How Drungtsho decides.
A short, transparent explanation of the parts of the system, what each part is allowed to do, and where humans stay in charge.
- Listen
The AI interprets your wording. It maps free text to a supplied intention taxonomy. It does not decide eligibility.
- Safety rules
A deterministic rule system decides whether product-fit guidance is permitted. Rules run on the server and cannot be changed by the language model.
- Approved records
Only published products, active labels, approved claims, and reviewed evidence can be used.
- AI explanation
The AI turns validated facts into a short, bounded explanation. It cannot invent, paraphrase, or upgrade a claim.
- Server validation
Every ID, warning, direction, and disclosure is validated before rendering. If validation fails, we fall back to a safe, non-personalized response.
- User decision
You decide. You can also request deletion, contact a human, or report a concern.
The system may make mistakes and is not medical care.