Mealer

Plan meals. Waste less. Spend smarter. Live better.

Comment

Inspiration

Planning meals sounds simple, but it quickly becomes a repetitive cycle of checking the fridge, finding recipes, managing leftovers, buying missing ingredients, and fitting meals around a changing schedule. This friction often leads people to spend more on takeaway while food at home goes unused.

We created Mealer to make home cooking easier, more affordable, and less wasteful by connecting pantry tracking, recipe discovery, meal preparation, weekly planning, and grocery spending in one place.

Plan meals. Waste less. Spend smarter.

What it does

Mealer helps users manage the journey from grocery receipt to prepared meal:

  • Scan a grocery receipt to extract food items, quantities, and prices.
  • Review and correct scanned items before adding them to the pantry.
  • Track pantry quantities, expiry dates, and the estimated value of food on hand.
  • Generate AI-powered recipes using available ingredients, prioritising food that expires soon.
  • Compare recipes by ingredient availability and cost per serving.
  • Bookmark recipes and record prepared meals with their available portions.
  • Drag and drop prepared meals into a visual weekly calendar.
  • Automatically reduce the number of available portions as meals are scheduled.
  • Track grocery spending and meal-preparation costs by week.
  • Generate shopping-list requirements from planned meals and current pantry contents.

How we built it

Mealer is a full-stack mobile application built with:

  • React Native and Expo for the cross-platform mobile experience.
  • TypeScript and Expo Router for typed screens, navigation, and API integration.
  • ASP.NET Core and C# for authentication, validation, business logic, and REST APIs.
  • Entity Framework Core and PostgreSQL for relational and user-scoped data storage.
  • OpenRouter-powered language and vision models for receipt understanding and recipe generation.
  • JWT authentication to protect individual user data.

Receipt photos are sent to a vision model using a strict structured-output schema. Users review the result before anything is written to their pantry.

Recipe generation is also grounded in backend data. The model receives the user’s current ingredients, expiry dates, saved recipes, and an approved ingredient catalogue. The backend validates generated ingredient IDs and units before saving the recipes.

Challenges we ran into

One of our biggest challenges was turning inconsistent receipt text into reliable pantry data. Product descriptions can wrap across lines, use abbreviations, contain package sizes, or show quantities differently. We developed detailed extraction rules and added a review step so users remain in control.

Inventory measurement was another challenge. Recipes and receipts can describe the same ingredient using grams, kilograms, millilitres, litres, or individual pieces. Keeping quantities consistent is essential for recipe matching, cost calculations, and pantry deductions.

We also had to keep AI behaviour grounded. A model can generate creative suggestions, but it should not invent database records, ingredient identifiers, or quantities that cannot be verified. Structured schemas and backend validation became an important part of our design.

Finally, implementing an intuitive drag-and-drop weekly planner across mobile devices required careful coordination between gestures, calendar drop zones, backend updates, and prepared-meal portion counts.

Accomplishments that we're proud of

We are proud that Mealer is more than a collection of disconnected screens—it supports a working journey from buying groceries to planning when they will be eaten.

In particular, we built:

  • An AI receipt scanner with a user verification workflow.
  • Pantry and grocery-cost tracking backed by PostgreSQL.
  • AI recipes grounded in the user’s real pantry and expiry dates.
  • Recipe cost-per-serving calculations.
  • Prepared-meal and remaining-portion tracking.
  • A flexible weekly planner with drag-and-drop interaction.
  • Weekly grocery spending and meal-preparation summaries.
  • A C# backend that verifies and persists user data instead of allowing the AI to write unchecked results.

What we learned

We learned that successful AI products need strong conventional software around the model. The AI is useful for interpreting receipts, generating recipes, and explaining suggestions, while the backend should remain responsible for authorization, validation, calculations, and data integrity.

We also learned that reducing food waste is not only about recommending recipes. The entire workflow must be convenient: capturing groceries, understanding what is available, preparing food, scheduling portions, and seeing the financial impact.

Most importantly, users need flexibility. AI should reduce planning effort without removing control, so generated results should always remain editable and understandable.

What's next for Mealer

Our next step is to let Mealer generate a complete weekly meal-plan proposal based on pantry contents, expiry dates, budget, preparation time, dietary preferences, and the user’s schedule. Users will be able to accept the proposal or adjust it through the existing planner.

We also plan to add:

  • Automatic shopping lists that account for pantry quantities and unit conversions.
  • Fastest and easiest shopping-route recommendations based on selected stores.
  • Expiry reminders and proactive “use this next” suggestions.
  • Nutrition and dietary-goal tracking.
  • Shared household pantries and collaborative meal planning.
  • Better personalisation based on saved recipes and planning history.
  • More detailed spending insights and estimated savings from eating at home.

Built With

Share this project:

Updates