Inquiry triage and case extraction
Reads a free-text message and returns a structured case: concern, medications, conditions, allergies, dates and what is still missing. Uses Claude tool use for reliable structured output.
Pilot program for medical travel clinics
Answer every patient inquiry, in their language.
GraftlyAI is an AI coordinator for clinics that treat patients from abroad. It reads photos and history from WhatsApp and email, drafts replies in the patient's language, and gives your doctors a clean case summary. Humans approve every medical reply.
Powered by Claude, the AI model from Anthropic
Built on Claude by AnthropicHuman approval on every medical replyDesigned for GDPR, KVKK and HIPAA requirementsFull audit trail
From first message to a case your doctor can review
Patient writes in
WhatsApp, email, or your website form, in any language your patients use.
GraftlyAI collects the details
It asks follow-up questions, sorts photos and medical history, and marks anything that needs a doctor's attention.
Your team reviews
Coordinators approve the reply. Surgeons see a structured summary instead of a long chat thread.
Follow-up after the procedure
Scheduled check-ins, progress photos sorted by date, and alerts when a patient reports something unusual.
Try the triage simulator
Paste a sample patient message or pick an example. The simulator turns it into a case summary and a draft reply, the way the product does.
Use fake data only. When the live demo is available, your text is sent through our server to the Claude API to produce the result, and it is not stored. Otherwise a simple rule-based simulation runs in your browser.
Results appear here.
Built around how clinics actually work
Multilingual replies
Draft answers in the patient's language, in your clinic's tone.
Case summaries
Concern, history, medication, and dates in one clear card.
Photo intake
Patients get guidance on which angles to send, so fewer photos get rejected.
Aftercare tracking
Reminders and check-ins that match each patient's recovery timeline.
Multi-office inbox
One place for your main clinic and consultation offices abroad.
After-hours response
Inquiries from other time zones get a first answer right away.
Where Claude powers GraftlyAI
Each module uses the Claude API for one specific job, inside strict rules and with a person in the loop. The label shows what exists today.
Multilingual reply drafting
Writes a short, warm draft in the patient's own language and asks at most two follow-up questions. No prices, no diagnosis.
Safety screening and escalation
Flags medications, conditions and post-operative red flags such as fever, bleeding or severe swelling, and routes them to a person at once.
Doctor handoff summaries
Condenses a long conversation into a one-page brief for the surgeon, so nothing hides in a chat thread.
Photo intake guidance
Uses Claude's vision to check whether photos show the right angles and lighting, then asks the patient for what is missing. It does not judge the medical case.
Aftercare check-in analysis
Reads patient updates after the procedure and highlights anything that departs from the expected recovery for staff to review.
Document and report extraction
Pulls medication lists and relevant history out of PDFs and scans that patients upload, for a doctor to verify.
Clinic knowledge answers
Answers general logistics questions (transfers, hotel, what is included) using only your clinic's approved content, and hands everything else to staff.
In every module AI drafts and extracts, and people decide. We recommend testing each module on sample data before any real patient data is used.
How the pieces fit together
Channels
WhatsApp, email, web form
Intake
Language detection and consent notice
Claude API
Extraction, drafting, safety check
Human approval
Coordinator and doctor review
Send and log
Reply goes out, every step is recorded
A day at the coordinator desk, before and after
Today
- Photos and history are scattered across WhatsApp threads.
- Replies wait until someone who speaks the language is online.
- Doctors read long chats to find the medication line.
- Aftercare check-ins depend on someone remembering.
With GraftlyAI
- Every inquiry becomes one case with photos and history attached.
- A draft reply in the patient's language is ready for approval.
- Doctors open a summary with red flags already marked.
- Follow-ups are scheduled and urgent reports are escalated.
Human approval on every medical replyUrgent symptom escalationAudit log of every draftConsent-first data handling
Made for clinics that treat patients from abroad
Hair restoration
Photo-heavy inquiries, multiple languages, and long recovery follow-ups.
