AI add-on

AI study tutor for students

A student picks a subject from their own programme and a level, and the tutor writes them a lesson they can work through, keep and ask questions about. It runs on Amazon Bedrock, within a daily allowance the institution sets.

Request a demo of the study tutor See capabilities

Who it serves

Built for the people who use it

Students revising or catching up on a subject
Deans of academics deciding whether to offer it
Administrators setting allowances and budgets
Data protection officers deciding on consent

The problems it addresses

What usually prompts the call

Students want an explanation at their own pace, outside class hours.
A topic needs explaining again from first principles, or taken further, for different students.
Students are already using general AI tools, with no guidance from the institution and no limit on what they are told.
An institution wants to offer this without an open-ended bill.

Capabilities

What it does

Which of these are included in a deployment is set out in the proposal for that institution.

Subjects of the student’s own programme

The student chooses from the core subjects of their programme and the electives they have chosen, grouped by term. A subject outside their programme is refused, and the check is made on every request.

Four levels

Fundamentals, Core concepts, Application and Advanced. Each level assumes the one below it, and each lesson ends by pointing to the next level.

A topic, or the whole subject

The student can name a topic within the subject, or leave it blank for the material that matters most at that level.

Structured lessons

A lesson in ordered sections, with terms defined and worked examples, followed by check-your-understanding questions for the student to attempt and a suggested next step.

Follow-up questions

The student can ask about something in the lesson. The tutor is instructed to answer the question asked and to steer questions off the subject back to it.

Keep the lesson

A lesson can be downloaded as a PDF.

Safeguards

What the student is told, and what staff control

Before the first lesson

It is a study aid written by a computer, not their teacher, not issued course material and not a source to cite.
It can be wrong, and confidently so; anything that matters should be checked against their notes, textbook and teacher.
It does not know their syllabus or their exam; it teaches the subject in general.
Handing its words in as their own is plagiarism under the institution’s rules.
Sessions are paid for by the institution and recorded against their account, and they should not type anything private or sensitive.

The student ticks that they understand before starting, and each lesson is labelled as a machine-written study aid.

Staff controls

The tutor needs the AI add-on on the institution’s plan and starts switched off; the institution’s administrator switches it on.
A daily number of sessions per student, from 1 to 200, resetting at midnight in the institution’s own time zone. A lesson or a follow-up question each uses one session.
Optional daily and monthly spending limits for all student use of AI.
Optional consent: where the institution asks for it, only students who have agreed are taught, and a withdrawal takes effect at once.
Usage and spending reports for the institution.

Example workflow

A study session, start to finish

An illustration of a typical sequence. Your own process is configured during implementation.

The institution switches the tutor on and sets the daily allowance, and any consent question and spending limits.
The student opens the tutor, gives consent if asked, and reads and accepts the notice.
They choose a subject from their programme, an optional topic and a level.
The tutor writes the lesson, with questions for the student to attempt.
The student asks follow-up questions, or moves up a level.
They download the lesson as a PDF to keep.

Data and the model

A lesson request sends the subject and topic names and the level; a follow-up sends the lesson text and the student’s question. The student’s name and other records are not sent. If the model service is busy, the student is asked to try again and the session is not counted. The tutor is part of the AI add-on to the OLT SIS platform, where the data handling and controls are set out in full.

Academics

The student record, programmes, subjects and electives the tutor draws its list of subjects from.

Learning management

Course pages, materials and assignments: the taught material the tutor sits beside, not in place of.

Online quizzes

A question bank, quiz structures and schedules for assessing what students have learnt.

Questions

Frequently asked

Does the tutor teach from our own course material?

No. The tutor is given the subject name, the topic the student asks for and the level, and teaches the subject in general. It does not read the institution’s notes, syllabus or past papers, and students are told so before they start.

Can staff read what students asked?

Each lesson and follow-up question is recorded against the student as a usage entry, with the subject, the topic and level of a lesson, and the charge, and administrators see usage and spending for the institution. The text of the lesson and of the student’s question is not kept in that record.

What stops it costing more than we expect?

Three things: a daily number of sessions per student, set by the institution between 1 and 200; optional daily and monthly spending limits for all student use; and the switch that turns the tutor off. Each is checked before the model is called.

All questions

See the study tutor on your own subjects

Tell us which programmes you have in mind and the demonstration will use subjects from them.

Request a demo of the study tutor Build your requirements