PHI encrypted, always
Protected health information is encrypted in transit and at rest. Patient data is never used to train shared models.
Patient data stays protected. Your team stays in control.
Doctors and clinic owners are right to be careful about patient confidentiality. This page explains exactly how Caddie handles PHI, what the platform does and does not do, and how your team trains and supervises it.
HIPAA compliance
How we keep patient data secure.
Caddie is built to handle protected health information the way HIPAA requires — and every safeguard here is one you can inspect on a demo.
Role-based access, full audit logs
Every member of your team sees only what their role allows, and every access is logged. You can always answer who saw what, and when.
Every conversation recorded
Full transcripts of every call, chat and message are logged and searchable. Nothing the AI says to a patient is invisible to you.
Answers only from approved knowledge
Your AI employees answer from the treatments, pricing and policies your clinic has signed off on. If it is not in their knowledge, they do not say it.
Never medical advice
Hard rules, not preferences: no diagnoses, no treatment recommendations, no clinical judgment. The moment a conversation turns clinical, it goes to your staff with the full history attached.
99.99% uptime, monitored
The platform is monitored around the clock and runs on the latest frontier AI models, with each task routed to the model best suited for it.
A clear line
What the platform does, and what stays with your team.
The boundary is explicit, and you control it. Caddie runs the front office; your team owns every clinical decision and the patient relationship.
The platform handles
- Answering inquiries at 2am, and in seconds at 2pm
- Follow-ups that would otherwise slip through the cracks
- Scheduling, reminders, confirmations, reschedules
- Inbox triage, social replies, review requests
- Logging every conversation for your review
Your team keeps
- Every clinical conversation and consult
- Diagnosis, treatment advice, and medical judgment
- The patient relationship itself
- Any conversation the AI is not certain about
- Final say over what the AI can and cannot do
Training & control
You train it. You supervise it.
Trust should not rest on promises. It rests on mechanisms you can inspect — here are the four that govern every Caddie AI employee.
Trained on your clinic before it goes live
Your AI employees learn your services, pricing, policies and tone in a training environment, and talk to no patient until you have reviewed how they answer.
Correct it once, in plain English
Coach it the way you would coach a new hire. The correction applies immediately, permanently, and it does not repeat the mistake.
Escalation rules you write
Decide which topics always go to a human: pricing negotiations, complaints, specific treatments. Your rules are enforced on every conversation, on every channel.
Hands off instead of guessing
When a question falls outside its approved knowledge, it does not improvise. The conversation goes to your staff with full context — a warm handoff, not a dead end.
In practice: the platform runs the top of the funnel — answering, qualifying, booking, and following up. The moment a conversation needs clinical judgment, or a patient walks through your door, your team takes over with the full history in front of them.
Performance & reliability
Built for the industry where a wrong answer isn't an option.
Caddie's AI employees are trained specifically for healthcare, not adapted from out-of-the-box agents. Because they only speak from knowledge your clinic has approved, they don't make answers up — when they don't know, they say so, and bring in your staff.
99.99%Uptime, monitored around the clock
100%Of conversations logged and reviewable
0Medical advice given, ever — by design
See the safeguards for yourself.
Book a demo and we will show you the knowledge base, the transcripts, the audit logs, and the handoff — live.