Project Info

SmartRation

Inspiration

We wanted to tackle two huge pain points for under-resourced households: stretching a tight grocery budget and minimizing food waste. By combining off-the-shelf OCR and computer-vision models with powerful LLMs, we realized you could turn a simple snap of your fridge plus yesterday’s receipt into a fully personalized, budget-smart meal plan.

What it does

SmartRation packs five core features into one seamless dashboard: FoodGPT Chat Assistant – Ask for recipe ideas, cooking tips, dietary advice, and personalized meal suggestions in natural language. – Handles follow-up questions about allergies, household size, spice tolerance, and budget limits. Upload Receipt OCR – Snap or upload your grocery receipt to instantly extract items, quantities, and prices. – Automatically logs your spending to inform every downstream feature. AI Meal Plans – Generate a 7-day, 3-meals-per-day plan tailored to your budget, preferences, and what’s in your fridge. – Slide through daily cards, pick between variations, or hit “Regenerate” for fresh ideas. Smart Shopping List Generator – Based on your remaining ingredients, meal plan, and spending cap, build a budget-optimized shopping list. – Group items by store sections or preferred grocery APIs (Kroger, Walmart, etc.) for the fastest trip. Saved & Rated Meals – Bookmark your favorite dishes and rate each meal after you cook it. – View past successes at any time and let FoodGPT learn from your ratings to refine future recommendations.

How we built it

Frontend: typescript, tailwind css, shadn UI. Backend: Supabase for auth, Postgres tables (profiles, receipts, meal_plans, foodgpt_messages) with RLS. OCR: Google Vision Cloud API LLM: Anthropic Claude 2

Challenges we ran into

We had trouble training the OCR to recognize different receipt formats and types to improve its accuracy. We had to become prompt engineers for claude and its uses to create a curated shopping list and our own fine-tuned chatbot.

Accomplishments we're proud of

We're proud of the accuracy of our image detection software and our use of 3 apis.

What we learned

We learned a lot about prompt engineering and how providing excellent context is most beneficial. We learned about creating an amazing impact in such a short period. We all feel motivated by this to continue to create and learn.

What's next

We want to continue making our OCR detection more accurate and then integrate grocery shop apis to further get accurate food details. We want to get a customer and user base going so that our app creates value for those who need it.

Hackathon Project

Welcome to the hackathon project repository!

Getting Started

This repository is ready for your hackathon project. You can start adding your code, documentation, and other project files.

Project Structure

.
├── README.md
├── .gitignore
└── [Your project files will go here]

Contributing

Make your changes
Commit your changes with descriptive commit messages
Push to your branch
Create a pull request if working with a team

License

[Add your license information here]

Analysis

Compare with all teams

View

Metric

Sodbayar Ganbat
19
Robert Flores
1

Figures cover GitHub contributors during the hackathon window. A co-authored commit counts in full for each author, so per-member totals add up to more than the whole-team figures.

Technology

Found in codeClaimed only

AnthropicIn code
CSSIn code
Next.jsIn code
ReactIn code
SQLIn code
SupabaseIn code
Tailwind CSSIn code
TypeScriptIn code

8 of 8 appear in the indexed code.

AI coding agents

No AI coding agent signals were found in this repository.

Detected from committed agent config files and commit authorship. Absence of a signal is not proof an agent was unused.

Codebase size

Source size

359 KB

Source files

Counts recognized source files only; vendored directories, binaries and lockfiles are excluded, so this is smaller than the repository on disk.