Campus Bot

AI That Understands Your Campus

Comment

Inspiration

Every college student faces the same frustration — endless searching for academic information, rules, circulars, and contacts, and waiting days for complaint resolutions. We wanted to fix this by creating one intelligent platform where students can instantly ask anything about the campus and get verified answers, and also raise complaints directly to department heads. The idea was simple: “What if our campus had its own ChatGPT — trained only on official documents?”

What it does

We built Campus Desk, an AI-powered platform combining:

AI Chatbot with RAG

A custom chatbot built using Gemini 2.5 and a Retrieval-Augmented Generation system. It answers student queries only using:

Official PDFs

Website text

Uploaded circulars

Department docs

Complaint Management System

Students can submit complaints, attach images, track status, and view admin replies. Admins can update status, send responses, and manage complaints for their department.

Admin Console

Admins can upload documents (PDF/text), which automatically get indexed, chunked, and embedded, becoming part of the knowledge base used by AI.

Smooth UI/UX

A dual-dashboard interface (student + admin) with classic dark UI, clean cards, and neatly separated sections.

How we built it

Flask for backend APIs (auth, chatbot, complaints, file uploads).

Supabase PostgreSQL for scalable database storage.

Gemini API for embeddings + intelligent responses.

RAG Pipeline for accurate campus-specific answers.

HTML, CSS, JS for the frontend dashboards.

Designed a custom embedding + indexing workflow to make the chatbot reliable and document-aware.

Challenges we ran into

Embedding and indexing errors when Gemini keys failed or rate limits hit.

Database migrations while updating document schema for RAG.

Frontend state management for dynamic dashboards.

Ensuring the chatbot answers only from college documents, not hallucinating.

Handling PDF extraction errors and large files.

Making complaint routing work correctly for each department admin.

But solving these challenges is exactly what made the project solid and hackathon-ready.

Accomplishments that we're proud of

We built a smart, unified campus assistant that reduces manual workload, improves communication, and gives students instant access to verified campus information.

Campus Desk → “Ask. Upload. Resolve.”

What we learned

During this project, we learned how powerful Retrieval-Augmented Generation (RAG) can be when paired with real institutional data. We explored how embeddings work, how documents are chunked and indexed, and how LLMs retrieve knowledge with high accuracy. We also gained hands-on experience with Flask routing, Supabase database design, and building a multi-role authentication system. On the frontend side, we learned to build clean dashboards, maintain state across pages, and handle API communication efficiently. Most importantly, we learned how to transform raw college information—circulars, rules, departments, complaints—into a unified and intelligent system that actually solves a real campus communication problem.

What's next for Campus Bot

Our next step is to make Campus Desk even smarter and more integrated. We plan to expand the knowledge base by automatically scraping the latest notices from official college websites, add multilingual support for students, and introduce voice-based queries. We also want to include analytics dashboards for administrators, improve complaint categorization using AI, and implement push notifications for new circulars or resolved complaints. Ultimately, we want Campus Desk to evolve into a full digital campus companion that streamlines communication, reduces confusion, and makes official information accessible to every student—instantly.

Share this project:

Updates