🚀 AutoGen Multi-Agent Workflow Tutorial

This repository contains a beginner-friendly AutoGen 0.4 tutorial demonstrating how to create an AI-powered multi-agent system with external API integrations (text-to-speech, image generation) and Ollama integration.

🎥 Watch the YouTube video here
📖 Read the blog post here

🛠️ Features

• 🤖 Multi-Agent System: Script Writer, Voice Actor, Graphic Designer, and Director agents working together.
• 🎙️ Text-to-Speech: Converts AI-generated text into voiceovers using ElevenLabs API.
• 🖼️ Image Generation: Creates AI-generated visuals using Stability AI.
• 🏡 Local LLM Support: Optional integration with Ollama for running AI models offline.
• 🎮 Interactive Console: Type prompts and watch agents generate content dynamically.

---

📂 Folder Structure

autogen-multi-agent-workflow/
│── tools.py               # Utility functions (text-to-speech, image generation)
│── main.py                # Entry point for running the workflow
│── .env                   # API keys (not included, create your own)
│── .gitignore             # 
│── requirements.txt       # Dependencies for the project
│── README.md              # Documentation

---

🚀 Quick Start

1️⃣ Clone the Repository

git clone https://github.com/gswithjeff/autogen-multi-agent-workflow.git
cd autogen-multi-agent-workflow

2️⃣ Create & Activate a Virtual Environment

For macOS/Linux:

python -m venv venv
source venv/bin/activate

For Windows:

python -m venv venv
venv\Scripts\activate

3️⃣ Install Dependencies

pip install -r requirements.txt

4️⃣ Set Up API Keys

Create a .env file and add your API keys:

OPENAI_API_KEY=your-openai-api-key
ELEVENLABS_API_KEY=your-elevenlabs-api-key
STABILITY_API_KEY=your-stability-ai-api-key

5️⃣ Create Required Accounts

Before running the workflow, you'll need to create accounts for the following services:

• ElevenLabs (Text to Speech): Sign up here
• Stability AI (Image Generation): Sign up here

These accounts provide API access for text-to-speech and image generation, which are required for the agents to function.

6️⃣ Run the Workflow

python main.py

You’ll be prompted to enter a task, and the agents will collaborate to generate a script, voiceover, and images dynamically.

---

🛠️ How It Works

1️⃣ Script Writer Agent generates structured captions.
2️⃣ Voice Actor Agent converts text to speech.
3️⃣ Graphic Designer Agent creates images based on captions.
4️⃣ Director Agent orchestrates the final output.

---

🎯 Example Usage

User Prompt:

Create a short AI-generated video about space exploration.

Generated JSON Response:

{
    "topic": "Space Exploration",
    "takeaway": "The future of space travel is closer than we think!",
    "captions": [
        "What lies beyond our galaxy?",
        "Humans are reaching new frontiers.",
        "AI is shaping space exploration.",
        "New planets are waiting to be discovered.",
        "The universe is limitless!"
    ]
}

✅ Voiceovers are generated
✅ Images are created
✅ Final video assembly is handled

---

🔧 Customization

• Use a different LLM: Swap OpenAI for Ollama to run locally. (See blog for more)
• Modify agent behaviors: Edit the system_message for each agent.
• Integrate new APIs: Extend tools.py for additional functionality.

---

🤝 Contributing

Pull requests are welcome! If you find issues or want to improve the workflow, feel free to open an issue.

---

📜 License

This project is licensed under the MIT License.
You are free to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the software,
as long as the original copyright notice and permission notice appear in all copies.

For full details, see the LICENSE file.

---

🌟 Support & Feedback

If you find this project helpful, ⭐️ star the repo and share your thoughts!