Inspiration

As students, we constantly juggle deadlines, quizzes, midterms, and grading requirements across multiple Canvas courses, especially when planning and applying for events on other third-party platforms. Critical information is often buried deep inside PDFs, announcements, or scattered across different tabs. Finding something as simple as “When is my midterm?” or “How many assignments are left?” can take several minutes, and missing one detail can mean missing a deadline, or mis-scheduling of important events.

We built AskMyCanvas to remove this friction. Our goal was to let students ask questions while they are on any platform, and get instant, reliable answers without digging through files or guessing where instructors posted something.

What it does

AskMyCanvas is an AI-enabled browser extension chatbot for Canvas that allows students to ask questions about their courses in plain English.

Students can ask:

  • “How many assignments do I have due this week?”
  • “When is my midterm for CS2040S?”
  • “What do I need to score on the final to get an A?”
  • “Which assignments are still unsubmitted?”

The extension intelligently reads relevant Canvas data (assignments, grades, schedules, pages) and surfaces precise answers instantly - helping students stay organized, productive, and stress-free.

How we built it

  • The extension accepts prompts from the user
  • The prompt is send to the backend server we have built, which then sends it to Gemini
  • Gemini based on the "tools" that we have provided, will choose which tools to execute
  • Tools are essentially functions with extensive descriptions fetching data from Canvas
  • This has been optimised by making sure that Gemini runs the fewest tools possible for fastest data retrieval
  • If Gemini is not able to find the answer using the first tool given, it will run upto 4 tools, ensuring it doesn't give hallucinated answers
  • The AI interprets the student’s query, performs calculations if needed (e.g. grade projections), and returns a clear, contextual answer.
  • All interactions happen seamlessly within whatever app you have open(think Gmail, Calendar, Other Apps), without disrupting the student’s workflow.

We designed the system to be fast, privacy-aware, and intuitive, focusing on real student use cases rather than generic chatbot responses.

Challenges we ran into

The first challenge was getting Canvas API and integrating into the backend. Afterwhich, we struggled with LLM tooling and how to call appropriate functions to Gemini. Also, ensuring the chatbot gives accurate, non-hallucinated answers, especially for grades and deadlines was tricky, but we too smart!!

Accomplishments that we're proud of

We hope to pull the average NUS GPA up by atleast 0.2 points.

Share this project:

Updates