Use cases by industry2 use cases
AI automation for clinic and practice administration
A multi-site physiotherapy or imaging group receives around 300 referrals a week as faxes, letters and PDFs. Front-desk staff re-key patient, insurer, diagnosis codes and the requested service into the practice system, and an incomplete referral sits in a pile until someone calls the referring doctor. Meanwhile late cancellations leave slots empty while the same staff phone waitlisted patients one by one. Referral intake and a waitlist agent take those two jobs.
The US market is served by funded products: Tennr raised $101M at a $605M valuation for referral intake, and Assort Health raised $76M in October 2025 for patient phone calls. European clinics run different systems under different rules: Doctolib for booking in France, Germany and Italy, THEORG or Starke in German physiotherapy, Dampsoft or CGM Z1 in dental, Cliniko, Semble or WriteUpp in the UK, and KIM for secure messaging between German practices.
The line I hold on every healthcare build is simple: administration yes, clinical judgment no. The system reads, sorts, checks completeness, books and chases; clinicians triage, diagnose and decide. Health data is special-category data under GDPR Article 9, so EU hosting, a data protection impact assessment and strict access control are part of the build, not extras.
What I would build for healthcare admin
Referral faxes and letters turned into patient records, with missing details chased automatically
Reads each incoming referral, checks it against what the exam needs, links the right patient and chases the referrer for gaps; clinicians keep every clinical call.
Fax-to-email and practice inboxes / KIM (Kommunikation im Medizinwesen) / RIS or practice management system / Scheduling / EU-hosted extraction and review service
Cancelled clinic appointments refilled from the waitlist before the slot goes empty
When a patient cancels, the agent offers the slot to waitlisted patients who fit, books the first confirmed reply and tells everyone else the slot is gone.
Practice management system (Cliniko, Jane, Doctolib or a German practice system) / WhatsApp Business Platform / Twilio SMS / Google Calendar / Waitlist database
The admin work that fills a clinic's day
Referrals that arrive incomplete
Each service has its own requirements: a prescription with specific fields, an insurer approval, prior imaging. Checking completeness on arrival and requesting the missing piece from the referrer the same day keeps referrals from going stale.
Prescriptions that cost money when wrong
In Germany, physiotherapy prescriptions with formal errors lead statutory insurers to cut payments. Checking each one against the therapy rules before the first treatment, and flagging it for correction with the prescribing doctor, is a rules problem with an extraction front end.
Eligibility and pre-authorization
Dental and physio front desks check insurer portals patient by patient, and UK private insurers such as Bupa and AXA Health want pre-authorization for outpatient care. Assembling the packet and tracking its status is automatable; the clinician attests to the clinical content.
Empty slots after cancellations
Matching a freed slot to waitlisted patients by treatment type, practitioner and availability, messaging them in priority order and booking the first valid reply fills gaps a phone list cannot.
Boundaries that do not move in healthcare
No clinical triage by the model
Urgent-sounding referrals are flagged to a clinician at once. The model never decides priority, diagnosis or whether a patient needs to be seen.
Messages carry no medical detail
Waitlist and reminder messages say an appointment is free on Tuesday at 10:00 and nothing about the treatment, because WhatsApp and SMS are no place for health data.
Patients are never merged automatically
Two patients with similar names and birth dates are a classic error. The system proposes a match, and a person confirms it before records are linked.
Scribes and phone bots are products
Ambient scribes are already off the shelf, and US patient phone agents are well funded. I would buy those and build the integration and document work around them.
Software these builds usually connect to
The systems do not get replaced. The build sits across them, reads from them through their APIs, and writes results back.
Doctolib / Cliniko / Jane / Semble / WriteUpp / THEORG / Dampsoft / CGM Z1 / Epic / athenahealth / WebPT / KIM
Frequently asked questions
Can AI process patient referrals and faxes?
Yes, as an administrative workflow. It classifies each incoming document, extracts patient, insurer, referring doctor, diagnosis codes and the requested service, checks completeness against that service's requirements, creates or updates the record and requests anything missing. It makes no clinical decisions: urgent referrals are flagged to a clinician immediately, and uncertain patient matches go to a person.
Is it GDPR compliant to use AI with patient data?
It can be, with the right design. Health data is special-category data under Article 9, so you need a lawful basis, a data protection impact assessment, a processing agreement with every provider, EU hosting or self-hosted models, strict access control and minimal data in messages. That design work is most of the effort, and it has to happen before the first patient record is processed. More on how I approach it under GDPR-compliant AI development.
Should a clinic use Doctolib's or Cliniko's waitlist feature instead of a custom agent?
If it fits, yes. Doctolib, Cliniko and Jane have waitlist and reminder features that work for a single practice with ordinary rules. A custom agent earns its cost for multi-site groups with rules those features cannot express, such as practitioner preferences, treatment durations and priority categories, or for practice systems that have no waitlist at all.
Does an AI that messages patients have to say it is an AI?
In the EU, yes. Since 2 August 2026, Article 50 of the EU AI Act requires telling people when they are interacting with an AI system, unless it is obvious. A waitlist or reminder agent says so in its first message, and a patient can always reach the front desk instead. Beyond disclosure, the messages stay free of treatment details.
Also browse: Document processing · Operations and logistics · WhatsApp · all use cases