Trust & control
Capable by design. Limited on purpose.
An AI Employee you cannot inspect is a liability. Ours carry an identity, narrow permissions, written boundaries, named approvers and a trail for every action.
Controlled intelligence
Six questions, answered before go-live.
Before a role goes live, each of these has a written answer. Select one to see it on the Core.
Every AI Employee has a name, a role and a voice. It never pretends to be a person.
Example · Support AISupport AI for the front desk, French, Darija, Arabic and English.
Principles
How we handle your data.
JDM does not hold a security certification today. When it does, the certificate will be published here. Until then, these are the engineering rules every project follows.
- Data
- We map where information goes before anything is built, and keep it in the fewest places possible.
- Access
- Each AI Employee sees only the sources its role needs.
- Permissions
- Actions are granted one by one, never as blanket access.
- Human oversight
- People stay in control of money, commitments and exceptions.
- Auditability
- Every action can be traced to a source, a rule and an owner.
- Privacy
- Customer information is used to serve that customer, and not for anything else.
Autonomy
Autonomy is earned, not assumed.
Every role starts in assisted mode. It moves up a level only when its own trail shows it is ready, and you decide when.
Assisted
The AI Employee drafts. A person sends. Nothing leaves without a human click.
Semi-autonomous
Routine cases run on their own. Anything outside the rules waits for a named person.
Autonomous
The full task runs end to end inside fixed boundaries, and every action is logged for review.