Dental travel
X-rays and treatment history collected before the patient books a flight.
Cosmetic surgery
Careful screening of medical history, with a doctor reviewing every case.
Your coordinators' new inbox
Every conversation becomes a case with a status, a language, and a clear next step. Product preview with sample data.
Open casesSample data
| Patient | Language | Channel | Status | Next step |
|---|---|---|---|---|
| Marco T. | Italiano | Doctor check | Review medication | |
| James R. | English | Photos needed | Request side angle | |
| Ali K. | Türkçe | Ready for plan | Surgeon sign-off | |
| Sara L. | English | Web form | Aftercare day 14 | Send progress check |
Channels
WhatsApppilot targetEmailpilot targetWeb formpilot targetCalendarlater
Estimate the time your team could save
Adjust the numbers to match your clinic. This is an illustrative estimate, not a guarantee. The saved-time share is an assumption you control.
40 h
staff hours saved per month
A simple, inspectable pattern
The core step is one Claude API call that returns a structured case through tool use. The API key stays on the server and never reaches the browser.
# Illustrative. The same pattern runs in the live demo on this page.
import anthropic
client = anthropic.Anthropic() # API key comes from the server environment
resp = client.messages.create(
model="claude-sonnet-5-5",
max_tokens=900,
system=RULES, # no diagnosis, no prices, escalate red flags
tools=[RECORD_CASE], # structured output through tool use
tool_choice={"type": "tool", "name": "record_case"},
messages=[{"role": "user", "content": patient_message}],
)
case = next(b.input for b in resp.content if b.type == "tool_use")
# Staff review case["draft_reply"] before anything is sent.
Doctors decide. GraftlyAI prepares.
No diagnosis, no price quotes
Medical answers and prices only go out after your staff approves them.
Red-flag escalation
Medications, allergies, and health conditions are flagged for a doctor before any plan is made.
Privacy by design
We are designing around consent and data minimization, with GDPR, KVKK and HIPAA requirements in mind.
Full audit trail
Every AI draft and every human approval is logged.
Compliance roadmap
We do not claim certifications we do not hold. This is what we plan to have in place before any real patient data is processed.
PlannedPatient consent and a clear notice at intake
PlannedData processing agreement with every clinic (GDPR, KVKK)
PlannedData minimization and retention limits
PlannedReview of HIPAA requirements for US-facing clinics
What GraftlyAI never does
- Diagnose a condition or recommend a treatment.
- State prices or graft counts on its own.
- Promise results.
- Send a medical message without staff approval.
Where we are
Now
Working prototype
Inquiry triage, case summaries, and approval flow, tested on sample data.
Next
Pilot clinics and aftercare module
Real workflows with partner clinics, plus automated follow-ups and progress photo tracking.
Later
More clinics and specialties
Dental, cosmetic, and other medical travel providers.
Built by people who know the inquiry inbox
GraftlyAI is built by a team with hands-on experience handling multilingual patient inquiries at a medical travel clinic. We are building the tool we wished we had.
Questions clinics ask
Does GraftlyAI replace our coordinators?
No. It handles the repetitive first steps so coordinators spend their time on patients who are ready to book.
Can it give medical advice?
No. It collects information and drafts replies. A doctor reviews anything medical.
Which channels does it support?
WhatsApp, email, and web forms are the pilot targets.
How is patient data handled?
Before any real patient data is processed, we put in place patient consent, a data processing agreement with each clinic, and safeguards required by laws such as GDPR, KVKK and HIPAA, as they apply. This website does not process patient data.
Which AI does it use?
GraftlyAI is built on Claude, made by Anthropic, for multilingual understanding and structured case extraction.
What does it cost?
Pilot clinics will help us set pricing. Tell us your monthly inquiry volume and we will discuss options.
See it on your own inquiries.
Join the pilotHelp shape the first version
Tell us about your clinic and how you handle inquiries today. Please do not include patient information.