EcoSight

Inspiration

We were inspired by the growing desire among consumers to make more sustainable choices, contrasted with the difficulty of accessing reliable, easy-to-understand information about product impacts at the point of purchase. We wanted to bridge this gap by leveraging AI, big data, and machine learning to provide immediate, actionable insights, empowering everyone to shop and live more sustainably.

What it does

EcoScan is a AI-powered Progressive Web App (PWA) that acts as a personal sustainability companion. It allows users to:

Analyze Products: Get detailed sustainability assessments of products through text input or potentially scanning.
Discover Alternatives: Receive suggestions for greener product alternatives based on environmental impact data.
Ask Sustainability Questions: Chat with a AI assistant (powered by Gemini) for real-time answers to eco-related queries.
Get Personalized Tips: Receive tailored recommendations and lifestyle tips based on user preferences and interactions.
Learn & Engage: Access educational content on sustainability and engage with community-submitted tips.

How we built it

Frontend: A Progressive Web App built with React and TypeScript, using Vite for fast development and optimized builds. We used Tailwind, ShadCN, and three.js for the UI.
Backend: MongoDB Atlas and Cloudflare provide a scalable, DDOS-resistant backend that can handle a large number of users easily.
AI: Google Gemini 2.5 Pro and 2.0 Flash power the image recognition and response generation features.

Challenges we ran into

This was our first time making a PWA and a sustainability app, and we dsefinitely learned a lot along the way.

Finding out what sources to use to get accurate data
Understanding how to prompt LLMs to minimize hallucinations
Designing efficient database schemas in MongoDB Atlas and managing the flow of data between the PWA, Gemini, and the database securely and efficiently.

Accomplishments that we're proud of

Successfully integrating a powerful LLM to provide real-time sustainability insights within a user-friendly interface.
Building a PWA that offers an installable, native experience.
Designing and implementing a scalable backend.
Helping people reduce their carbon footprint

What we learned

LLM Integration: Gained practical experience in leveraging large language AIs for specific application tasks, including prompt design and API interaction.
PWA Development: Learned techniques for building performant PWAs with rich user experiences using Vite and React.
Full-Stack Application Design: Learned how to design and connect different components (frontend, backend, database, external AI service) into a cohesive and functional application.

What's next for EcoSight

Expand our Product Database: Integrate with more comprehensive datasets/APIs to increase the coverage and accuracy of product analyses.
Enhance Barcode Scanning: Implement or refine barcode scanning functionality for quicker product lookups.
Refine AI Responses: Iprove the Gemini prompts based on user feedback to provide even more accurate and helpful insights.
Develop Community Features: Build out the community engagement aspects, allowing users to share tips, rate products, and discuss sustainable practices.