NutrAIcionist

Inspiration

For my daily routine, I wanted to bridge the gap between complex nutritional data and actionable daily advice. NutrAIcionist was inspired by the idea of having a personal dietician in your pocket—someone who knows your goals, allergies, and dietary preferences, and can instantly analyze your food just by looking at a photo.

What it does

NutrAIcionist is a comprehensive AI-powered food analysis web application:

• Visual Analysis: Users upload a photo of a meal or raw ingredients. The app uses Google Gemini to identify the food items and classify them as a prepared "Dish" or "Ingredients"(it reasons to deliver different results according to the photo).
• Personalized Insights: It calculates macronutrients (protein, carbs, fat, calories) and provides a specific "Health Score" (1-10) based on nutrient density and processing levels.
• Context-Aware Advice: It checks the food against a user's specific profile (Vegan, Keto, Allergies, Muscle Building, etc.). It warns about allergens and calculates a Goal Alignment Score to show how well the meal fits the user's health objectives.
• Smart Recommendations:
  • For Dishes: It suggests portion sizes, immediate meal advice, recommendations for the next meal, and offers two types of alternatives: similar healthier versions and completely different dishes that better fit the user's goals.
• For Ingredients: It acts as a chef, suggesting recipes to cook with the identified items and highlighting the specific health benefits of key ingredients.
• Search Grounding: Using Gemini's Google Search tool, it provides real-world links on where to buy the dish (restaurants/stores) or tutorials on how to cook the ingredients.
• Tracking: It saves scan history to visualize macronutrient trends over time and allows users to favorite specific items or alternatives.

How we built it

The platform was built with AI Studio, including the following technologies: a

• Frontend: Built with React (TypeScript) and Vite for a fast, responsive UI.
• Styling: Used Tailwind CSS for a clean, modern, and accessible design with custom color schemes for health ratings.
• AI Engine: Powered by the Google GenAI SDK (@google/genai) using the gemini-3 model, with ultimodal capabilities to analyze images, sructured Output to guarantee the AI returns valid JSON for the UI to render.
• Data Visualization: Integrated Recharts to render interactive pie charts for macronutrient breakdown and line charts for historical trends.
• Persistence: Leveraged browser localStorage to persist user profiles, scan history, and favorites without needing a backend database.

Challenges we ran into

The biggest challenge was creating a prompt that effectively synthesizes visual data with abstract user goals (e.g., "I want to build muscle") into concrete numerical scores (Goal Alignment Score), which required several iterations of prompt engineering.

Accomplishments that we're proud of

I successfully created a single flow that intelligently switches between "Meal Analysis" (macros, portions) and "Ingredient Analysis" (recipes, shopping) based solely on the image content. And the "Goal Impact Analysis" gauge is a standout feature that translates abstract nutritional data into a simple "Good/Bad for YOUR specific goal" metric.

What we learned

The most I learnt was the capabilities of Gemini 3 to make a platform with multi-modal functions, not just to understand an image, but also to mix with large text to deliver the best result as possible.

What's next for NutrAIcionist

• Social Features: Allowing users to share their "Healthy Streaks" or favorite AI-generated recipes with friends.
• Pantry Integration: expanding the "Ingredients" mode to scan a whole fridge and generate a weekly meal plan to minimize waste.

Built With

• aistudio
• gemini