Oishi

Inspiration

We have all thought about setting nutrition goals, many have made plans, several have tried to follow through by collecting recipes, and some have given up when recipes don't translate well to grocery budgets and time constraints. It takes herculean effort to make and stick to meal plans, but a lot of that friction comes not from our own lack of willingness and motivation, but rather from the lack of tools that bring together personalized meal planning, grocery shopping list management, recipe collections, nutrition tracking and expert consultations in one place. We wanted to address this gap in the market.

What it does

Oishi collects the user's preferences, goals, grocery budget and dietary restriction to create a AI-generated meal plan that is trained on inputs from professionals like dieticians, nutritionists and trainers. Then, it collects recipes from various sources on the internet, extracts ingredients lists and nutrition content, which you can add to your tracker. This helps you visualize your entire meal plan in a dauly, weekly and monthly view, and track progress against your goals. Once you have populated your meal plan, you can generate your grocery list, and while you are shopping, you can have the built-in AI suggest alternatives if anything is out of stock. After you make and eat the meal, you can check that off on the tracker so that your progress is saved. Lastly, the social component of the app allows you to find and follow experts and influencers who share their own recipes to inspire you, and you can also share your own results with the recipe you followed to motivate you to stay on track with your plan and also see what other people are experimenting with.

How we built it

We first came up with our idea based on unmet user needs we identified in this space. Then we came up with our initial roadmap and optimized our prompts using Perplexity and Gemini. We then generated our prototype on loveable and iterated as we refined our UI.

Challenges we ran into

Lack of a trained LLM with food related data LLMs don't understand nuances in preferences - when I dislike something doesn't mean I hate it and never want it featured in my recipes.