The real argument for it is embarrassment, not efficiency
The strongest case has nothing to do with saving teacher time. It's that the student who most needs to ask a question a third time is exactly the one who won't — not in a class of forty, and not to a teacher who has already explained it twice and is visibly moving on.
That student currently gets nothing. An AI tutor answers the fourth attempt with the same patience as the first. For a child who is quietly behind and knows it, that is not a marginal improvement over the status quo — the status quo is silence.
The real argument against it is that it can teach the wrong thing
A general-purpose chatbot answering from the open internet will confidently contradict what was taught in class — different notation, a different method, a syllabus that isn't yours. A student who follows it loses marks and, worse, learns that the school's tool is unreliable.
This is the failure mode that makes teachers stop supporting the tool, and once teachers stop supporting it, it's over. Whether the tutor is grounded in the school's own material is the whole question, not a detail. Applying that to ourselves: SkoolBook's tutor is told which board, subject and grade it is helping with and to draw examples from that syllabus, but it does not retrieve from material you have uploaded. That is conditioning, not grounding, and the two get used interchangeably by people who are hoping you will not ask.
It should not be a separate app
An AI tutor that lives in its own application with its own login and its own account gets used for a fortnight. If it isn't in the place students already go for homework and marks, it competes for attention with everything else on a phone, and it loses.
Being inside the school portal also keeps it inside the school's data boundary rather than sending student questions into a consumer product with its own terms.
Ask how it's metered, and be suspicious of "unlimited"
AI has a real per-use cost that the vendor pays. "Unlimited AI" at a low per-student price is not generosity; it's a promise that gets quietly throttled, degraded to a cheaper model, or repriced once usage is real.
A vendor who tells you there's a fair-use cap is telling you the number they quoted is the number they can sustain — so ask them to show you where that cap is enforced, not just where it is written. Our own position, since it would be absurd to set this test and dodge it: SkoolBook does not have a per-student cap today, and the AI Tutor is enabled per deployment rather than running by default. If you are sizing this for a large cohort, that is a conversation to have with us up front.
What it doesn't do
It doesn't replace a teacher, and any vendor implying otherwise is selling to the wrong buyer. It doesn't fix a student who isn't attending. And it won't help at all if nobody tells students it exists — adoption is a communication problem, not a technology one.
Ask two questions: how grounded is it really, and how is it metered. Push past the marketing word on both — ask to see where material is retrieved from, and where a cap is actually enforced. Ungrounded answers destroy teacher trust; unmetered AI destroys the price you agreed.