ConsumeWise v0.1
Decode your choices, elevate your health.
Created on 1st October 2024
•
ConsumeWise v0.1
Decode your choices, elevate your health.
Describe your project
Now a days, there is a growing demand for healthier alternatives of regularly consumed FMCG food products, but there is a gap in awareness. Many consumers express a desire for healthier options, but they often lack the knowledge to discern what is genuinely beneficial. This confusion is made worse by companies that exploit this lack of understanding, making misleading claims on packaging that deceive consumers into purchasing substandard products. While there are some online/youtube reviewers who try to bridge the information gap, the vast product variety makes comprehensive analysis and conveying it in an easy to understand manner challenging and time-consuming.
ConsumeWise addresses this issue by serving as a product review catalog and consumption tracker powered by google gemini. The platform analyzes health products, monitors users’ consumption patterns, and suggests small, manageable steps for improvement, leading to significant health benefits over time.
It includes:
Data Collection: Scrapes data from major FMCG platforms (Bigbasket for prototyping) to compile a comprehensive product list.
AI-Generated Insights: Provides ingredient breakdowns, nutritional information, identifies misleading claims, and offers tailored recommendations based on users' health needs.
Regular User Consumption Analysis: It generates regular reports evaluating how users’ choices align with their health goals and provides recommendations for improvement and to healthier alternatives.
For the hackathon prototype, My focus was on personalization and user-friendly information presentation. In future I will aim to expand product listings from all major FMCG sites, enhance personalization with user location and budget data, integrate with health apps like HealthifyMe, and enable retailers to list their products directly on the platform. These enhancements will make ConsumeWise more comprehensive resource for health-conscious consumers.
Challenges I ran into
Generating a personalized product overview using genAI on the fly with limited resources was little tricky. I wanted to make the personalized overview page as detailed and informative as possible.
Initially, I made it such that when users logged in and navigated to a particular product details page, to speed up the process I tried to sending separate requests to the API.
However, this approach caused two issues. First, I had to send the same large context to the LLM for multiple requests, and it mostly triggered a "too many requests" error. To resolve this, I finally (for now) decided to send the request only from the top-level component and cache the response in the database for later use. While this solution is a bit slower, it works reliably.
Additionally, I wanted to implement a personalized recommendation feature using collaborative filtering approach. Unfortunately, my implementation performed very poorly with limited user data. Given the limited time, I had to drop the recommendation feature for now.
Other than that everything went smoothly and I plan to continue to add more features moving forward.
Tracks Applied (1)
4. Problem statement shared by People+ai (ConsumeWise)
Cheer Project
Cheering for a project means supporting a project you like with as little as 0.0025 ETH. Right now, you can Cheer using ETH on Arbitrum, Optimism and Base.