The human stays in control
On matters that commit the student (assessment, guidance, recommendation), the decision always rests with the advisor. The twin prepares, proposes and handles the routine. The human approves, rules and signs.
Product principles
One twin per staff member, in their name.
Every admissions advisor has their own agent, calibrated to their voice. So does every programme manager and every coach. Students always see a person's name.
The three principles
The three non-negotiable principles
The student always knows who is speaking
By default, a transparency line at the foot of every message prepared by a twin (AI Act Art. 50 compliant), and a “Write to [advisor's name] in person” button on the twin's public page. No student can be trapped in a dialogue with an agent.
Sensitive topics stay human
Distress, conflict, vulnerability, a legally binding matter, a delicate academic situation: the twin automatically steps back and alerts the advisor. It is written in the code.
The twin, step by step
Calibration · Action · Oversight · Learning
Calibration
It reads past work (emails, drafts, documents). A style guide is injected into every generation. Continuous recalibration as the advisor makes corrections.
For staff
Advisors run their twin from their phone
Decisions always remain human: the twin prepares drafts, the advisor approves, edits or rejects them. Five minutes in the morning, five minutes in the evening, like checking your notifications.
- Routine replies arrive as drafts ready to send: one read-through, one click.
- Committing replies and targeted follow-ups are edited in the advisor's voice before sending.
- Legal matters and exceptions to the regulations go straight to the advisor, with no reply generated.
- Distress, conflict, vulnerability: a human is mandatory, the twin steps aside and raises an alert.
9:11●●●·100%
← InboxMy drafts⚙
12 pending · 2 urgent
Draft · To approve3 min ago
To: Yuxin Zhao
Subject: Re: Enrolment deadline
您好赵雨欣，关于您的注册截止日期问题，我们的目标是在 11 月 15 日前确认……
Approvedsent
Carlos M. · Re: Application complete?
For recipients
Students know. The relationship stays intact.
By default, a transparency line at the foot of every message prepared by a twin. A “Write to [advisor's name] in person” button on the twin's public page. AI Act Article 50 compliant.
inbox · applicant
From: Sophie Martin · My School
Subject: Re: Enrolment deadline
Hi Yuxin,
Thanks for getting back to me. I'm aiming to confirm on 15 November, as we discussed. In the meantime, feel free to ping me if you have any questions about the visa or accommodation.
speak soon,
Sophie
Message prepared with the help of an AI assistant and reviewed by Sophie Martin before sending.
Why it is an advantage
An advantage for the institution
Failed AI deployments have one thing in common: an anonymous bot that nobody trusts. The twin removes this obstacle by design.
When an applicant receives a message in the name of Sophie Martin, programme advisor, with its transparency line and, on the twin's public page, a “Write to Sophie in person” button, they know that responsibility lies with an identified human.
And when Sophie leaves the institution, her twin belongs to her. It's in the contract.
Principles FAQ
The fundamental questions
Yes. By default, every message prepared by an advisor's twin carries a transparency line at its foot (“Message prepared with the help of an AI assistant and reviewed by Sophie Martin before sending.”), and the twin's public page offers a “Write to Sophie in person” button. It is Article 50 of the AI Act, implemented by design.
Let's discuss how it applies to your institution
We'll show you the principles in action on a real case (yours) in 30 minutes.