can AI eliminate medical billing denials?
No. Some denials arise from payer policies, coverage
restrictions, disputed medical necessity decisions, or other circumstances.
AI may help reduce certain preventable errors and support
denial management, but eliminating all denials is not realistic.
Why do claims get denied?
Claims may be denied because of eligibility issues, missing
authorization, documentation problems, coding discrepancies, coverage
limitations, medical necessity determinations, or payer processing errors.
The cause should be identified before choosing a solution.
What is the difference between denial prevention and
denial management?
Denial prevention focuses on identifying avoidable
problems before claim submission.
Denial management focuses on investigating,
correcting, appealing, or otherwise resolving claims that have already been
denied.
Both are necessary.
Why is data quality important in medical billing?
Medical billing depends on accurate patient, insurance,
provider, service, diagnosis, and administrative information.
Incorrect or incomplete information can lead to delays,
denials, rework, or improper reimbursement.
Is FHIR a medical billing standard?
FHIR is primarily a healthcare information exchange
standard.
It can support access to clinical information, but medical
billing also relies on other standards, transactions, code sets, and
payer-specific requirements.
Can small clinics benefit from AI?
Potentially.
The strongest use cases are those that address measurable
problems without creating excessive implementation costs or staff burden.
What should a clinic measure first?
Start with first-pass claim acceptance, denial rate, A/R
aging, rework hours, and preventable denial categories.
Use consistent definitions and establish a baseline.
Should physicians outsource billing?
Outsourcing may be appropriate depending on staffing, cost,
specialty, internal expertise, and vendor performance.
The practice should retain adequate visibility and
oversight.
Can AI select medical diagnosis codes automatically?
Some systems can suggest codes.
However, recommendations must be supported by documentation
and reviewed as appropriate.
AI should not invent clinical facts or recommend unsupported
diagnoses.
Will AI replace medical billers?
AI may automate repetitive tasks and change the
responsibilities of billing professionals.
Complex coding, payer disputes, compliance decisions, and
exception management will continue to require expertise.
How can physicians reduce administrative burnout?
Begin by identifying recurring interruptions and rework.
Improve workflows, clarify responsibilities, eliminate
unnecessary duplication, and introduce technology only where it reduces
meaningful burden.
What is the best AI billing platform?
There is no universal best platform.
The right choice depends on the practice's needs, existing
systems, compliance requirements, costs, and evidence of measurable
improvement.
How long does revenue cycle improvement take?
A focused 60- to 90-day assessment and improvement cycle
can help evaluate specific changes.
More complex financial results may require longer
observation.
What is the biggest mistake clinics make when adopting
AI?
Buying technology before identifying the operational
problem.
The most expensive software is often the software
purchased without a clear reason to use it.
Final Thoughts: Stop Celebrating the Fire Department.
Start Preventing Fires.
Healthcare has developed remarkable capabilities.
We can image the human body in extraordinary detail.
We can detect diseases earlier.
We can perform complex procedures with extraordinary
precision.
We can develop AI systems capable of interpreting
increasingly sophisticated clinical information.
Yet an independent physician can still spend weeks trying to
get paid because an insurance identifier was outdated.
There is something almost comical about that contrast.
Almost.
Because the consequences are real.
Administrative waste consumes staff time.
Delayed reimbursement creates financial uncertainty.
Repeated documentation requests interrupt physicians.
And operational complexity makes independent practice harder
to sustain.
We should not accept these problems as unavoidable simply
because they have become familiar.
The industry has invested heavily in recovering denied
claims.
Now it should invest more attention in preventing avoidable
denials.
It has invested in processing information faster.
Now it should invest in making that information more
trustworthy.
It has invested in making billing systems more intelligent.
Now it should ask whether those systems can make billing
less burdensome.
The future of healthcare AI should not be measured by how
many mistakes it can correct. It should be measured, in part, by how many
mistakes never happen.
Because the ultimate goal is not to build a more impressive
billing industry.
It is to give physicians more time to practice medicine.
And that may be the most important innovation of all.
