Routine replies drafted in seconds. Anything clinical goes straight to a person.
Safety rules run first, every draft is fact-checked, and a person always presses send.
Concept for Dispensed. Not affiliated with Dispensed. All patients, orders and messages are fictional.

Two example messages, run through the checks
Loading the live example
How a message moves
From a patient's phone to a person's reply.

Step 1: A patient writes in
By email or chat: an order, a refill, or a question about their health.
- Personal details removed
- Safety rules
- Sorted by type
- Sources found
- Reply drafted
- Facts checked
- A person decides
Step 2: Care Desk checks it
Personal details removed, safety rules first, every fact checked against the records.

Step 3: A person replies
An agent sends routine replies. Anything clinical goes straight to a clinician, and the orders go on hold.
What keeps it safe
Safety rules run first
Every fact is checked
Dates, amounts, order and tracking numbers must match the patient's records, or the draft is blocked.A person always sends
AI only drafts routine replies. An agent reads each one, then sends, edits or passes it on.
The safety rules are real. The sorting and drafting shown here are samples.

Talk to me
I'm Chanon Poovaviranon (Beam). I designed and built Care Desk, from the safety rules to the test set, and I can set it up with your team.
Or write to chanonbcp@gmail.com
How I would set it up with your team
- Weeks 1 to 2Connect your helpdesk and order records, and agree the safety rules with your clinicians.
- Weeks 3 to 4Order-status drafts go live for one agent. A person still sends every reply.
- Days 31 to 90Add more routine message types and compare the weekly numbers with before.

