Built for Real Students, Guided by Real Textbooks
We believe AI should empower students to master their school curriculum with verified facts and exact page references, rather than making up answers from the internet.
The Problem with Generic AI
Standard language models hallucinate facts, use college-level vocabulary with primary school students, and cannot cite official state board textbooks or page numbers.
The Study Buddy Solution
We built a specialized Retrieval-Augmented Generation (RAG) engine that indexes actual school syllabus textbooks into vector embeddings and restricts AI answers strictly to that content.
Privacy by Default
Student data should never be monetized or used for model training. Every chat session is stored in an encrypted, isolated MongoDB document tied only to that student.
World-Class Local Tech Stack
Crafted with modern open-source technologies for blazing speed, high precision, and absolute data sovereignty.
Ollama Local LLM
Executes state-of-the-art open-weights models locally on your machine with low latency and complete offline independence.
ChromaDB & MiniLM
High-performance vector database storing 384-dimensional dense semantic embeddings generated by sentence-transformers.
MongoDB Vault
NoSQL database architecture with separate users and chats collections linked via authenticated ObjectId references.