CoursePilot

CoursePilot Logo
Automatically parse syllabus and create Slack canvas
Importing from Google Drive MCP
Automatically create new Slack channel based on threshold throuhg Slack RTS API
New slack channel created from hitting threshold
Automatically ping teaching assistants
Architecture Diagram
Quiz Making UI
Student view of quiz
Quiz Summary

Inspiration

Throughout college, Slack became the center of every course. Announcements, questions, study groups, and discussions all happened there. Yet despite being the hub of course communication, it remained little more than a messaging platform.

When we became teaching assistants, we realized we barely interacted with Slack beyond answering questions and posting announcements. It was where learning conversations happened, but it did very little to actively support learning itself.

That led us to ask:

What if Slack wasn't just where learning happened, but where learning was actively coordinated?

Could it recognize when students were struggling, organize peer learning, generate course resources, and reduce the administrative workload for instructors?

That question inspired CoursePilot.

What it does

CoursePilot is an AI-powered, Slack-native learning coordinator that transforms Slack into an active teaching assistant.

It helps instructors by:

Detecting students who may be struggling through conversation patterns
Alerting instructors with relevant context
Automatically forming study groups around shared learning gaps
Turning syllabus PDFs into searchable Slack Canvas resources
Generating low-stakes quizzes from lecture materials
Summarizing quiz results so instructors can quickly assess class understanding

By combining Slack Events, Search, Canvas, and AI, CoursePilot helps instructors spend less time managing courses and more time teaching.

How we built it

Frontend

Slack as the primary interface for students, instructors, and TAs
Slack messages, DMs, reactions, and channels for collaborative workflows
Slack Canvas for syllabus quick facts and course FAQs
Emoji-based quiz responses for frictionless participation

Backend

Python + Flask
Slack SDK and Slack Real Time Search API
Gemini and Ollama for syllabus extraction, quiz generation, and summarization
Google Drive OAuth + MCP integration for importing lecture materials
Lightweight JSON and in-memory storage for hackathon-friendly state management

Challenges we ran into

Separating fast Slack event handling from slower AI tasks like quiz generation and message analysis
Designing a seamless experience for both students and instructors with very different needs
Managing Slack permissions across messaging, reactions, channels, DMs, and Canvas APIs
Integrating Google Drive through OAuth and MCP while maintaining a reliable fallback path
Keeping AI instructor-safe by requiring review before student-facing content is published
Choosing lightweight storage for rapid prototyping while accepting tradeoffs in persistence

Accomplishments that we're proud of

Built an AI learning coordinator that works entirely inside Slack
Created an end-to-end workflow that detects confusion, forms study groups, generates quizzes, and summarizes learning outcomes
Designed a product that benefits both students and instructors without disrupting existing workflows
Integrated AI in a human-in-the-loop workflow instead of replacing instructor judgment
Connected Google Drive so lecture materials can immediately become learning resources
Built a modular architecture that is easy to extend beyond the hackathon

What we learned

Designing for Slack means designing conversations rather than traditional interfaces
Instructor and student experiences must be balanced for adoption
Slack's event model and permission system strongly influence application architecture
AI is most valuable when it assists instructors rather than replaces them
Integrating external resources like Google Drive requires as much attention to authentication and reliability as AI itself