LinkedIn Post Generator Agent 🤖 [POC]

Note: This project is an educational proof of concept exploring AI-driven content generation and workflow automation using LangGraph and Ollama.

📋 Overview

An intelligent agent that transforms plain text into engaging LinkedIn posts through an iterative, feedback-driven approach. Built with LangGraph and Ollama, it demonstrates how AI can simulate a team of content experts working together.

📋 Quick Start

Prerequisites

• Python 3.11+
• Ollama installed and running
• Required Ollama model: llama3:latest

Installation

# Clone the repository
git clone [REPO_URL]
cd linkedin-agent

# Create and activate virtual environment
python -m venv venv
source venv/bin/activate  # or `venv\Scripts\activate` on Windows

# Install dependencies
pip install -e .

# Start Ollama service
ollama serve

# Pull required model
ollama pull llama3:latest

🔌 API Usage

Start the API

uvicorn src.api:app --reload

The API will be available at http://localhost:8000

Endpoints

Generate Post

curl -X POST http://localhost:8000/generate \
  -H "Content-Type: application/json" \
  -d '{
    "text": "We are launching a new AI product",
    "target_audience": "Tech leaders",
    "n_drafts": 3
  }'

Response:

{
  "final_post": "...",
  "all_versions": [
    {
      "version": 1,
      "content": "...",
      "feedback": null
    }
  ],
  "workflow_status": "completed"
}

Health Check

curl http://localhost:8000/health

Response:

{
  "status": "healthy",
  "timestamp": "2024-12-27T09:00:00.000Z",
  "version": "1.0.0",
  "environment": {
    "python_version": "3.11.0",
    "platform": "..."
  },
  "services": {
    "ollama": {
      "status": "healthy",
      "model": "llama3:latest",
      "available": true,
      "available_models": ["llama3:latest", "..."]
    }
  },
  "system": {
    "cpu_usage": 45.2,
    "memory": {
      "total": 17179869184,
      "available": 8589934592,
      "percent": 50.0
    },
    "disk": {
      "total": 499963174912,
      "free": 374972680192,
      "percent": 25.0
    }
  }
}

🐳 Docker Deployment

Quick Start with Docker

# Build and start services
docker-compose up --build

# The services will be available at:
# - API: http://localhost:8000
# - Ollama: http://localhost:11434

Features

• Self-contained environment with all dependencies
• Automatic model download during build
• Persistent model storage
• Health monitoring
• Automatic service orchestration

Environment Variables

OLLAMA_BASE_URL=http://localhost:11434

Container Management

# View logs
docker-compose logs -f

# Stop services
docker-compose down

# Remove volumes (will delete downloaded models)
docker-compose down -v

🏗 Technical Architecture

Core Components

1. API Layer (src/api.py)

   • FastAPI application
   • Health monitoring
   • Request validation

2. Workflow Engine (src/workflow.py)

   • LangGraph state management
   • Node orchestration
   • Conditional routing

3. AI Agents (src/services/linkedin_agent.py)

   • Editor: Improves text clarity
   • Writer: Creates LinkedIn-optimized content
   • Critic: Provides feedback
   • Supervisor: Manages iterations

4. Models (src/models/)

   • State management
   • Post structure
   • Configuration

Configuration

Key parameters in src/utils/constants.py:

OLLAMA_MODEL = "llama3:latest"
OLLAMA_TEMPERATURE = 0.7
MAX_POST_LENGTH = 1300
MIN_HASHTAGS = 3
MAX_HASHTAGS = 5

📄 License

This project is released under the MIT License.

⚠️ Disclaimer

This is a proof of concept for educational purposes. While it demonstrates interesting possibilities in AI-driven content generation, it should not be considered production-ready software.

📚 Resources

• LangGraph Documentation
• Ollama Documentation
• LinkedIn Content Best Practices