Nourishr

The work

Nourishr is a 2024 health mobile app for iOS and Android, live at https://nourishr.app/. The tagline is a time claim: decide what to eat in five seconds. The product is an AI meal companion that reads cravings, dietary needs, and what is already in the fridge, then recommends a meal. Computer vision and a recommendation engine sit at the center. A user can snap ingredients and get recipes, keep a pantry with expiration alerts, run photo-based nutritional analysis, and send an order path through Uber Eats and DoorDash. Twelve or more dietary patterns are supported. Public results list 10,000+ active users, 250+ hours saved per user each year, 95% recommendation accuracy, and a 40% reduction in food waste. The average person, per the challenge, already spends 250+ hours a year deciding what to eat. Nourishr is the product built to make that decision instant and personal.

Decision fatigue around food is the problem the overview names. The challenge repeats the 250+ hour figure and asks for an AI-powered solution that makes meal decisions instant and personalized. The solution is a recommendation engine that learns user preferences, combined with computer vision for ingredient scanning and nutritional analysis. Mood is part of the input, alongside the fridge and the diet pattern. Five seconds is the target for a recommendation. That is both the tagline and a listed feature: AI meal recommendations in five seconds. The rest of the feature set exists to feed that recommendation or to act on it. Snapping ingredients fills the pantry. Expiration alerts keep that pantry honest. Photo analysis estimates nutrition. Uber Eats and DoorDash cover the nights when cooking is not the answer. Twelve-plus dietary patterns keep the engine from recommending a meal the user cannot eat.

Ingredient scanning is the computer-vision path. A user photographs what is on the counter or in the fridge, and the app turns that photo into recipes that can be made from those items. OpenAI Vision and TensorFlow are the listed vision stack, with Python on the model side and React Native on the phone. The same photo path supports nutritional analysis, so a plate or a set of ingredients can be read for more than recipe fit. Smart pantry tracking then holds those items over time and fires expiration alerts, which is how the product connects a single snap to a week of decisions. The 40% food-waste reduction in the results is the outcome attached to that loop: see what you have, cook it before it expires, and stop buying a third jar of the same sauce. The pantry is a living input to the five-second recommender.

Dietary patterns and delivery close the other half of a meal decision. Nourishr supports 12+ dietary patterns, so a recommendation can respect those patterns and the user's learned preferences. When the answer is not a recipe from the pantry, Uber Eats and DoorDash are integrated so the same companion can point at a meal that arrives. The product is still making a five-second call. It is choosing among cook-from-what-you-have, cook-from-a-recipe, and order-in, using mood, diet, and inventory. Photo-based nutritional analysis gives that choice a numbers layer when the user wants to see what a plate means. Supabase holds the account, pantry, and preference data. AWS hosts the heavier Python and TensorFlow work. The mobile clients stay on React Native for both iOS and Android.

The recommendation engine is described as something that learns. Preferences accumulate. A mood can change the suggestion even when the pantry stays the same. Accuracy is published at 95% for recommendations, which is the figure attached to that learning loop. Time saved is published as 250+ hours per user per year, which matches the challenge's estimate of how long people already spend deciding. Those two numbers are how the product claims the five-second decision is worth installing. The 10,000+ active-user count is the scale attached to the same claims. Nothing in the source adds restaurant partnerships beyond Uber Eats and DoorDash, and nothing adds a new diet count beyond 12+. The industry is Health. The groups are mobile and AI. The tags are AI, Computer Vision, and Health.

A typical path through the app, using only listed features, looks like this. Open Nourishr on iOS or Android. Snap the fridge or the counter. Let pantry tracking store the items and watch expiration dates. Ask for a meal. Get an AI recommendation in about five seconds, filtered by a supported dietary pattern and by what the engine has learned. If the meal is homemade, use the instant recipes from the snap. If the meal is bought, hand off to Uber Eats or DoorDash. If the question is what a plate contains, run photo-based nutritional analysis. That is the companion loop. It is meant to replace the long stare into a fridge or a delivery app. The 2024 product at nourishr.app is that loop on a phone, with Python, TensorFlow, and OpenAI Vision doing the seeing and the suggesting.

Results stay inside four published claims: 10,000+ active users, 250+ hours saved per user annually, 95% recommendation accuracy, and a 40% reduction in food waste. The stack is React Native, Python, TensorFlow, OpenAI Vision, Supabase, and AWS. The live URL is https://nourishr.app/. Platforms are iOS and Android. Year is 2024. Features are the five-second recommendation, ingredient snaps, pantry expiration alerts, photo nutrition, Uber Eats and DoorDash, and 12+ dietary patterns. The challenge, the overview, and the results all return to the same time problem: people spend hundreds of hours a year deciding what to eat, and Nourishr is the 2024 mobile companion built to make that decision in seconds while using what is already in the house.