PlateWise

PlateWise transforms any restaurant menu into personalized AI meal recommendations tailored to your allergies, dietary needs, and health goals.

Comment

PlateWise

Inspiration

Eating out should be enjoyable, not stressful. However, for people with food allergies, dietary restrictions, or specific health goals, choosing what to eat from a restaurant menu can be difficult. Nutritional information is often unavailable, ingredients may be unclear, and comparing dishes manually is time-consuming.

We wanted to build an AI-powered assistant that helps people make smarter dining decisions by turning any restaurant menu into personalized recommendations based on their individual food profile. Our goal was to create a solution that makes healthy and informed dining more accessible without requiring restaurants to change their existing menus.

What it does

PlateWise is an AI-powered restaurant menu recommendation platform that personalizes meal suggestions based on each user's dietary preferences and nutrition goals.

Users can:

  • Create a personalized food profile with allergies, dietary restrictions, cuisine preferences, and health goals.
  • Upload restaurant menus as PDFs or images, or simply paste a restaurant website URL.
  • Automatically extract menu items using OCR and menu parsing.
  • Estimate nutritional information such as calories, protein, carbohydrates, fats, and potential allergens.
  • Receive AI-generated rankings of the most suitable meals, along with clear explanations for each recommendation.
  • Save recommendation history for future reference.

PlateWise helps users spend less time guessing and more time enjoying meals that align with their personal needs.

How we built it

PlateWise was developed as a full-stack web application.

Frontend

  • Next.js
  • TypeScript
  • Tailwind CSS
  • Modern responsive UI components

Backend

  • FastAPI
  • Python
  • Pydantic
  • REST APIs

Database

  • MongoDB
  • PyMongo

AI Pipeline

The recommendation engine combines:

  • User dietary profile
  • Uploaded restaurant menu
  • OCR for scanned images and PDFs
  • Website menu parsing
  • OpenAI-powered nutrition estimation
  • AI-powered personalized ranking

Instead of relying on fixed nutritional databases, the AI estimates nutritional values directly from the provided menu descriptions before comparing every dish against the user's preferences, dietary restrictions, allergies, and health goals.

Challenges we ran into

One of the biggest challenges was converting restaurant menus into structured information. Menus vary significantly in formatting, making it difficult to consistently extract dishes and descriptions from PDFs, images, and websites.

Another challenge was generating recommendations responsibly. Since nutritional values are estimated from menu descriptions, we designed the system to avoid making medical claims and instead present recommendations with clear explanations and transparency.

We also spent considerable time integrating the AI recommendation pipeline, handling API communication, prompt engineering, and ensuring recommendations remained relevant while respecting user allergies and dietary restrictions.

Accomplishments that we're proud of

  • Successfully built an end-to-end AI-powered recommendation platform.
  • Created a unified workflow that supports menu uploads, restaurant website analysis, and personalized recommendations.
  • Developed a transparent recommendation engine that explains why each meal is recommended rather than providing black-box results.
  • Integrated OCR, menu parsing, nutrition estimation, and AI ranking into a seamless user experience.
  • Built a scalable architecture using modern technologies that can easily support future enhancements.

What we learned

Throughout this project, we gained valuable experience in integrating large language models into a production-style application. We learned how important prompt engineering, structured API design, and data validation are for producing reliable AI outputs.

We also discovered that building trustworthy AI goes beyond generating recommendations—it requires providing explanations, acknowledging uncertainty, and designing systems that prioritize user safety and transparency.

Working as a team also strengthened our collaboration skills, from coordinating frontend and backend development to integrating AI services into a unified product.

What's next for PlateWise

Our vision is to evolve PlateWise into a comprehensive AI-powered dining companion.

Future improvements include:

  • Live restaurant integrations through official APIs.
  • Barcode and packaged food scanning.
  • More accurate nutrition estimation using verified food databases.
  • Meal tracking and daily nutrition summaries.
  • Fitness wearable integration.
  • Personalized meal planning across multiple restaurants.
  • Multi-language menu support.
  • Restaurant analytics dashboards for businesses.
  • Stronger authentication and production-ready cloud deployment.
  • Continuous learning to provide increasingly personalized recommendations while maintaining user privacy.

Ultimately, we want PlateWise to empower people to make healthier, safer, and more confident dining choices wherever they eat.

Built With

Share this project:

Updates

Submission history