Inspiration

Students often reread notes for hours without knowing whether they truly understand the material. We wanted to turn passive reading into active learning, helping students test their knowledge and learn from their mistakes.

What it does

StudyHive turns PDFs into an interactive study experience. Students receive plain-language explanations one section at a time, then take a five-question multiple-choice quiz generated from the document. They get a score and explanations for incorrect answers, with references to the relevant sections. Students can also retry with fresh questions.

How we built it

StudyHive is designed around three specialized AI agents:

Explainer: Converts document sections into clear, plain-language explanations.
Quizzer: Generates five multiple-choice questions with four options each, grounded in the PDF content.
Tutor: Explains incorrect answers using the relevant source section.

The workflow takes students from uploading a PDF to understanding, testing, and reinforcing their knowledge.

Challenges we ran into

Keeping AI-generated content grounded in the uploaded PDF is a central challenge. Quiz questions, correct answers, and explanations must reflect the source material rather than introduce unsupported information. Another challenge is ensuring quiz responses follow a consistent, valid JSON structure.

Accomplishments that we're proud of

We designed a focused learning workflow that brings explanation, assessment, and feedback together. StudyHive uses three specialized agents to support the learning process, with an emphasis on source-grounded answers and helping students understand why they made mistakes.

What we learned

Effective AI-assisted learning is about more than generating explanations. Students also need opportunities to test their understanding, receive meaningful feedback, and revisit the source material when they get something wrong. Specialized agents can give each stage of this process a clear responsibility.

What's next for StudyHive

We want to improve the quality and variety of generated questions, strengthen source references in feedback, and make retries more useful for reinforcing weak areas. Future iterations could also explore broader document support and more ways to help students track their understanding.