What it does
Smart Bite innovates with Agentic AI by dynamically generating recipes using ingredients that are nearing expiration, helping users reduce food waste. The app empowers users to manage their pantry more efficiently by allowing them to scan food item images with AI, which tracks expiry dates and organizes items. With timely notifications, the app ensures users are always aware of expiring items, encouraging sustainable and proactive pantry management. By seamlessly processing user inputs with multiple AI models, Smart Bite bridges the gap between intelligent decision-making and practical utility.
How I built it
The project combines a robust tech stack with powerful AI models to deliver a scalable and intelligent experience:
Frontend: Built with Vite.js to ensure a fast, visually appealing, and responsive user interface, making interactions with the AI intuitive and enticing. Backend: Developed with Node Express, featuring RESTful APIs that interact with a MongoDB database to store and manage pantry data, including expiry dates. Sambanova Cloud AI Models: Llama 3.2 Vision 90B model: Recognizes and tracks food items by processing images. Llama 3.1 70B instruct model: Generates recipes tailored to expiring ingredients using contextual prompts in a multi-model workflow. Notification System: Integrated to send proactive alerts (via email) when items approach their expiry dates, closing the loop on smart pantry management. Agentic Workflow: Prompts chain seamlessly across models and database queries, allowing the system to make proactive decisions, such as prioritizing recipes that reduce food waste.
Challenges I ran into
Integrating the AI models for both recipe generation and image recognition with the backend was one of the key challenges. Ensuring that the AI could process both text and images effectively required careful coordination between the models and the database. Additionally, tracking expiry dates accurately and ensuring timely recipe generation were important hurdles to overcome.
Accomplishments that I'm proud of
I’m proud of successfully integrating agentic AI models for both recipe generation and image scanning, creating a seamless and intelligent user experience. The AI not only generates recipes based on expiring ingredients but also actively manages pantry items by recognizing and organizing them through image scanning. The added notification feature, alerts users when items are nearing expiration, ensuring they never miss the chance to use their ingredients before they go to waste. This agentic approach helps reduce food waste and promotes sustainability by enabling proactive, intelligent pantry management.
What I learned
Through this project, I learned how to work with powerful AI models for both text and image processing. I also gained hands-on experience in using Vite.js, Node Express, and MongoDB to build a web application with real-world functionality. Integrating these components required me to think critically about how different systems interact and how to ensure a smooth user experience.
What's next for Smart Bite
In the future, I plan to improve the image recognition system to make pantry item identification even more accurate. I also want to introduce more customization options for recipe generation, such as allowing users to filter recipes based on dietary restrictions. Additionally, I aim to optimize the backend to ensure scalability as the app grows.
Built With
- emailjs
- express.js
- javascript
- llama-3.1-70b
- llama-3.2-vision-90b
- mongodb
- node.js
- restful
- sambanova-cloud
- tailwind
- vite.js
Log in or sign up for Devpost to join the conversation.