🚀 Autonomous Hacks 26 | Online Agentic Hackathon Submission

⏱️ 36-Hour Online Hackathon
20 December, 8:00 AM → 21 December, 12:00 PM

🧠 Problem Statement 6: Multi-Agent Creative Studio
Built using Agent-to-Agent (A2A) communication and iterative refinement.

🌐 Hackathon Website: https://autonomoushacks.co.in/

Multi-Agent Creative Studio 🎨🤖

A sophisticated multi-agent system that uses Agent-to-Agent (A2A) communication and iterative refinement to generate, critique, refine, and present creative ideas. Built with Google Gemini 2.5 Flash.

Now available with both CLI and Web UI interfaces!

🌟 Features

Multi-Agent Architecture

This project implements a complete creative workflow using four specialized AI agents:

💡 Idea Agent: Generates 3 unique creative concepts based on any topic
🔍 Critic Agent: Analyzes flaws, weaknesses, and opportunities in the ideas
✨ Refiner Agent: Improves ideas based on the critic's feedback
📊 Presenter Agent: Produces a final structured, professional presentation

Why Agentic?

Agent-to-Agent (A2A) Communication: Agents pass data to each other, creating a seamless workflow
Iterative Refinement: Ideas are progressively improved through multiple stages
Specialized Roles: Each agent has a specific expertise and responsibility
Autonomous Decision Making: Agents process information and make decisions independently

🛠️ Technology Stack

Language: Python 3.8+
AI Model: Google Gemini 2.5 Flash (gemini-2.5-flash)
Framework: Google ADK (Agent Development Kit)
Interfaces: CLI and Web UI (Streamlit)
Libraries:
- google-genai - Google's modern Generative AI SDK
- python-dotenv - Environment variable management
- streamlit - Interactive web interface

📋 Prerequisites

Python 3.8 or higher
Google API Key (for Gemini API access)
pip (Python package manager)

🚀 Installation

Clone the repository:

git clone https://github.com/mayankgautam-dev/multi-agent-creative-studio.git
cd multi-agent-creative-studio

Install dependencies:

pip install -r requirements.txt

Set up environment variables:

cp .env.example .env

Add your Google API Key to the .env file:

GOOGLE_API_KEY=your_google_api_key_here

💡 Get your API key: Visit Google AI Studio to obtain a free API key.

💻 Usage

Web UI (Recommended)

🎥 Watch Web UI Demo

Launch the interactive web interface:

streamlit run app.py

The app opens automatically in your browser at http://localhost:8501. Features include:

🎨 Interactive topic input
📊 Real-time progress tracking
📑 Tabbed results display
💾 One-click download

For detailed Web UI documentation, see README_UI.md.

Command Line Interface (CLI)

🎥 Watch CLI Demo

Run the creative studio with a default topic:

python main.py

Specify your own topic for creative idea generation:

python main.py "Your creative topic here"

CLI Examples

# Example 1: Mobile app idea
python main.py "A mobile app for learning languages through games"

# Example 2: Business concept
python main.py "An eco-friendly food delivery service"

# Example 3: Product idea
python main.py "Smart home device for elderly care"

💡 Quick Start: See QUICKSTART.md for a 5-minute setup guide.

📊 How It Works

Agent-to-Agent Communication Flow

┌─────────────────┐
│   Idea Agent    │ Generates 3 creative concepts
└────────┬────────┘
         │ A2A Communication
         ▼
┌─────────────────┐
│  Critic Agent   │ Analyzes flaws and opportunities
└────────┬────────┘
         │ A2A Communication
         ▼
┌─────────────────┐
│ Refiner Agent   │ Improves ideas based on feedback
└────────┬────────┘
         │ A2A Communication
         ▼
┌─────────────────┐
│Presenter Agent  │ Creates final professional output
└─────────────────┘

Workflow Steps

💡 Generation Phase:
- Idea Agent receives a topic
- Generates 3 unique creative concepts with titles, descriptions, and USPs
🔍 Analysis Phase:
- Critic Agent receives ideas from Idea Agent (A2A)
- Analyzes strengths, weaknesses, feasibility, market potential, and risks
✨ Refinement Phase:
- Refiner Agent receives critique from Critic Agent (A2A)
- Improves ideas by addressing weaknesses and enhancing strengths
📊 Presentation Phase:
- Presenter Agent receives refined ideas from Refiner Agent (A2A)
- Creates comprehensive presentation with executive summary, recommendations, and action items
- Saves output to a markdown file

📁 Output

The system generates a markdown file with the complete creative process and final recommendations:

Filename format: creative_studio_output_YYYYMMDD_HHMMSS.md
Contents:
- Executive Summary
- Creative Process Overview
- Final Recommendations (Top 3 Ideas)
- Comparison Analysis
- Next Steps and Action Items

🏗️ Project Structure

multi-agent-creative-studio/
│
├── main.py                 # Main orchestration script
├── app.py                  # Streamlit web UI
├── idea_agent.py          # Idea generation agent
├── critic_agent. py        # Critique and analysis agent
├── refiner_agent. py       # Idea refinement agent
├── presenter_agent.py     # Final presentation agent
├── requirements.txt       # Python dependencies
├── .env.example          # Environment variables template
├── .gitignore            # Git ignore rules
├── .streamlit/            # Streamlit configuration
│   └── config.toml       # UI theme settings
├── README.md             # This file
├── README_UI.md          # Web UI documentation
├── QUICKSTART. md         # Quick start guide
├── ARCHITECTURE.md       # Technical architecture
├── SAMPLE_OUTPUT.md      # Example outputs
├── test_structure.py     # Structure validation tests
└── test_ui.py            # UI tests

🔧 Configuration

Model Selection

By default, the system uses gemini-2.5-flash. To use a different model, modify the model_name parameter in main.py:

studio = CreativeStudio(api_key, model_name="gemini-2.5-flash")

Customizing Agents

Each agent is modular and can be customized independently:

Modify prompts in the respective agent files
Adjust response formats
Add additional analysis criteria

🎯 Use Cases

Product Development: Generate and refine product ideas
Marketing Campaigns: Create campaign concepts
Business Strategy: Develop business ideas and strategies
Content Creation: Generate content themes and approaches
Innovation Workshops: Facilitate brainstorming sessions
Startup Ideas: Validate and improve startup concepts

📚 Documentation

README_UI.md - Complete Web UI guide with features and troubleshooting
QUICKSTART. md - Get started in 5 minutes
ARCHITECTURE. md - Technical architecture and A2A communication details
SAMPLE_OUTPUT.md - Example outputs and use cases

🤝 Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

📝 License

This project is open source and available under the MIT License.

🙏 Acknowledgments

Built with Google Gemini 2.5 Flash
Powered by Google ADK (Agent Development Kit)
Inspired by modern agentic AI architectures

📧 Contact

For questions or feedback, please open an issue on GitHub.

Made with ❤️ by InnoV8 Team

Multi Agent Creative Studio

Getting Started

README