CoursePilot was inspired by a simple problem: students often know their goals and constraints, but turning them into a workable schedule is still tedious, fragmented, and confusing. We built a conversation-first planning system that lets users describe preferences in natural language, translates them into structured scheduling constraints, and then generates multiple comparable plans on top of a deterministic scheduler. Along the way, we learned that the hardest part was not just generating schedules, but designing a workflow that keeps the AI understandable, editable, and grounded in explicit constraints. Our main challenge was bridging open-ended dialogue with reliable planning logic, so we introduced a visible restrictions layer and a plan workspace to make the system both flexible and controllable.
Built With
- fastapi
- python
- react
- tailwind
- tailwindcss
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.