AI for OMS Administrators:
Practical Uses Without the Compliance Panic
AI tools are already changing how practices communicate, organize information, and manage workflows.
The goal is not to replace people. The goal is to reduce repetitive work, save time,
and help teams focus on higher-value tasks.
The key is learning where AI can help… and where it absolutely should not be used.
Use this guide as a practical starting point for exploring AI safely and strategically in your OMS practice.
Disclaimer: The information and examples on this page are provided for educational purposes only and should not be considered legal, medical, dental, financial, or professional advice. AI-generated content may be inaccurate, incomplete, or outdated and should always be reviewed by an appropriately qualified professional before being relied upon, distributed, or used in patient care, business operations, or compliance decisions. Users are responsible for verifying the accuracy and appropriateness of all AI-generated content and for ensuring compliance with applicable laws, regulations, and organizational policies.
3 Basic Rules
Regardless of the AI system you are using, following these three rules can save you lots of trouble.
- Human in the middle: Always verify accuracy before distributing.
- Don't ask, “Is AI involved?” Ask “Is the workflow accurate, transparent, and compliant?”
- Never assume a tool is compliant because it says “AI-powered.”
Good Places to Start Using AI
Social Media & Marketing
Low risk. High time savings.
Examples:
- Social captions
- Blog drafts
- Referral office newsletters
- Event promotions
- Patient education content
- Rewriting long text into shorter formats
Reminder: Always review for tone, accuracy, and professionalism before posting.
Hear industry experts discuss practical AI applications on our podcast:
- "The Future of OMS Marketing: Harnessing AI and Digital Strategy" with hosts Jamie Leos (OMS Powerhouse Consulting) and Nathan Greenberg (Firm Media) and guest Chris Suchánek (Firm Media).
- "The Administrator's Secret: Automating Busy Work to Humanize the Patient Journey" with host Daniel Boyce (Digimatiq Marketing).
Email Drafting
AI can help organize thoughts and improve clarity.
Good uses:
- difficult email drafts
- policy announcements
- meeting recaps
- staff reminders
- rewriting emotional emails into calmer language
Do not enter into public AI systems:
- patient names
- complaints tied to individuals
- HR documentation
- confidential contracts
Meeting Notes & Action Items
Use AI tools for:
- team meetings
- brainstorming sessions
- vendor calls
- training sessions
Then generate helpful content, such as:
- summaries
- task lists
- follow-up reminders
- cleaned-up notes
- onboarding/staff training guides
AI-generated content may contain errors. Ensure all medical and scientific information is reviewed by appropriately licensed or otherwise qualified professionals before distribution or use.
Spreadsheet Help
AI is useful for:
- Excel formulas
- Google Sheets troubleshooting
- conditional formatting
- report cleanup
- organizing exports
- creating templates
This may be one of the safest and highest-ROI entry points for administrators.
Insurance Narratives & Appeals
AI can help:
- organize arguments
- improve readability
- simplify repetitive writing
But!
- Never paste complete patient records, radiographs, or identifying details into public AI systems.
- Use generalized language and templates whenever possible.
Call Management & Phone Systems
Many practices already interact with AI without realizing it. Voicemail transcription, after-hours answering, intake, call summaries, and scheduling can all involve PHI. Verify HIPAA compliance with your vendors, confirm a signed Business Associate Agreement (BAA) when PHI is involved, and understand how patient data is stored, accessed, and retained.
Examples:
- call routing
- voicemail transcription
- after-hours answering
- scheduling assistance
- FAQ automation
The important question is not “Is AI involved?” It’s “Is the workflow accurate, transparent, and compliant?”
Some phone service providers reference the HIPAA Conduit Exception to avoid signing a BAA. That exception is narrow and generally applies to transmission-only services with only temporary storage incidental to transmission. If call recordings, voicemails, transcripts, summaries, or patient data live on the vendor's servers, practices should expect a BAA to be in place.
Patient Charting & Documentation
Extreme caution area.
AI scribes and charting assistants are expanding rapidly, but practices should:
- verify HIPAA compliance
- review vendor BAAs carefully
- understand where data is stored
- confirm whether recordings are retained or used for training
Never assume a tool is compliant because it says “AI-powered.” For OMS administrators, vendor vetting should include whether the tool is built for oral surgery, integrates with the practice's PMS, reduces duplicate data entry, supports customizable workflows, signs a BAA when PHI is involved, and solves a specific operational problem rather than adding another disconnected platform.
This page has been reviewed by SOMSA’s IT and cybersecurity partners listed below. If you have any questions, don't hesitate to reach out to info@somsa.org or the experts below.
5 Low-Risk Ways to Experiment With AI This Month
- Ask AI to draft a social media post about an upcoming office event.
- Use AI to summarize meeting notes into action items.
- Ask AI to help build an Excel formula or reporting dashboard.
- Use AI to rewrite a long email into a shorter, clearer version.
- Test AI-generated brainstorming:
- staff appreciation ideas
- referral marketing themes
- patient FAQ wording
- job posting drafts
5 Smart Rules for Using AI Responsibly
1. Never Enter PHI
Rule of Thumb: If you wouldn’t post it publicly, don’t paste it into a public AI tool.
Do not enter:
- patient names
- birth dates
- medical record numbers
- radiographs
- clinical photos
- treatment details tied to identifiable patients
In addition to PHI, handle other personally identifiable information (PII) with care. Call recordings, voicemails, and transcripts can contain all of the below, and removing a patient's name is not enough if other details can still identify the patient.
- phone numbers
- email addresses
- home addresses
- appointment/service dates
- insurance/member IDs
2. Turn Off AI Training When Possible
Many AI platforms allow you to disable use of your conversations, uploaded files, or data for model training.
This reduces—but does not eliminate—risk. Review privacy settings carefully.
Free consumer AI tools are usually not appropriate for professional or healthcare workflows. If the product is free, the practice should assume the vendor's business model may involve using data, usage patterns, or user activity in ways that are not appropriate for sensitive business or patient information. Unfortunately, the free tools often do not provide the option to disable the model training or selling of aggregate data.
3. Treat AI Output Like a First Draft
AI can sound confident while being wrong.
Always review:
- facts
- compliance language
- tone
- calculations
- legal or insurance statements
Human oversight matters. According to Seth Godin, "The question is now, 'Who approved this?' not 'who made this?'"
4. Avoid Uploading Sensitive Business Information
Do not upload:
- payroll data
- contracts
- reimbursement rates
- legal disputes
- disciplinary records
- private financial reports
AI convenience is not worth creating operational risk.
5. Start Small
The best AI implementations are usually boring at first.
Save 10 minutes here, reduce one repetitive task there.
That compounds faster than most people expect.
The Bottom Line
AI is not replacing OMS administrators. But OMS administrators who learn how to use AI thoughtfully may eventually outperform those who refuse to engage with it entirely.
The sweet spot is neither panic nor blind trust. It’s informed experimentation.