Overview
Every question you send creates a doubt session. The first response is generated by the AI solver and grounded in your subject and grade. You can keep chatting in the same session to ask follow-up questions; messages are saved so you can revisit the explanation later.
Doubts use GPT-4o-mini for text and GPT-4o for image OCR. The solver pulls relevant context from approved tutor notes (RAG) where available, so explanations match the way concepts are taught on Tuition.in rather than the open internet.
How to ask a doubt
- Open Dashboard → Doubts and click New doubt.
- Type your question, or paste it from a screenshot. Optionally set the
subjectandgradeso the solver narrows its context. - Click Solve. Within a few seconds you get a step-by-step explanation. The model used and approximate token cost are logged for transparency.
Snapping a photo (OCR)
Click Upload image instead of typing. The image uploads to your tuition.in storage and the solver runs OCR on it before answering.
- Supports printed textbook questions, handwriting, math expressions, and diagrams.
- Best results: keep the question well-lit and within the frame. The extracted text is shown back to you so you can correct it before the AI answers.
- You can leave the text field empty when uploading — the OCR result becomes the question.
Follow-up chat
Each doubt session is a thread. After the first answer you can ask clarifying questions like "why is step 3 like that?" or "show me a similar problem". The solver has the whole conversation in context.
Sessions remain open indefinitely. You can come back the next day and continue the conversation.
Doubt history
The Doubts page lists every session you've started, newest first, with the subject, grade, question preview, and current message count. Click any row to reopen it.
Sessions count toward your daily XP — see Streak & XP.
Rating an answer
At the bottom of each answer is a thumbs-up / thumbs-down. Your feedback trains future answers and lets us spot patterns where the solver is wrong. Unhelpful or incorrect answers are reviewed weekly and used to update the RAG context.
Languages supported
Your preferred language is passed to the solver. English, Hindi, Telugu, Tamil, Kannada, Bengali, and Marathi are supported as the answer language. The OCR works on English text regardless of UI language.