Regulated & high-trust · website AI assistant
AI chatbot for Healthcare & clinics
Updated
The short version
Answer practical patient questions — hours, locations, preparation, insurance, referrals — from your own pages, and hand off everything clinical.
A clinic's website gets two kinds of question. The first is logistical: opening hours, parking, what to bring, whether you take a given insurer, how a referral works. The second is clinical, and no automated system should be answering it.
Most chatbot deployments in healthcare fail by blurring that line. The safe design draws it hard: answer the logistics from published content with a citation, and route anything clinical to a human, immediately and visibly.
MySiteGPT is built around content-only answers, so that boundary is the default rather than a prompt you have to defend. It answers from your practice pages and patient information, cites them, and returns your fallback — "please contact the practice" — the moment a question goes beyond what you've published.
What clinics and healthcare providers get asked
Real questions, in visitors' own words. Every one of these is answerable from content you probably already publish.
- “What are your opening hours, and are you open on Saturdays?”
- “Do you accept my insurance?”
- “How do I prepare for this procedure?”
- “Do I need a referral to book an appointment?”
- “Where do I park, and is the entrance accessible?”
- “How do I get my results?”
How MySiteGPT helps clinics and healthcare providers
Takes the phone-line load off reception
Hours, location, preparation instructions, and insurance questions are answered on the site, so the phone is free for the calls that need a person.
A hard line at clinical questions
It answers only from your published content. Symptom and treatment questions fall outside it, hit the fallback, and route to your practice — by design, not by luck.
Preparation instructions, exactly as written
Fasting, medication, and appointment-prep guidance is cited to your own patient information page, so nobody follows a paraphrase before a procedure.
Patients get answers in their language
Replies come back in the visitor's language across 50+ languages, still grounded in the patient information you published.
What to train it on, in order
The answers are only as good as the index behind them. Start at the top — these are ordered by how much question volume they absorb per page trained.
- 1
Practice information — hours, locations, access, parking
The highest-volume calls reception takes, and completely stable content.
- 2
Insurance, billing, and referral pages
Fiddly, frequently asked, and easy to get wrong verbally. Cite them.
- 3
Patient preparation and aftercare instructions
Where citation matters most. Train the exact published wording, ideally as Q&A pairs.
- 4
Your fallback message, written carefully
Not a source, but the most important sentence in the setup — it should name the phone number and say plainly that clinical questions need a clinician.
Not sure your pages are ready to be retrieved from? Run one through the RAG readiness checker first, or audit the whole site with the sitemap analyzer.
What to measure once it's live
We don't publish other people's numbers as if they were yours. Track these instead — they are all visible in your own dashboard within a fortnight.
- Reception call volume for logistical questions
- Out-of-hours questions answered on the site
- Fallback rate on clinical questions — this should be high, and that's correct
- Unanswered logistical questions, which point at missing patient information
AI chatbot for Healthcare & clinics: frequently asked questions
Can it give medical advice?
No. It answers only from the content you train it on, and clinical judgement isn't in that content. Write your fallback to say so explicitly and to give the contact route — the assistant will use it whenever a question falls outside your published material.
Does it handle patient data or health records?
It has no access to any record system. It answers from published pages. Conversations and any details a visitor types are stored on your account, so treat the widget like any other public web form and don't invite patients to type clinical details into it.
Is it HIPAA or GDPR compliant?
Compliance depends on how you deploy it, not on the tool alone. There's no BAA on offer, so the safe deployment is public, non-clinical information only, with lead capture limited to what your privacy notice covers. Review the privacy policy with whoever handles your data protection before going live.
Can it book appointments?
Not directly. It explains how booking works, cites the page, and captures a name and email so your existing booking process picks it up.
Try it on your own content first.
The chat bubble on our homepage is the product, trained on our own site — ask it anything and watch every answer cite its source. Then train one on yours: 14 days, full features, no card.