ClassSync — All your classes. One place.

ClassSync helps students save time, feel less overwhelmed, and get back on track when college gets busy. Between classes, work, and everyday life, it is easy to lose sight of a deadline or fall behind after a missed lecture. We built ClassSync to bring the semester into one place and help students answer a simple question: “What should I do next?”

Our goal is to give students more time for learning and less time spent hunting through course pages, piecing together deadlines, or wondering where they stand. Whether someone is planning their week, catching up, or working toward a target grade, ClassSync gives them a clearer starting point.

What it does

• Helps students see what needs attention with one view of assignments, deadlines, announcements, and to-dos across their classes.
• Offers an AI study companion for questions about coursework, notes, and upcoming work, alongside general explanations and study guidance. Guest chat works without creating an account.
• Keeps notes, course materials, and calendar dates together so students can spend less time searching and more time studying.
• Makes grades easier to understand with per-class calculators, category weights, what-if scenarios, and target-grade planning. Syllabus tools identify dates and grading information for students to review.
• Lets students build their workspace themselves or save setup time by importing D2L Brightspace coursework through a read-only browser extension. Students review each import before it is added.
• Keeps the working workspace in the student’s browser, with export/restore and optional Supabase cloud backup.

How we built it

The app uses Next.js, React, TypeScript, Tailwind CSS, and Zod, and is deployed on Vercel. The D2L integration is a Manifest V3 browser extension that reads information through the student’s own signed-in session. Imported records are validated and merged only after approval.

For AI, the live chatbot uses OpenAI’s open-weight GPT-oss-20b model hosted on DigitalOcean Serverless Inference. Our server calls DigitalOcean’s OpenAI-compatible endpoint; the model access key stays in a Vercel server-side secret. The model requests read-only tools, the browser looks up relevant items in the current workspace, and only those results are sent back for the answer. We verified both a live model response and a lookup of the student’s actual classes.

Challenges and accomplishments

The hardest parts were reconciling concurrent team changes, keeping repeated imports from duplicating coursework, extracting useful syllabus information, and connecting a deployed AI service while keeping credentials private and usage bounded. We added request limits and a DigitalOcean spending guard, and clearly label rule-based local answers separately from model-generated responses.

We are proud of shipping a working deployed app and a chatbot grounded in the workspace. Our latest merged version passed 107 automated tests and a production build. Grade calculations remain deterministic; the AI does not submit assignments or change course records.

What we learned and what’s next

We learned that useful AI needs reliable data, clear boundaries, and end-to-end testing as much as a good prompt. Next steps include broader LMS support, shared production-wide AI quotas, and more testing of syllabus layouts and school-specific configurations.

Try the live demo at https://classsync-two.vercel.app. New visitors start with an empty workspace and can add their own classes; D2L users can follow Connect your school to import their coursework. Guest AI chat is available immediately. DigitalOcean inference uses promotional credits and is usage-limited.