05/08/2026 David J. Freedman, DPM
What are the pitfalls of my EMR and templates used? (Lawrence Kosova, DPM)
I just came back from the 2026 AAPC Healthcon
Conference where AI was a significant topic of
discussion to over 3000 attendees. The purpose of
my post was to explain to the masses who use
Templates created by them or their EMR's was
really about the pitfalls and warnings they need
to be aware of in charting. So, let's be honest,
doctors do not change quickly when it comes to
charting opportunities. On the flip side, those
who are engaging AI need to understand it and the
fact of issues that have been brought up just at
this very conference this week.
While on the surface, AI can make each visit
"unique" to that patient's visit as you stated Dr.
Kosova. What the auditors are seeing is old habits
do not die. For example, "Copy & Paste", to save
time, many doctors just bring forward old
information even using AI. What most do not
understand is that information does not
necessarily count toward their E/M unless it
changed or has relevance at that specific visit.
So, I am sharing a list of potential concerns I
wrote down using AI that were discussed during the
2026 AAPC Healthcon Conference session:
1. AI is not a total solution. Copy/paste-forward
content is often not detected, and AI can
sometimes add work because all generated content
still requires review and validation.
2. Start slowly with defined workflows and phased
implementation. Processes must be vetted carefully
before broad rollout. Do not enable every AI
function immediately. Include input from your
practice stakeholders that might include RCM,
Compliance, and other physicians.
3. Consider establishing physician champions if in
a group or small practices instead of relying only
on the vendor or IT team.
4. If expecting the AI to properly code, make sure
this is correct even using demo tools. They should
prove they can accurately extract current ICD-10-
CM, CPT, and HCPCS codes from documentation.
Suggested codes must always be validated, with
efforts to avoid unspecified codes whenever
possible. Clear oversight is needed to determine
who is auditing for accuracy during
implementation.
5. AI is not realistically expected to reduce
FTE's immediately. The greater value may come from
digitization, workflow efficiency, and reduced
touch time. Content sources should come from
authoritative organizations such as AMA CPT and
CMS, just to name a few.
6. Expect gaps and errors early in implementation.
Organizations/Practices must identify issues
quickly and be prepared to disable functionality
if inaccurate documentation, coding, modifiers, or
billing occurs. One panelist through out an
example where they noted modifier automation had
to be turned off due to incorrect modifier
assignment.
7. Strong collaboration between coding,
compliance, physicians, and operational subject
matter experts is critical. Teams should meet
regularly to review charting, coding, compliance,
and workflow performance, with testing completed
early in the process.
8. Questions remain around what dashboards,
analytics, and measurable efficiencies currently
exist within your software AI, particularly
regarding RCM workflows versus physician-only
functionality. AI may improve RCM productivity by
reducing touch time and increasing chart
throughput per day.
9. Physician trust and engagement are essential.
Practice leadership should avoid forcing adoption
before workflows and risks are addressed.
Physicians do not want additional work or
increased compliance exposure.
10. Training and standards must be established in
advance. Adoption rates will vary, especially with
embedded AI tools and AI scribes. While AI
scribing may require minimal training, billing-
related AI functions require much stronger
validation, education, and governance.
11. AI scribes embedded within the EMR must remain
efficient with minimal clicks, or physician
adoption will suffer. Some organizations
successfully use multiple AI vendors and tools,
which improved physician engagement significantly
when providers were given choices. Physician
champions are important to help peers adopt and
troubleshoot technology with each other.
12. AI scribes and related tools should only be
used on secure practice company-approved devices,
such as corporate/practice iPads/computers, not
personal phones, due to PHI and security
requirements.
13. Job descriptions and responsibilities may need
updating for employees in RCM, Compliance, and
with physician roles that incorporate AI-supported
workflows.
14. Practice organizations should create SOPs and
governance processes for vendor evaluation and
engagement. Vendors should be questioned
thoroughly, and claims should be independently
validated rather than accepted at face value.
15. Payors are increasingly using AI for denial
management, making it important for organizations
to improve denial prevention and denial management
workflows through AI-enabled efficiencies as well.
16. Consider establishing an AI Practice
Governance Committee to oversee implementation,
risk, compliance, workflow standards, and adoption
strategies. Adoption may occur faster among
younger staff/physicians, but success depends on
demonstrating that AI helps teams “work smarter,
not harder”.
David J. Freedman, DPM, Silver Spring, MD
