ByteMyMood

• 

• 

  agent architecture

Inspiration

What inspired us to create ByteMyMoodis a mix of personal frustration and a shared desire to leverage technology to solve everyday problems. As recent graduates, we've all spent countless nights fueled by instant ramen and questionable leftovers during intense coding sessions and exam periods. We noticed it wasn't just us; nearly everyone we knew struggled with consistent, healthy meal planning.

The sheer mental load of deciding "What should I eat tonight?" or "What groceries do I need?" felt like an unnecessary drain on our limited time and energy. That's when the lightbulb went off. We thought, "There has to be a better way to automate and optimize this." The idea of building an intelligent system that could not only suggest recipes based on preferences and dietary needs but also manage grocery lists and even consider ingredients already on hand, was incredibly appealing to us.

The hackathon provided the perfect crucible for this idea. It was an opportunity for us to take a common pain point and apply our collective skills to create something tangible and genuinely useful. Ultimately, ByteMyMood was born from our desire to make healthy eating more accessible and less stressful for busy individuals, just like us and our peers.

What it does

It's designed to be your ultimate kitchen assistant, tackling food waste and mealtime dilemmas head-on. Here is a list of tasks that it does:

1 . Recipe Suggestions: Provides food suggestion based on how you are feeling and your location. It also considers your dietary restrictions.

1. Inventory recognition: Insert a picture of the inside of your fridge, and the agent goes to work to provide you with a list of everything it identifies inside. It's like having a super-smart inventory manager for your perishable goods!

2. Generates Smart Grocery Lists: Once ByteMyMood knows what you have on hand, it starts to generate a shopping list based on the ingredients that you are missing. If a recipe requires ingredients you don't have, it automatically adds them to a categorized grocery list, optimizing for efficiency and minimizing impulse buys.

4.Offers cooking instructions and tips: Once you pick a recipe, it provides clear, step-by-step instructions, and even suggests helpful cooking tips or substitutions, and pictures to follow along.

In essence, ByteMyMood takes the guesswork and stress out of meal planning, making it easier for users to eat well, save money, and contribute to a more sustainable lifestyle by reducing food waste.

How we built it

ByteMyMood was built using Google’s Agent Development Kit (ADK) to orchestrate a modular multi-agent system. Each sub-agent — Inspiration, Planning, and Execution — handles a specific phase of the user journey.

We integrated the following technologies:

• Google Weather API and Google Search Grounding Tool for real-time, context-aware recipe suggestions.
  
• Gemini 2.0 Flash to generate visual aids for each cooking step.
  
• ADK’s built-in session management to enable:
  • Smooth multi-turn conversations
    
  • Error handling and fallback logic
    
  • Dynamic task delegation across agents
    

This architecture allows us to deliver a personalized and responsive cooking assistant that adapts to user mood, environment, and preferences.

Challenges we ran into

One of the main challenges was maintaining agent reliability when dealing with unexpected or vague user input. Traditional programming offers predictable logic and debugging tools, but prompt-based agents rely on natural language, which introduces ambiguity. It was especially difficult to ensure agents responded accurately when user commands were incomplete, contradictory, or outside expected patterns.

Accomplishments that we're proud of

• Built a fully functional multi-agent system using Google ADK to automate recipe planning from mood to meal.
  
• Successfully integrated weather, mood, and dietary data into personalized recipe suggestions.
  
• Enabled fridge image parsing for inventory tracking using memory tools.
  
• Generated AI-powered cooking visuals with Gemini for better user experience.
  
• Designed a modular, extensible agent architecture that's easy to maintain and scale.

What we learned

• Gained hands-on experience with multi-agent orchestration and task delegation using Google ADK.
  
• Learned the importance of prompt engineering for achieving reliable, consistent agent behavior.
  
• Understood the challenges of managing context, fallback handling, and natural language variability.
  
• Discovered how to blend AI capabilities with real-world user needs in a cooking-related domain.

What's next for ByteMyMood

We plan to integrate ByteMyMood as the intelligent backend for a web or mobile app, enabling a more interactive and accessible user experience. Future features include:

• Calorie tracking and personalized dietary goal recommendations
  
• Voice-controlled cooking assistance for a hands-free kitchen experience
  
• Better fridge image parsing using custom-trained food detection models
  
• Integration with grocery delivery services for one-click ingredient ordering

Our long-term goal is to make ByteMyMood a complete smart kitchen companion for anyone looking to cook better, eat healthier, and save time.

Built With

• google-cloud
• python