Get Involved: The Conversation Independent Physicians
Need to Have
A question worth debating: Has healthcare become too
comfortable paying people and software to repair preventable administrative
mistakes instead of redesigning the processes that create them?
Have your say: If you own or manage a clinic, comment
with the single biggest source of administrative rework in your practice:
eligibility, authorization, documentation, coding, or payer denials.
Help change the conversation: Repost this article if
you believe independent physicians deserve fewer billing headaches, better
information, and technology that removes unnecessary work rather than creating
more of it.
About the Author
Dr. Daniel Cham is a physician, medical consultant,
and healthcare entrepreneur whose interests span medical technology, healthcare
management, medical billing, and the practical application of artificial
intelligence.
He is the founder of OnnX, an AI-powered medical
billing SaaS initiative focused on helping independent clinics address
administrative inefficiencies, improve information quality, and reduce
unnecessary intermediaries.
His work explores a central question: How can healthcare
technology remove complexity rather than add another layer of it?
Through writing, research, and professional discussions, Dr.
Cham shares practical perspectives on healthcare innovation, physician
entrepreneurship, revenue cycle management, and the future of independent
medical practice.
Connect with Dr.
Daniel Cham on LinkedIn
P.S. Visit the Featured section of my LinkedIn
profile for a free resource. No signup is needed here.
Continue the Conversation Beyond LinkedIn
Healthcare improves when physicians, innovators, and
practice leaders challenge assumptions and exchange practical ideas.
For more insights into medicine, technology, healthcare
operations, and the realities of building better systems, explore the
following channels.
Personal website: DrDanielCham.com
Podcast: Listen on Spotify
Video discussions: Subscribe on YouTube
Professional perspectives: Follow on X | Facebook | Substack
Short-form insights: TikTok | Instagram | Threads
Community discussions: Reddit | Quora
Better information creates better decisions. Better
decisions create better systems. Better systems give physicians more time to
care.
Professional Note
This article is intended for education, commentary, and
professional discussion. It does not provide individualized medical, legal,
financial, coding, or regulatory advice. Healthcare organizations should
consult qualified professionals and authoritative guidance before implementing
changes affecting clinical care, reimbursement, privacy, or compliance.
Illustrative examples are not representations of actual OnnX customer outcomes.
References and Further Reading
1. CMS:
Calendar Year 2027 Medicare Physician Fee Schedule Final Rule
Outlines changes to Medicare physician reimbursement, coding
requirements, remote monitoring, and payment policies. Essential context for
understanding how evolving regulations affect revenue cycle management and
independent practice operations.
2. CMS
Finalizes CY 2027 Physician Fee Schedule to Shift from Sick Care to Healthcare
and Curb Waste
Explains CMS's policy direction toward preventive care,
payment accountability, and reducing unnecessary healthcare spending,
reinforcing the need for efficient and accurate administrative processes.
3. CMS
Issues CY 2027 Physician Fee Schedule Final Rule
Provides an industry perspective on Medicare payment changes
and their implications for healthcare organizations, physician practices, and
operational planning.
4. Your
Billing Denial Has a Plot Twist: The Problem May Have Started Before the Claim
Existed
Explores how upstream data errors, fragmented workflows,
and information loss can create billing failures long before claims are
submitted. Extends the article's argument that prevention begins with
trustworthy information.
5. The
Denial Isn't the Problem. It's the Clue.
Reframes claim denials as signals of deeper operational
weaknesses. Examines denial patterns, root-cause analysis, and opportunities
to shift from reactive recovery toward proactive prevention.
6. Anthropic
Wants More AI Regulation. Physicians Should Be Asking Why.
Examines AI governance, physician accountability,
transparency, and regulatory oversight, highlighting why healthcare
organizations must evaluate more than efficiency when adopting intelligent
systems.
#HealthcareAI #MedicalBilling #RevenueCycleManagement
#DenialPrevention #IndependentPractice #PhysicianEntrepreneur
#HealthcareInnovation #HealthcareTechnology #MedicalPracticeManagement
#HealthcareData #PhysicianLeadership #ClinicalDocumentation #HealthcareAutomation
#PracticeManagement #HealthInformatics
