Course Correct is a space themed University scheduling assistant whose purpose is to help students create their class schedule using the most up-to-date class information, knowledge of their previously taken courses, and the requirements listed for their specific major, all powered by a Google Gemini-based AI agent. Users start by uploading a transcript, and the agent does the rest!
In addition to performing basic scheduling tasks, Course Correct maintains a conversation with the user and makes course suggestions based on their preferences, including gathering RateMyProfessor reviews to steer students toward professors more tailored to their learning style.
We used Gemini Flash 2.5 to power the AI agent's responses. This gave us the ability to have an agent with quite a few messages, allowed function calls, and lots of data in the context, while still being able to produce sensible and accurate responses to the user's queries.
ElevenLabs was use for speech-to-text and text-to-speech capabilities, allowing the user to maintain a conversation with the agent. We chose the Matilda voice, which we felt filled the role of an academic advisor quite well.
Flask was used for our backend api, allowing our AI agent to access tools written in Python for course scheduling, RateMyProfessor reviews, major requirements, and more.
Our frontend was built on React with TypeScript, providing a beautiful interface to interract with the Course Correct agent, generate schedules, and view chat logs in real time.