🤖 Multi-AI Agent Blog Generator

An intelligent multi-agent system that researches, writes, reviews, and publishes tech blog posts automatically

🚀 Features • 🏗️ Architecture • ⚡ Quick Start • 📱 Screenshots

---

🌟 Overview

This project showcases a production-ready multi-agent AI system built with LangGraph that automates the entire blog creation workflow. From research to publication on Dev.to, four specialized AI agents collaborate seamlessly to produce high-quality, humanized tech content.

Built by: Rcids | Live Example: My Dev.to Profile

✨ Features

🔍 Intelligent Research Agent

• Powered by DuckDuckGo Search API
• Gathers latest tech news and insights
• Contextual information extraction

✍️ Human-like Writer Agent

• Uses Llama 3.1 via Ollama
• Conversational, engaging tone
• Avoids robotic AI patterns

🧐 Quality Reviewer Agent

• Multi-criteria evaluation
• Detects AI-like language
• Iterative improvement loop (max 3 revisions)

🚀 Auto-Publisher Agent

• One-click publishing to Dev.to
• Draft mode for safety
• Manual override option

🎨 Modern Flutter UI

• Animated process visualization
• Dark theme with gradients
• Real-time stage tracking
• Responsive design (Windows/Web/Mobile)

🏗️ Architecture

graph LR
    A[User Input] --> B[Researcher Agent]
    B --> C[Writer Agent]
    C --> D[Reviewer Agent]
    D -->|Needs Work| C
    D -->|Approved| E[Publisher Agent]
    E --> F[Dev.to Draft]
    F --> G[Manual Publish]

Loading

Tech Stack:

• Backend: FastAPI, LangGraph, LangChain
• LLM: Llama 3.1 (via Ollama)
• Frontend: Flutter (Windows/Web/Mobile)
• Publishing: Dev.to API

⚡ Quick Start

Prerequisites

• Python 3.8+
• Flutter SDK
• Ollama with Llama 3.1
• Dev.to API Key

🐍 Backend Setup

1. Clone the repository

git clone https://github.com/Rcidshacker/Multi-AI-Agent.git
cd Multi-AI-Agent

2. Create virtual environment

python -m venv ai_agents_env
# Windows
.\ai_agents_env\Scripts\activate
# macOS/Linux
source ai_agents_env/bin/activate

3. Install dependencies

pip install -r requirements.txt

4. Install Ollama and pull Llama 3.1

# Install from https://ollama.ai
ollama pull llama3.1:latest

5. Get Dev.to API Key

• Go to Dev.to Settings
• Generate API Key
• Update publisher.py with your key

6. Run the server

python -m uvicorn server:app --host 127.0.0.1 --port 8000

📱 Flutter App Setup

1. Navigate to app directory

cd ai_agent_app

2. Get dependencies

flutter pub get

3. Run the app

# For Windows
flutter run -d windows

# For Web
flutter run -d chrome

# For Android (using emulator)
flutter run

📱 Screenshots

Main Interface

Clean, modern UI with gradient backgrounds and smooth animations

Workflow Visualization

The app displays real-time progress through 4 stages:

1. 🔎 Research - Gathering information
2. ✍️ Write - Generating content
3. 🧐 Review - Quality check
4. 🚀 Publish - Sending to Dev.to

🎯 How It Works

1. Enter a topic in the Flutter app (e.g., "The Future of Quantum Computing")
2. Click "Ignite Agents" to start the workflow
3. Watch the magic:
   • Researcher finds latest info
   • Writer creates engaging draft
   • Reviewer checks quality (loops if needed)
   • Publisher sends to Dev.to
4. Manual override: Click "Publish to Dev.to" if you want to publish even if reviewer requested changes

🛠️ Project Structure

Multi-AI-Agent/
├── ai_agent_app/          # Flutter frontend
│   ├── lib/
│   │   ├── main.dart      # Main app
│   │   └── widgets/       # Custom widgets
│   └── pubspec.yaml
├── main.py                # Core agent logic
├── server.py              # FastAPI backend
├── publisher.py           # Dev.to publishing
└── requirements.txt       # Python dependencies

🔧 Configuration

API Keys

Update publisher.py with your Dev.to API key:

API_KEY = "your_dev_to_api_key_here"

Backend URL (for mobile)

In ai_agent_app/lib/main.dart, update for Android emulator:

String baseUrl = Platform.isAndroid 
    ? "http://10.0.2.2:8000"  // Android emulator
    : "http://127.0.0.1:8000"; // Desktop/Web

🚀 Features in Action

Humanized Content

The Writer Agent is specifically prompted to:

• Use conversational language
• Avoid AI clichés like "Moreover" and "In conclusion"
• Include rhetorical questions
• Vary sentence structure
• Show personality

Quality Control

The Reviewer Agent checks for:

• Comprehensiveness
• Human-like tone
• Robotic transitions (flags them)
• Varied sentence structure

Auto-Publishing

• Posts are created as drafts by default
• Manual publish button for override
• URL displayed after successful publish

🤝 Contributing

Contributions are welcome! This project demonstrates:

• ✅ Multi-agent orchestration with LangGraph
• ✅ State management in AI workflows
• ✅ RESTful API design with FastAPI
• ✅ Modern Flutter UI development
• ✅ External API integration (Dev.to)
• ✅ Conditional workflow routing

📄 License

MIT License - feel free to use this project as a portfolio piece or learning resource!

🌐 Connect

Author: Rcids
Dev.to: https://dev.to/rcids
GitHub: https://github.com/Rcidshacker

---

⭐ Star this repo if you find it helpful!

Made with ❤️ using LangGraph, Llama 3.1, and Flutter