Inspiration
The idea for ShopSmart AI came from a real frustration while shopping online. I was looking for sustainable bedsheets for my family — something like organic cotton — but every product page was full of vague claims like “eco-friendly” and “all-natural.” At the same time, reviews looked fake and repetitive, all saying “amazing product!” After spending nearly an hour cross-checking reviews and brand certifications, I bought a set that fell apart after a few washes. It was a waste of money and bad for the environment. That’s when I realized online shopping forces consumers to be detectives when all we want is trust. I thought, “What if my browser could instantly tell me whether a product is truly trustworthy and sustainable?” — and that’s how ShopSmart AI was born.
What it does
ShopSmart AI is a Chrome extension that brings trust and transparency to e-commerce. It uses Google’s built-in Gemini Nano AI to analyze any product page and provide two smart insights: EcoScore: Evaluates environmental impact based on materials, packaging, and certifications. Trust Score: Detects fake reviews and measures how reliable a product’s rating really is. Users can toggle between modes with a single click. It works on sites like Amazon and Flipkart, providing clear, AI-powered insights — all processed locally for total privacy.
How we built it
We built ShopSmart AI as a lightweight, privacy-first Chrome Extension using HTML, CSS, and JavaScript. The real magic lies in Google Chrome’s Built-in AI APIs, which run directly on the user’s device. Prompt API (Gemini Nano): Powers EcoScore & Trust Score via structured prompts. Summarizer API: Condenses long product data for faster analysis. Writer & Rewriter APIs: Simplify technical outputs into user-friendly text. Side Panel & Storage APIs: Handle UI and caching for instant, offline results. Everything happens on-device, ensuring it’s fast, secure, and private — no user data ever leaves the computer.
Challenges we ran into
Our main challenge was prompt engineering — teaching the AI to look beyond buzzwords like “eco-conscious” and find verifiable proof such as GOTS or Fair Trade certifications. We also faced data extraction issues since each e-commerce site has different structures. To solve this, we built a flexible parsing system that adapts dynamically, allowing ShopSmart AI to work smoothly across Amazon, Flipkart, and more. Balancing accuracy, speed, and on-device performance required countless iterations and optimizations. The relationship can be expressed as:
$$\text{Performance} = \frac{\text{Accuracy} \times \text{Speed}}{\text{On-Device Latency}}$$
Accomplishments that we're proud of
Built a dual-mode AI system that simplifies complex analysis. Achieved fast, private, on-device AI performance using Gemini Nano. Solved fake review detection and greenwashing in one solution. Created a smooth user experience with instant, offline results. We’re proud that ShopSmart AI isn’t just a demo — it’s a tool that genuinely helps users shop smarter, greener, and with confidence.
What we learned
We learned the power of on-device AI and how privacy and intelligence can coexist. It’s possible to build complex, real-time analysis tools entirely on the client side. Our biggest takeaway was mastering prompt engineering — understanding how context and constraints shape AI accuracy. We also learned to balance UX with model performance, ensuring instant insights $\text{where } \Delta t \to 0$ without compromising privacy. This project showed us that the future of AI isn’t just in the cloud — it’s in your browser.
What's next for Shopsmart
This is just the start. We plan to expand ShopSmart AI into a full ethical commerce platform, featuring: Integrated price tracking with EcoScore & Trust Score. Carbon footprint estimation for entire carts. Integration with second-hand marketplaces like eBay and Poshmark. Community-driven ethics data to verify brand claims. Our goal is to make every online purchase transparent, sustainable, and trustworthy — powered by privacy-first AI.
Log in or sign up for Devpost to join the conversation.