Every demo looks impressive. That is what demos are for. This is a list of questions that separate a well-built product from a well-marketed one, written so you can take it into a sales conversation.
First: what problem are you actually solving?
Before any of the twelve, answer this one honestly, because the right product depends entirely on it.
- The student understands the work but has not done enough of it. Practice volume is the need. AI is genuinely strong here and you may not need much else.
- The student has a misconception nobody has found. Diagnosis is the need. This is the hardest thing for software and the clearest case for a person.
- The student has stopped trying. Motivation is the need. No current AI product solves this, whatever the marketing implies.
- The student is fine and you want an edge. Enrichment. Almost anything decent will do; do not overpay.
A product that is excellent for the first case can be useless for the third. Most disappointment comes from buying for the wrong one.
Questions 1 to 4: what it actually does
1. Who starts the session, my child or the system? If your child must always arrive with a question, it is a study assistant rather than a tutor. That may be fine, but it should be priced like one.
2. What does it remember, and how does that change tomorrow? "It saves chat history" is not an answer. Ask what happens differently tomorrow because your child missed three questions today.
3. When my child is stuck, does it explain or does it answer? Ask to see this live. Have them enter a problem and refuse to engage. A good system asks a narrower question. A weak one supplies the solution.
4. Can two students end up on different topics, or only at different speeds? This separates real path personalization from pacing. Both are legitimate; only one is what the phrase implies. See what personalized learning actually means.
Questions 5 to 8: evidence, safety and data
5. What evidence do you have that it works, and on whom? Testimonials are not evidence. Ask whether learning was measured after the tool was removed. Ask what was measured, on how many students, compared to what.
6. Does it have guardrails on giving answers, and what are they? This is question one in disguise, and it is the single most important thing on the list. In the PNAS study, unrestricted GPT-4 left students around 17% worse off than never having used it; a guardrailed version did not. Guardrails are the difference between the two results.
7. Where do the questions come from, and how do you know the answers are right? Freshly generated questions can be wrong, and a confident wrong explanation is worse than no explanation to a student who cannot catch it. A vetted bank is a meaningfully better answer than "the model produces them".
8. What data do you collect about my child, where is it stored, and who can see it? You are handing over a minor's academic performance over time. Ask for a straight answer, and ask what happens to it if you cancel.
Questions 9 to 12: fit, cost and exit
9. Which syllabus is it built for? A product aligned to a different curriculum will teach correct material in the wrong order and to the wrong depth. For Indian families this is the CBSE, ICSE or state board question, and "it covers everything" usually means it is aligned to nothing.
10. What does it cost in total, per year, including anything that is extra? Compare against the real alternative, which is usually a number of hours of human teaching, not zero.
11. What happens when my child is stuck in a way the system cannot handle? Is there a human? How quickly? Is that included?
12. How do I cancel, and what do I keep? Ask before you sign, not after.
The three most products quietly fail
From the twelve, these are the ones where answers tend to become vague:
- Question 3 (explain or answer). Many products are optimised to feel helpful, and giving the answer feels helpful.
- Question 5 (evidence after removal). Very few have measured this. Fewer publish it.
- Question 6 (guardrails). This is where the research points and where marketing is quietest, because a tool that withholds is a harder sell than one that solves.
If a company answers those three well, it is probably serious.
A note on what none of them can do
No product on this list will make a disengaged child want to study. Software can remove friction, supply practice and explain patiently. It cannot supply the reason. If that is your actual problem, a person who expects your child on Thursday is more likely to help than any subscription, which is the honest conclusion of AI tutor vs human tutor.
Our own answers
It would be poor form to publish a checklist and duck it, so here is eGunas against all twelve, as of today.
- Who starts? A human tutor does. We do not have an AI product live.
- What does it remember? The tutor does, and reports to you. No automated model yet.
- Explain or answer? Our tutors are taught to ask rather than tell. This is a training standard, not a technical guarantee.
- Different topics or different speeds? Different topics. That is what one-to-one teaching is.
- Evidence? We have no published trial of our own. We would rather say so than cite someone else's as if it were ours.
- Guardrails? Not applicable yet. When our AI launches, this page is the standard we expect to be held to.
- Question sources? Board-aligned material selected by the tutor.
- Data? Contact details and lesson records. Ask us and we will tell you exactly.
- Syllabus? CBSE, ICSE and state boards.
- Cost? Per-hour tuition, no published AI subscription, because there is nothing to subscribe to.
- When stuck? A person is already the answer.
- Cancel? Stop booking. Nothing is locked in.
Two of those are "not yet". We would rather you had a list that we currently fail in places than a list written to make us win.
A checklist assumes the decision to buy is already made. If it is not, work backwards first: whether your child needs a tutor at all, and then how many hours would actually be useful.
How eGunas helps your child
eGunas teaches one child at a time. A session follows the student's own board syllabus but is built around what that particular student needs next -- their pace, the topics they are actually weak at -- rather than one lesson delivered to thirty students at once. That is what personalised learning means in practice.
eGunas AI, our AI-powered learning platform, is planned for 2026 and not available yet. It is being designed to extend that same individual attention beyond the lesson hour: practice pitched at the right difficulty, available whenever your child sits down to work, so progress can continue between sessions instead of waiting for the next one.
- Browse our courses -- CBSE, ICSE and state boards, Classes 1 to 10
- How eGunas online tutoring works -- one-to-one lessons online, wherever in India you live
- Book a free demo class -- one-to-one with a tutor, at no cost
- Ask us about your child -- tell us the subject and year, and we will say honestly whether we can help
New articles are published on the eGunas blog.
