For university
An AI tutor for university students: what changes when the work is course-specific and rubric-graded
University work differs from school work in three ways that matter for an AI tutor: it is specific to one professor's course and notation, it is graded against a rubric the student can usually read, and it sits above the level where general curriculum libraries thin out. An AI tutor for university students therefore needs to work from the student's own problem set, paper brief and rubric, hold a real conversation about a proof or a derivation, and be honest about the line between feedback and authorship. CANtutor AI is built around those three.
By CANtutor AI team, EditorialPublished Updated
Why do school-level AI tools thin out at university?
Because their material is a curriculum, and a university course is not one. A second-year real analysis course uses the professor's definitions and the professor's order; a fourth-year seminar is graded on an argument nobody has made before. A tool built on a fixed library of lessons has nothing to say about either, and a general chat assistant will discuss the topic without ever having seen the course. The gap is not intelligence. It is that the tutor has not read the thing you will be marked on.
The second reason is tone. A product written for a fifteen-year-old will make a third-year engineering student feel it is not for them, and they will be right. The bar for a university tutor is that it reads the rubric, follows the notation, and does not talk down.
What does course-specific mean in practice?
It means the tutor works from your documents. In CANtutor AI each assignment has its own workspace holding the brief, the attachments and the rubric, and a chat scoped to that assignment that has already read them, so you never paste the question in first. In a live session, share the screen with the problem set open or hold the printed sheet up to the camera, and the tutor reads it itself. The whiteboard is where a derivation, a proof sketch or a free-body diagram gets drawn while the tutor talks, which is the part that a text explanation cannot replace.
For a paper, upload the brief and the rubric, and ask what each criterion is actually asking for. Rubrics are frequently misleading about what earns the marks, and the criterion that costs the most is usually the one written in the vaguest language. Then write the draft yourself, and run it through the AI Grader against that rubric: it returns the specific things costing marks, and Revise with AI shows proposed changes as a diff you accept or reject, so the paper stays yours.
How does it handle a proof or a derivation?
Out loud, in steps, with you doing the next one. Learn mode walks through the idea and asks questions rather than presenting a finished proof. Practice mode waits for your attempt, tells you exactly where the argument breaks, and gives the smallest hint that gets you moving; ask outright and it shows the full solution. That is the right shape for a problem set, because the exam will ask for the following problem, not this one.
Answer mode exists and is labelled, with a warning at the point of use that turning in AI-written work as your own may breach your institution's academic-integrity policy. The grader ignores the mode. For a university student the practical reading is: use Answer mode to see the shape of a method you have never met, then close it and do the set.
What about the rest of the term?
The session is the middle of studying, and the surrounding tools cover the rest. Lectures records or uploads a lecture, transcribes it, and turns the transcript into flashcards and a practice quiz. The exam study plan builds a dated, day-by-day revision plan from your own coursework and attachments, with written-answer practice built in. The grade calculator works out what you need on the final from your weighted categories, and the planner keeps the deadlines in one view.
Where your institution uses Google Classroom, it can connect. Google Classroom connects for personal Google accounts and for school accounts where the district permits third-party apps. Some districts block it; if yours does, everything else still works and you upload your material instead. Most universities use a different LMS, in which case you upload the brief and the rubric to the assignment workspace and everything works from that.
Which subjects, and what does it cost?
There is a /tutoring page per subject, from linear algebra, differential equations and organic chemistry through data structures, research papers and academic writing, each naming where students actually get stuck. Sessions and generated material spend credits; the plans and trial terms are on /plans, and the credits meter shows a per-feature breakdown, so a term's usage is visible rather than a surprise.
Questions people ask about this
Does it work for graduate-level material?
It works from whatever you bring, and the level follows the material. What it cannot do is know a field's unpublished conventions; give it the professor's notes and it will use them.
Will it write my paper?
Answer mode can produce text, and it says so with a warning. The product is built around the other direction: you write, it marks against the rubric, and you accept the fixes you agree with.
Can I use it for a lab report?
Yes. Upload the lab handout and the marking scheme, work the analysis in a session with the data on screen, and run the write-up through the grader before submitting.