AI for healthcare
AI for healthcare practices, starting with the front desk
Clinical AI gets the headlines. The hours in a private practice are lost at the front desk and in the fax queue: reminders, referrals, prior authorizations, and the same portal questions all day.
The short answer
AI helps a healthcare practice by handling scheduling, recall outreach, referral intake, and prior authorization paperwork, and by drafting clinical notes for the clinician to sign. It can only touch protected health information through vendors that have signed a business associate agreement, and it should never give clinical advice on its own.
Where it pays off
Five workflows worth looking at first
- 01
Scheduling and recall
Books, moves, and confirms appointments against the real schedule, and works the recall list for patients overdue for a visit, with messages kept free of clinical detail.
- 02
Prior authorization packets
Pulls the diagnosis, history, and prior treatments a payer wants from the chart and assembles a draft packet. Staff check it and submit it.
- 03
Referral and fax intake
Reads incoming faxes and referral PDFs, matches them to a patient, and files them for review, with low-confidence matches held for a person.
- 04
Portal message triage
Sorts patient messages into refills, billing, scheduling, and clinical questions, then drafts replies for the routine ones. Anything clinical goes to a nurse or provider queue.
- 05
Visit note drafting
An ambient tool drafts the note from the visit and the clinician reviews and signs it. This is usually a product to buy, and our job would be picking one and fitting it to your EHR.
Before you build
HIPAA applies to every vendor in the chain. If a model provider, hosting company, or texting service touches protected health information, you need a business associate agreement with them before any patient data moves, and a consumer AI account does not come with one. Texts and voicemails should carry the minimum necessary, and logs and prompts count as PHI too. Triage is the other danger: an assistant that tells a patient chest pain can wait is a patient safety event, so symptoms route to clinical staff. Software that steers clinical decisions can also raise FDA questions.
This is a practical caution, not legal advice. Check with your own counsel.
Where our experience stops
We have not shipped a production system for a healthcare practice, and we would bring in your compliance lead from the first conversation rather than learn HIPAA on your patients.
Questions
What people ask us
Is ChatGPT HIPAA compliant?
No tool is compliant on its own. What matters is whether the vendor will sign a business associate agreement for the plan you use and how you configure it. Without a signed agreement, keep patient information out of it entirely, and check the current terms because they change.
What can AI automate in a medical practice?
Administrative work first: scheduling, reminders, recall, referral intake, prior authorization packets, and billing follow-up. Those tasks are high in volume and low in clinical risk. Leave diagnosis and treatment guidance to clinicians, and treat any AI output in the chart as a draft.
Can an AI receptionist answer calls for a doctor's office?
For scheduling and routine questions, yes, provided the vendor signs a business associate agreement and can connect to your scheduling system. It must recognize urgent symptoms and move those callers to staff or tell them to call 911 immediately. Test that path harder than any other.
Tell us what is eating your week
A free discovery call with a solutions architect, or a blueprint in about a minute.