MessMitra — AI Hostel Mess Feedback Analyzer

Inspiration

As a 2nd-year B.Tech AI/ML student living in a hostel, I have seen how food and mess complaints often get lost in WhatsApp groups or verbal discussions. Students may report issues such as cold food, watery dal, undercooked rotis, or hygiene problems, but there is usually no simple way to see which issues are repeated, which are urgent, or whether the mess committee’s response actually worked. I wanted to build a small but practical AI tool that turns scattered complaints into clear priorities and measurable action.

What it does

MessMitra lets hostel students submit quick, anonymous mess feedback in English or Hinglish. A student selects the meal, gives a rating, and writes feedback such as “Dinner me daal me insect mila!” or “Roti kacchi thi aur cold thi.”

The app instantly analyzes the feedback, identifies the issue category, sentiment, urgency, and specific problem. It then groups similar complaints into clear issue cards for the mess committee.

For committees, MessMitra provides a dashboard with urgent safety alerts, clustered issues, complaint counts, meal filters, and AI-suggested corrective actions. Its most important feature is the closed feedback loop: after an action is marked as implemented, the app compares later feedback and shows whether complaints for that issue reduced.

How we built it

I built MessMitra as a mobile-friendly web app using Next.js, React, TypeScript, Tailwind CSS, and Lucide React. Feedback and demo data are stored in the browser using localStorage, so the app works without a backend database, login system, or paid cloud service.

The AI analysis uses a built-in offline NLP pattern-matching engine. It understands common English and Hinglish hostel complaints, classifies them into categories such as hygiene, food quality, quantity, temperature, service, and menu variety, and assigns urgency levels. Critical issues, such as insects or hygiene risks, are highlighted immediately.

I followed a spec-first workflow using scope.md, prd.md, spec.md, and a verified build checklist. The app was built in small, testable slices, with each completed slice committed to Git.

Challenges we ran into

One challenge was handling Next.js server-side rendering with localStorage, since browser storage is only available on the client. I solved this by loading saved feedback after the app mounts on the client.

Another challenge was building a reliable offline analyzer for informal Hinglish feedback. Instead of depending on an external AI API, I created a keyword and pattern-based analyzer that can understand common campus expressions while keeping the demo fast, free, and reliable.

I also had to keep the project focused. It was tempting to add login, photo uploads, real-time chat, and multi-hostel support, but I cut those features to deliver one complete proof of concept.

Accomplishments that we're proud of

I am proud that MessMitra is a complete, working end-to-end product rather than just a technical demo. A student can submit feedback, the app can classify and group it, a committee can act on it, and the app can show whether the action reduced future complaints.

I am especially proud of the closed-loop impact tracker. It does not only say “there are complaints”; it shows whether a corrective action actually improved the situation. I am also proud that the core AI analysis works fully offline, making the demo dependable and free to run.

What we learned

I learned how to turn an idea into a structured plan before writing code. Using scope.md, prd.md, and spec.md helped me define the user, core loop, features, boundaries, and success criteria clearly.

I also learned to build in thin, verifiable slices instead of asking an AI agent to build everything at once. Testing after each slice made it easier to catch issues early and keep the project stable.

Finally, I learned that a good AI product does not need to be complicated. A focused workflow—feedback, insight, action, and measurable improvement—can create real value.

What's next for MessMitra — AI Hostel Mess Feedback Analyzer

Next, I would add: Multi-hostel and multi-mess support Student upvoting on existing complaints WhatsApp or email notifications for critical hygiene issues Photo upload for food-quality and hygiene evidence Admin roles for mess committee members Weekly analytics comparing multiple hostels Optional LLM-powered deeper analysis while keeping the offline analyzer as the reliable fallback The long-term goal is to make MessMitra a practical accountability tool for hostel messes: helping students be heard, helping committees act faster, and proving that improvements actually work.

Built With

• lucide-react
• next.js
• react
• tailwind-css
• typescript