Canvas LMS AI Assistant

canvas login
courses page
collaboration feature
Course resources
Memory management

Inspiration

Every single minute of this presentation, over 200 U.S. college students experience significant stress and anxiety directly tied to their coursework. That’s not just a number - it’s a symptom of a system that isn’t built for them. The traditional education model is showing its cracks: •⁠ ⁠Most U.S. college students report major stress and anxiety around academics (National Education Association, 2024). •⁠ ⁠A large share of learning time is wasted on inefficient studying and passive content review (Global Learning Council, 2025). •⁠ ⁠Many students feel their learning materials don’t fit their individual learning styles, causing disengagement (U.S. Dept. of Education, 2024). •⁠ ⁠These numbers tell a story: students aren’t failing - the system is failing them. We were inspired to create a solution that reimagines how students learn and interact with educational content, using technology to make studying personalized, efficient, and mentally sustainable.

What it does

Our platform transforms the way students learn by: •⁠ ⁠Unifying all learning materials - from PDFs, DOCs, lectures, and even YouTube videos - into one smart, searchable hub. •⁠ ⁠Providing an adaptive AI tutor that interacts with you in real time, answers questions, and adjusts to your learning pace and style. •⁠ ⁠Recommending personalized resources like articles and videos to bridge your knowledge gaps and reinforce your understanding.

How we built it

We built our project using a powerful combination of cutting-edge technologies: Canvas API Token – to securely fetch course materials and student data from learning platforms. LangChain – to orchestrate the flow between different data sources, vector databases, and the LLM. OpenAI LLM – as the core reasoning engine for tutoring, summarization, and adaptive learning conversations. MemMachine – to manage memory across chat sessions and enable contextual, long-term understanding. DuckDuckGo Search Library – to fetch real-time and relevant web content, enhancing recommendations.

Challenges we ran into

MemMachine Setup & Integration: Syncing it with our LLM pipeline was complex - we had to optimize how memory was stored and retrieved across sessions.

Accomplishments that we're proud of

Seamless MemMachine Integration: Successfully implemented persistent memory to retain learning context across sessions, enabling more natural and personalized tutoring interactions. Automated Notes Generation: Developed a pipeline that produces well-formatted, concise study notes from diverse content sources including lectures, PDFs, and documents. Intelligent Q&A Chatbot: Integrated a Retrieval-Augmented Generation (RAG) chain powered chatbot that answers questions using a knowledge base built from professor-provided materials. Guided Learning Mode: Designed a personalized learning system that adapts to each student’s preferred learning style - whether they learn best through videos, research articles, or written explanations - ensuring a more engaging and effective study experience.

What we learned

MemMachine Setup & Integration: Setting up MemMachine and syncing it with our LLM pipeline was initially challenging. We had to fine-tune how memory was stored and retrieved across sessions to maintain context without overloading the model.

What's next for Canvas LMS AI Assistant

Enhanced Data Privacy: Give users full control over what gets stored through private sessions and customizable data retention settings. Private Search Mode: Enable secure, anonymous searches that protect user identity and prevent data tracking. Smarter Personalization: Improve guided learning by refining content recommendations based on user feedback and learning behavior.