How Scholexis uses AI
One feature, opt-in every time, with a screen that tells you exactly what is sent before anything leaves. What we send, what we never send, and what is coming.
Most products bury this. Ours is written on the screen at the moment it matters, and this page is the same text, so you can read it before signing up for anything.
There is one AI feature today
Breaking an assignment into steps. You press a button on an assignment, and a model drafts the steps. Nothing is added to your task list until you read the steps and accept them. You can edit them, delete them, or throw the whole draft away.
That is the only place AI touches your account right now. Everything else — your planner, your energy check-ins, your calendar, your accommodations — is ordinary software doing ordinary things.
Before anything is sent, you get this screen
Not a checkbox at signup. Every single time, before the request leaves:
Sent to the model Assignment title · a bounded description · assignment type · due date and time · work kind and planning defaults · the current date
Not sent Your notes · your profile, accommodations, or history · your course or instructor identity · your existing task text
The model is Claude, made by Anthropic, reached through Cloudflare’s AI gateway. The returned steps stay on the page until you review them.
What that means in practice
A model sees the words of an assignment. It does not see who you are, what you struggle with, what accommodations you have, what your grades are, or anything you have written privately.
We do not train models on your work.
What is coming, and how it will arrive
We want AI to help with more than one thing. Drafting the awkward email to an instructor. Rehearsing a hard conversation before you have it. Turning a photographed syllabus into real dates.
Each of those will arrive the same way this one did: as its own feature, with its own screen listing exactly what it sends, asked every time you use it. Nothing gets added to the list of what leaves your account quietly, and no capability starts sending more than it says.
If that ever stops being true, it is a bug and we would like to hear about it: [email protected].