-
-
This is not final how it looks like it need to be modified.
Inspiration
In schools, colleges, hospitals, and offices, canteens often face long queues during peak hours. Students and staff waste valuable break time waiting for food. Hospitals face delays for patients’ families and staff. This problem leads to stress, food wastage, and inefficient canteen operations.
We wanted to build a solution that makes dining smooth, fast, and smart.
What it does
DineFlow AI is an Agentic AI-powered cafeteria assistant that:
Lets users pre-order meals and choose time slots.
Uses AI to predict rush hours based on historical orders.
Provides cashless QR/UPI payments for faster service.
Gives canteen staff demand forecasts to reduce food wastage.
Scales across colleges, hospitals, schools, and offices.
How we built it
Backend: Python + Flask/FastAPI.
AI: IBM Granite models + ADK for natural language order processing.
Prediction: Time series forecasting for rush hour analysis.
Frontend: HTML/CSS/JS with Bootstrap (mobile-friendly).
Database: SQLite/MySQL for order management.
Integration: QR/UPI payment simulation.
Challenges we ran into
Designing a scalable system that works across different types of canteens.
Integrating AI natural language understanding (“I want a dosa at 1:15”) into structured orders.
Building a prototype in limited time with full workflows.
Accomplishments that we’re proud of
Built a working prototype that demonstrates AI-driven canteen ordering.
Showed how queues can be reduced by 50% with time-slot ordering.
Created a real-world India-centric solution that judges and users can directly relate to.
What we learned
How to use Agentic AI in real workflows.
Practical challenges of time-series forecasting for demand prediction.
Importance of user-centric design in AI applications.
What’s next
Deploy DineFlow AI in a real college/hospital canteen.
Add voice ordering in local Indian languages.
Connect with digital payment gateways for real transactions.
Scale to other domains (schools, corporate offices).
Log in or sign up for Devpost to join the conversation.