MASala

Home Page
Agent working
Curated Recipe
Recipe Card
Agent Logs
Dashboard
Dashboard

💡 Inspiration

As students, we often find ourselves drained after long days of lectures, staring into our fridges and wondering what to cook. But this challenge goes far beyond student life , it affects busy professionals, health-conscious individuals, people with dietary restrictions, and anyone who simply wants to make better use of what's already in their kitchen.

Finding the right recipe online is time-consuming, juggling ingredient availability, allergy filters, cooking time, and nutrition can feel overwhelming.

That’s why we created MASala, a Multi-Agent System that takes all that cognitive load off your plate. It’s a smart, personalized cooking assistant that works for everyone, whether you're a student with limited groceries or a home chef looking to try something new, safely, quickly, and deliciously.

🏷️ Why the name "MASala"

The name MASala is a blend of meaning and emotion. "MAS" stands for Multi-Agent System, the core technology powering our project. "Masala" is a staple spice mix in Indian cuisine, evoking memories of flavor, warmth, and home-cooked meals.

🍽️ What it does

MASala takes your:

Available ingredients
Allergies
Dietary restrictions
Preferred dish type

...and generates three fully customized recipes, filtered for safety, tailored for your needs, inspired by culture, and beautifully presented.

Behind the scenes, a team of intelligent agents works together in perfect sync, just like a professional kitchen brigade, to deliver recipes that are smart, safe, and satisfying.

🛠️ How we built it

We used a combination of powerful tools and clear agent design to bring MASala to life:

CrewAI for orchestrating agent interactions and task flow
LangChain + Gemini (Google Generative AI) for understanding user inputs and generating recipe content
Custom tools for ingredient filtering, nutritional analysis, web search, and structured formatting
React frontend to provide a smooth, user-friendly experience
JSON + dashboard to visualize each agent's progress, status, and outputs

Each agent in MASala was designed with a distinct, specialized role:

🧠 Web Analyzer: Understands your context and cooking goals
🥗 Nutritionist: Filters out unsafe ingredients and ensures dietary compliance
👨‍🍳 Chef: Creates unique, constraint-aware recipes using the filtered ingredients
🎁 Presenter: Formats the recipes into clean, visually engaging cards with images and optional video links

🚧 Challenges we ran into

Creating distinct yet collaborative agent behaviors
Designing prompts that kept agents focused within their role
Ensuring structured output (JSON) from language models
Managing shared data between agents while keeping them autonomous
Building a dashboard that reflects behind-the-scenes agent actions

🏆 Accomplishments that we're proud of

A fully functioning Multi-Agent System with clear agent pipelines
Intuitive and elegant frontend to capture inputs and show results
Real-time logging of each agent's role and output
Seamless integration of AI creativity and real-world constraints
The joy of watching our system turn a handful of fridge items into delicious meals!

📚 What we learned

Agent-based thinking can solve real-world problems beautifully
Prompt clarity and tool design are critical for LLM-based systems
CrewAI is powerful for orchestrating tasks across intelligent agents
Creating a smart system isn’t just about output, it’s about how decisions are made