-
-
dashboard
-
chat
Inspiration
As a student, I constantly struggled with the chaos of academic life — assignments scattered across platforms, files lost in Drive, deadlines sneaking up, and ten browser tabs open just to stay on top of one class. Study Agent was born from that frustration. I wanted to build something that actually understands a student's world: not just a chatbot, but a real companion that knows your pending tasks, your documents, your schedule, and helps you learn — all in one place. My goal is simple: reduce the stress my fellow students feel every day, and give them space to actually focus on learning.
What it does
Study Agent is an AI-powered academic assistant that connects to the tools students already use — Google Classroom, Drive, Gmail, YouTube, and Notion — and brings everything together in one intelligent interface. With Study Agent, students can:
Track assignments and deadlines pulled directly from Google Classroom Get AI-generated summaries and study flashcards from their own Drive documents Receive personalized video recommendations from YouTube relevant to their coursework Search semantically across multiple Drive files using Elasticsearch indexing to find exactly what they need Create Google Calendar reminders for upcoming tasks and exams automatically Chat with a student-focused AI (powered by Gemini) that explains concepts, reviews rubrics, and keeps them on track
The agent doesn't just answer questions — it reasons about the student's actual workload and context to give relevant, actionable help.
How we built it
Gemini — core reasoning engine powering the chat agent, summaries, flashcard generation, and recommendations Google Cloud Run — hosts the containerized backend (Docker), deployed via gcloud CLI Google Classroom API — fetches active courses, assignments, and announcements Google Drive API — reads and downloads documents (Docs, PDFs, Sheets) for indexing and analysis Gmail API — reads inbox for academic-related notifications Google Calendar API — creates study reminders and deadline events YouTube Data API — recommends videos related to the student's current topics Elasticsearch — indexes Drive file content for semantic search across the student's own documents Notion API — stores notes, study plans, and summaries Netlify + Cloud Run — frontend and backend deployment Python / FastAPI — backend API
Challenges we ran into
I faced several challenges throughout the project, from compilation errors to complex configuration issues. One of the most memorable was a subtle bug in production: Google OAuth credentials were silently failing in Cloud Run, even though they worked perfectly locally.
Accomplishments that we're proud of
Getting a fully functional, multi-service AI agent deployed and running in production is something I'm really proud of. Connecting six different APIs into a single, coherent agent that truly understands the student's context—and doesn't just respond to commands—was a real milestone. The Elasticsearch integration for semantic searching of personal Drive files is something I'm especially proud of: it transforms the student's scattered documents into a searchable knowledge base that they can explore conversationally. The fact that it was launched as a real, hosted product, rather than just a demo, is also significant to me.
What we learned
Building Study Agent taught me that the hardest part of a real-world agent isn't the AI — it's the connective tissue. Auth flows, API quirks, environment parity between local and cloud, handling multiple file formats gracefully — that's where the real complexity lives. I also learned how powerful it is to combine semantic search with an LLM that has actual context about the user's workload. And honestly, I learned that students have far more unmet needs than I initially realized. Every feature I built revealed three more that would make a difference.
What's next for study-agent
The vision is a full academic companion that grows with the student. Next steps include a mobile app for on-the-go access, support for more LMS platforms like Canvas and Moodle, a voice interface for hands-free study sessions, and collaborative study groups where students can share agent context and work together. Long term, I want Study Agent to proactively surface what a student needs before they even think to ask — reminders, recommendations, summaries — all timed to their actual academic rhythm. Since Study Agent is fully open source, I also hope this becomes a collaborative project. If you're a student, developer, or educator with ideas — pull requests are welcome. There are endless directions this could grow, and the best features will probably come from the students who use it every day.
Log in or sign up for Devpost to join the conversation.