Login Screen
All Courses - Home page
Flashcards section
Course Details & Uploaded material
Quizzing/Practice
Report of current progress so far
Doubt asking Section
AI Voice assistant - on PPT material
Course Map - By topic
Topics & Sub-topics across the uploaded material
AI Voice assistant - on video material scolled
Course Map - By learning order
AI Voice assistant - on video material

Students often study from scattered PDFs, slides and lecture videos, while generic AI chatbots may give answers that are not grounded in their actual course material. We wanted to build a more reliable and personalized way to learn.

We built a Multimodal Personalized Course Companion that brings these resources together into a searchable knowledge base. It uses RAG with source-linked citations to answer doubts, verifies generated claims, organizes concepts into a topic tree and course map, and supports quizzes, flashcards and adaptive revision through a learner model. Our interactive Explain with Agent Video feature also helps students explore course content through narrated, source-linked explanations.

Building the system taught us how to combine multimodal processing, retrieval, LLM verification and learner modelling into one application. One of our biggest challenges was making ingestion and taxonomy preparation efficient while maintaining accurate source grounding and reliable question generation. We addressed these challenges by integrating background processing, hybrid retrieval, reranking and independent claim verification.

Our goal is to make learning more connected, transparent and personalized, helping students find trustworthy explanations and focus their revision on what they need to understand better.