Why We Built an "Ask a Question" Feature

We just shipped a feature that lets students ask follow-up questions when they get stuck on an MCQ.
Pretty simple idea. Kid gets a question wrong, reads the explanation, still confused — now they can just ask.
Why bother?
Right now, the options aren't great:
- Take a photo and ask tutor tomorrow
- Post on some Facebook parent group
- ChatGPT (which half the time gives answers that are too complex, or just wrong for PSLE)
By the time anyone gets back to them, they've moved on. The confusion just... stays there.
What we did differently
We didn't just slap a chatbot on it. Every conversation is grounded in:
- The PSLE Science syllabus
- The actual question they're looking at
- The explanation for that question
- What answer they picked
So the AI already knows what they got wrong and why. It's not starting from scratch.
This keeps things on-topic — no drifting into sec school content — and way more relevant than a generic "explain photosynthesis" answer.
What students can actually ask
The obvious one is "why is this the answer?"
But the more interesting stuff is what's hard to ask a real tutor:
- Explain simpler
- Give me a different analogy
- "Is this like when..." (connecting to something they've seen)
- More examples
- Real world examples
- "But why is my answer wrong?"
You're not gonna ask your tutor to explain it five different ways. You're not gonna ask in front of 30 classmates. With AI, there's no friction.
It's not just about getting unstuck. It's a space to poke around and be curious without feeling stupid.
The patience thing
AI doesn't sigh. Doesn't say "I already told you this."
Kids can ask the same thing over and over until it clicks. They can admit they don't get it. No judgment.
Sounds minor but we think it's a big deal. A lot of kids stop asking questions not because they understand — but because they're embarrassed. This removes that.
Hopefully the habit carries over to asking questions in class too.
What we're watching
This gives us data we've never had before:
Which MCQs keep triggering questions — if one question gets way more follow-ups than others, our explanation probably isn't good enough. Or the question is badly written. Either way, we'll know.
What kids are actually confused about — probably different from what we'd guess. If we see the same misconceptions coming up, we can fix them upstream.
What kinds of questions they ask — are they just asking "what's the answer" or are they asking "why does this happen"? Are they making connections? This tells us if we're helping them think or just helping them pass.
Still figuring out
- How do we nudge kids toward asking better questions? Not just "what" but "why" and "what if"
- Can we use the patterns to improve our explanations before kids even need to ask?
- Should parents or teachers see what their child is asking?
We're early. But the bet is simple: kids have way more questions than they let on. They just need somewhere to ask.