An AI study companion for students
An AI-powered study companion that gives students context-aware help. It is built on a Retrieval-Augmented Generation architecture with open-source models running locally, so every answer is grounded in the course materials stored in its vector database.
Integrated with the indigo learning platform of the Technical University of Crete. The current beta release answers questions on heat transfer and runs on the group's own server.
What it does for students
Answers on the course material
Students ask questions in plain language and get answers drawn from the material of their course.
Step-by-step explanations
Difficult concepts are broken down one step at a time.
Guidance at each student's pace
Help adapts to the pace and needs of each student.
Direct access to course notes
Answers lead to the notes on the platform for focused study.
The assistant's start page
How an answer is produced
- 01
Ask
A student types a question in the chat.
- 02
Retrieve
The system finds the relevant passages of the course material in its vector database.
- 03
Generate
An open-source language model running locally writes an answer grounded in those passages.
- 04
Verify
Students are reminded to check important information against the course material.