White Coat World

This is the agreement you accept before using the writing tools. It is short on purpose. The short version: the words in your application have to be yours, and we have built the tool so that ours never become yours by accident.

What our AI does

Reads the draft you wrote and tells you what is working and what is weak.
Asks you pointed questions designed to pull specific, real detail out of your memory.
Flags clichés, vague claims, and telling-instead-of-showing in your own sentences.
Points you back at experiences you have already logged, so the fix comes from your record.

What our AI does not do

It does not write your personal statement, an opening line, a closing line, a paragraph, or any sentence you could paste into your application — not even as an example.
It does not write your activity or experience descriptions.
It does not invent experiences, patients, quotes, dates, or numbers. It can only coach on what you actually gave it.

How that rule is enforced

Two layers, and the second is the one that matters. First, the model is instructed in absolute terms never to produce prose in your voice. Second — because an instruction is not a guarantee — every response is inspected before it reaches you. If it contains drafted first-person prose, it is discarded and regenerated. If it fails twice, you get an explanation instead of an essay.

The writing analysis that flags clichés and vague language is not AI at all. It is fixed, deterministic logic, which is why the same draft always gets the same read.

What you agree to

You will write your own application. Coaching, feedback, and questions from this tool are aids to your writing, not a substitute for it.
You will not attempt to instruct, trick, or prompt the tool into drafting application prose for you. Attempts to do so are a violation of the Acceptable Use Policy.
You understand that admissions offices and their detection tools penalise machine-written prose, and that a statement which is not yours will not survive an interview.
You are responsible for the certification you sign with your application service. Nothing this tool produces relieves you of it.

What we do with what you write

We do not use your work to train or fine-tune AI models. The model is run by Google (Gemini) under its paid API terms, so Google does not use your text to improve its products and keeps it only to detect abuse. Before your text reaches the model, contact details and ID numbers are removed and the passage is length-capped. Full detail is on the Security page, and the wider commitments are set out in The Integrity Guarantee.

Honest limits

The output check described above looks for prose in your voice. It is deliberately tuned to avoid blocking legitimate coaching, which means it is possible for a borderline sentence to get through. If you ever receive something that reads like a draft of your statement rather than advice about it, tell us — that is a defect, not a feature.