Blog

How Does One Photo of a Problem Become an Explanation in Minutes?

Up to 9 photos at a time. We transcribe, answer, then explain, checking against the textbook with page numbers. Usually 3-6 minutes. It can make mistakes.

The Duowenwen team · Updated

Some parents have asked what actually happens when you photograph a problem, and why they should believe it can explain it. This article lays out the process, and covers only what we really do, not what we can't.

What happens after you take the photo?

You can photograph a worksheet or an exam paper, up to 9 photos at a time. First the photos are transcribed into text, so every problem is read correctly, including any labels on diagrams. If the problem is misread, a brilliant explanation afterwards is worthless. Only once transcription is done does it answer, and only after answering does it explain. The three steps are separate, so when your child asks a follow-up, it works from the text and doesn't have to read the image again. A slightly crooked or dim photo is usually fine, but if a thumb covers part of the problem it can't read it. In that case it will tell you, "I couldn't see this one clearly", rather than guess.

Why does the answer have to match the textbook?

Because teachers go by the textbook, and the textbook is the only thing your child can look up. When it answers, we have it align with the textbook text word for word and give a page number, so your child can open the book and check. Where the explanation differs from the book, go with the book. The page number isn't there to make it look reliable. It's there to make it easy to spot when it's wrong.

What tone does it use?

A tutor's tone: it goes step by step and doesn't skip any. It never says "you should know this". It praises the thinking, not the cleverness, and it never compares a child with anyone else. At the end of each explanation it offers 2–3 "keep asking" prompts, so your child chooses what to ask next. It doesn't hurry them and it doesn't give out scores.

How long does it take?

Usually 3–6 minutes from photo to explanation. That covers all three stages, transcribing, answering and explaining, and it runs slower when there are many problems or diagrams. It isn't an instant-answer machine, because finding the original text and page number in the textbook takes time. We think those few minutes are worth it.

Do follow-up questions need a new photo?

No. Follow-ups carry on from the original problem, which is why we keep the transcribed text for 7 days; after that it's deleted automatically. Photos are kept for up to 90 days, only the operators can see them, and they're used to improve recognition. We don't use them to train our own models. Explanations do go through third-party AI services; see our privacy notice. Your child can ask "Why does this step work like this?" or "What if the conditions were different?", or can say "I don't think that's right", and it will explain again.

Can it make mistakes?

Yes. It can misread a problem, miscalculate, or give a page number that's slightly off. That's why we regularly check explanations by hand, looking at three things: the answer, the page number, and the tone. When we find a wrong explanation, we fix the rules. An explanation is there to help you understand, and the textbook and your teacher have the final say. We don't claim it never makes mistakes.

Next: photograph a problem from tonight's homework, check whether the page number is right, and see what it asks you at the end.

Common questions

How long does it take after I take a photo?
Usually 3–6 minutes from photo to explanation. Follow-up questions don't need a new photo; just carry on with the same problem.
Can it make mistakes?
Yes. It can misread a problem, miscalculate, or give the wrong page number. Page numbers are there so a person can check; the textbook and your teacher have the final say.

Try one question