🎯
Retrieval-Augmented Generation (RAG) Core
When you ask a question, our local sentence-transformers embed your query into high-dimensional vector space, query ChromaDB for top-matching textbook chunks, and feed only verified source excerpts to the language model.
1. Student Query Vectorization
all-MiniLM-L6-v2 · 384 Dim
2. ChromaDB Semantic Search
Top-3 Chunks · 98.4% Confidence
3. Grounded LLM Synthesis
Ollama · Zero Hallucination Mode
🎓
Class 1–12 Adaptation
Automatically scales explanation complexity. Explains gravity using apples for Class 5, and orbital mechanics and vectors for Class 11.
🔒
Isolated MongoDB Storage
Every conversation is securely saved in a dedicated user-specific MongoDB document. Your chat history remains 100% private to your account.
⚡
100% Private Local AI
Runs locally via Ollama. No expensive cloud APIs, no data mining, and no external tracking of student questions.
📖
Verifiable Page & Chapter Citations
Every single answer includes verified book titles and page numbers. Students can flip to the exact page in their physical textbook to verify and continue studying.
📚 GSEB Science Textbook
📘 NCERT Mathematics
📙 Social Science History & Civics