Add-on

AI assistance for teaching and study

An optional add-on to the OLT SIS platform that drafts questions, study material and assignments for faculty, suggests marks for a teacher to check, and gives students a study tutor. It runs on Amazon Bedrock, is billed by use, and every feature is switched on by the institution itself.

Request a demo of AI assistance See capabilities

Who it serves

Built for the people who use it

Faculty setting questions, notes and assignments
Markers working through submissions
Students revising a subject on their programme
Deans of academics deciding what to switch on
Administrators managing the AI budget

The problems it addresses

What usually prompts the call

Building a question bank large enough for fair, varied quizzes takes more faculty time than a term allows.
Notes, problem sets and teaching cases are written from scratch for every offering.
A large batch of submissions needs marking and individual feedback by a deadline.
Institutions want to try AI tools without losing control of what they cost, who uses them, or who decides a mark.

Capabilities

What it does

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

Question drafting

Faculty ask for multiple-choice questions on a subject and topic at a chosen difficulty. Drafts are reviewed and edited, and only the ones the teacher ticks are saved to the quiz bank.

Study material

Notes on a topic, optionally following an outline the teacher gives. The teacher reads the draft and either saves it to the document locker as a PDF or discards it.

Assignment drafting

A problem set or a teaching case with questions, marks that add to the total asked for, and marking guidance that is kept for the marker and not printed on the students’ copy.

Marking suggestions

A second reading of a student’s submitted file against the brief and marking guidance, suggesting a mark and feedback that the teacher checks before anything is saved.

Study tutor

Students ask for a lesson on a subject of their own programme at one of four levels and ask follow-up questions, within a daily allowance the institution sets. More on the tutor.

Usage and spending

Every successful call is recorded with its charge. Administrators see spending today, this month and over 30 days, and can set daily and monthly limits for staff and for students.

Data

What is sent to the model, and what is not

Each feature sends only what it needs for that request. The lists below follow what the software actually sends.

Sent

Question drafting: the subject and topic names, the difficulty, the number of questions and any syllabus text the teacher types.
Study material: the subject and topic names, the intended audience and any outline the teacher types.
Assignment drafting: the subject and topic names, the number of questions, the total marks and any context the teacher types.
Marking suggestions: the subject name, the assignment title, brief, marking guidance and maximum marks, and the student’s submitted file itself, up to 4 MB.
Study tutor: the subject and topic names and the level; for a follow-up, the lesson text and the student’s question.

Not sent

Student names, roll numbers or other identifiers are not added to any request.
The student’s own file name is replaced before a submission is sent.
Marks, attendance, fees and other records held in the platform are not sent.
The tutor does not send the student’s identity, and the teacher’s course material is not sent to it.

A submitted file is sent as it is, so anything the student wrote in it, including their name, reaches the model. Students are told not to type private or sensitive information into the tutor.

Model requests go to Amazon Bedrock from the platform’s own AWS environment using its AWS role; there is no separate API key. By default they use an Asia-Pacific inference profile, which keeps processing within AWS’s Asia-Pacific regions, though not necessarily within India. Amazon states in its Bedrock documentation that prompts and responses are not used to train the models or shared with the model providers; that is Amazon’s policy for the service, not a separate guarantee from us. See security and data handling for hosting and data protection.

Controls

Who decides what runs

Switched on by the platform

The AI add-on is a module of the institution’s plan. Without it, no AI feature can be switched on, and the check is made on every call, not only when a page is drawn.

Each feature needs its own module

Question drafting needs quizzes. Study material, assignment drafting and marking suggestions need the document locker. The study tutor needs only the add-on.

The institution’s own switches

The institution’s administrator switches each of the five features on or off. A switch can narrow what the plan includes, never widen it. Assignment drafting, marking suggestions and the tutor start switched off.

Consent where required

An institution can ask students to agree before the tutor teaches them, or before their work is read for a marking suggestion. Without a current yes, the request is refused, and a withdrawal takes effect at once.

Limits before spending

Daily and monthly spending limits for staff and for students, and a per-student daily tutor allowance, are checked before each model call. When a limit is reached, those features stop until the period rolls over.

Records

Each successful call is recorded with its type, subject, model, token counts and charge, and tutor calls against the student. Changes to the AI settings are entered in the audit trail.

Example workflow

From switching on to first use

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

The AI add-on is included in the institution’s plan, with the quiz and document locker modules it needs.
The administrator chooses which features to switch on, sets the tutor’s daily allowance and any spending limits.
A teacher drafts questions, notes or an assignment, reviews the draft, and saves only what they accept.
When submissions arrive, the teacher asks for a reading of a submission, checks it against the work, and saves the mark they decide.
Students use the study tutor on the subjects of their own programme, within their daily allowance.
The administrator reviews usage and spending and adjusts the switches or limits.

Part of the OLT SIS platform

AI assistance is an add-on to the multi-tenant OLT SIS platform and works inside its quiz bank, document locker and student pages. It is separate from Eudemonic, the wellbeing analytics product. See the platform for how modules and plans work.

Study tutor for students

Lessons and follow-up questions on the subjects of a student’s own programme, at four levels.

AI tools for faculty

Drafting questions, notes and assignments, and marking suggestions, all reviewed by the teacher before use.

Security and data handling

How the platform is hosted and how institutional and student data is protected.

Questions

Frequently asked

Does the AI decide marks or grades?

No. A marking suggestion is shown only to the teacher, who reads the work, edits the mark and feedback, and saves the mark through the normal marks entry. Nothing reaches the marks grid or the student until the teacher saves it, and term locks and mark ranges still apply.

Is our data used to train the model?

The features call models through Amazon Bedrock. Amazon Web Services states in its Bedrock documentation that prompts and responses are not used to train the models and are not shared with the model providers. That is Amazon’s stated policy for the service, which institutions should read for themselves; it is not a separate guarantee from us.

How is it charged?

By use. Each successful call is metered on the model tokens it used and recorded with its charge, and the institution can set daily and monthly spending limits for staff and for students. The rate for your institution is set out in the proposal.

All questions

See AI assistance with your own subjects

Tell us which subjects and features you have in mind and the demonstration will use those.

Request a demo of AI assistance Build your requirements