Complete Marketing Team Automation Pipeline Guide

📋 Overview

This project is an AI-powered Instagram marketing automation pipeline built with CrewAI framework that simulates a complete marketing team with 4 specialized AI agents:

1. Market Researcher - Analyzes trends, competitors, and hashtags
2. Content Strategist - Creates content calendars and strategies
3. Visual Creator - Generates detailed image descriptions
4. Copywriter - Crafts engaging captions and copy

🏗️ Project Structure

Automating-a-Marketing-Team-with-agents/
├── src/
│   └── instagram/
│       ├── __init__.py
│       ├── crew.py           # Main crew orchestration
│       ├── main.py           # Entry point
│       ├── config/
│       │   ├── agents.yaml   # AI agents configuration
│       │   └── tasks.yaml    # Tasks configuration
│       └── tools/
│           ├── __init__.py
│           └── search.py     # Search tools for research
├── .env                      # Environment variables
├── pyproject.toml           # Poetry dependencies
├── poetry.lock              # Locked dependencies
└── README.md

🚀 Step-by-Step Manual Setup and Execution

Step 1: Prerequisites

Required Software:

• Python 3.10-3.13
• Poetry (for dependency management)
• OpenAI API key

Step 2: Environment Setup

2.1 Install Poetry

pip install poetry

2.2 Navigate to Project Directory

cd C:\Users\abhis\Automating-a-Marketing-Team-with-agents

2.3 Lock and Install Dependencies

python -m poetry lock
python -m poetry install

Step 3: Configuration

3.1 Environment Variables Setup

Create/Edit .env file in project root:

OPENAI_API_KEY=your_actual_openai_api_key_here
# Optional: Add Serper API key for enhanced search
SERPER_API_KEY=your_serper_api_key_here

3.2 Agents Configuration (src/instagram/config/agents.yaml)

• Market Researcher: Analyzes Instagram trends and competitors
• Content Strategist: Develops content calendars
• Visual Creator: Creates image descriptions
• Copywriter: Writes engaging captions with SEO optimization

3.3 Tasks Configuration (src/instagram/config/tasks.yaml)

• market_research: Research trends and hashtags
• content_strategy: Create weekly content calendar
• visual_content_creation: Generate image descriptions
• copywriting: Write post captions and copy

Step 4: Pipeline Execution

4.1 Manual Execution Command

python -m poetry run instagram

4.2 Interactive Inputs Required

When you run the pipeline, you'll be prompted for:

1. Instagram Account Description:

   • Example: "A fitness coaching page focused on home workouts and healthy lifestyle tips"

2. Topic of the Week:

   • Example: "New Year fitness resolutions and motivation"

Step 5: Pipeline Flow

Phase 1: Market Research

• Agent: Market Researcher
• Action: Analyzes current trends, hashtags, competitor activities
• Output: Research report with trending hashtags and insights

Phase 2: Content Strategy

• Agent: Content Strategist
• Action: Creates 3-5 day content calendar based on research
• Output: Structured markdown calendar with themes and keywords

Phase 3: Visual Content Creation

• Agent: Visual Creator
• Action: Generates detailed image descriptions for each post
• Output: Detailed visual prompts for AI image generation

Phase 4: Copywriting

• Agent: Copywriter
• Action: Writes engaging captions with hashtags and CTAs
• Output: Ready-to-post Instagram captions

Step 6: Expected Outputs

The pipeline generates several files:

• Market Research Report (market_research.md)
• Content Calendar (integrated in final output)
• Visual Content Descriptions (visual-content.md)
• Instagram Copy (captions with hashtags)

🔧 Manual Customization Options

Modify Agents (agents.yaml)

market_researcher:
  role: "Your custom role"
  goal: "Your specific goal"
  backstory: "Agent personality and expertise"

Adjust Tasks (tasks.yaml)

market_research:
  description: "Custom task description with {variables}"
  expected_output: "Specific output format requirements"

Add Custom Tools (tools/search.py)

• Extend search capabilities
• Add new data sources
• Integrate APIs

🎯 Manual Execution Examples

Example 1: Fitness Coach Account

python -m poetry run instagram
# Inputs:
# Page description: "Personal trainer specializing in strength training for beginners"
# Topic: "January fitness motivation and goal setting"

Example 2: Food Blog

python -m poetry run instagram
# Inputs:
# Page description: "Healthy recipe blog with quick meal ideas for busy professionals"
# Topic: "Meal prep Sunday inspiration"

📊 Pipeline Performance Metrics

• Processing Time: 5-10 minutes depending on complexity
• API Calls: ~10-15 OpenAI API calls per run
• Output Quality: Professional-grade marketing content
• Customization Level: Highly configurable through YAML files

🛠️ Troubleshooting

Common Issues:

1. Poetry not recognized:

   python -m poetry [command]
   

2. API Key errors:

   • Verify .env file location and format
   • Check API key validity

3. Dependency conflicts:

   python -m poetry lock --no-update
   python -m poetry install
   

🔄 Alternative Execution Methods

Method 1: Direct Python Execution

python src/instagram/main.py

Method 2: Poetry Shell

poetry shell
python -m instagram

Method 3: Custom Script

Create run_pipeline.py:

from src.instagram.crew import InstagramCrew
import datetime

inputs = {
    'current_date': datetime.datetime.now().strftime("%Y-%m-%d"),
    'instagram_description': 'Your page description',
    'topic_of_the_week': 'Your weekly topic',
}
InstagramCrew().crew().kickoff(inputs=inputs)

🏃‍♂️ Quick Start Commands

For Windows PowerShell:

# 1. Install Poetry
pip install poetry

# 2. Navigate to project
cd C:\Users\abhis\Automating-a-Marketing-Team-with-agents

# 3. Install dependencies
python -m poetry lock
python -m poetry install

# 4. Run pipeline
python -m poetry run instagram

📁 File Locations and Purposes

Configuration Files:

• .env - Environment variables (API keys)
• pyproject.toml - Project dependencies and metadata
• src/instagram/config/agents.yaml - AI agent definitions
• src/instagram/config/tasks.yaml - Task descriptions and workflows

Core Code Files:

• src/instagram/main.py - Entry point and user interaction
• src/instagram/crew.py - CrewAI orchestration logic
• src/instagram/tools/search.py - Search and research tools

Output Files (Generated):

• market_research.md - Research findings and trends
• visual-content.md - Image descriptions for posts
• Various markdown files with content strategies and copy

🔐 Security Notes

• Never commit your .env file with real API keys
• Use environment variables for sensitive data
• Consider using .env.example for template sharing

🚀 Advanced Usage

Batch Processing:

Create multiple input configurations and run them sequentially:

# batch_run.py
from src.instagram.crew import InstagramCrew
import datetime

campaigns = [
    {
        'current_date': datetime.datetime.now().strftime("%Y-%m-%d"),
        'instagram_description': 'Fitness coaching page',
        'topic_of_the_week': 'New Year resolutions',
    },
    {
        'current_date': datetime.datetime.now().strftime("%Y-%m-%d"),
        'instagram_description': 'Travel photography blog',
        'topic_of_the_week': 'Winter destinations',
    }
]

for campaign in campaigns:
    print(f"Running campaign: {campaign['topic_of_the_week']}")
    InstagramCrew().crew().kickoff(inputs=campaign)

📈 Monitoring and Analytics

Track Pipeline Performance:

• Monitor API usage and costs
• Log execution times and success rates
• Analyze output quality and engagement metrics

Integration Possibilities:

• Connect to Instagram Business API
• Integrate with scheduling tools (Buffer, Hootsuite)
• Add analytics dashboards
• Implement A/B testing for content variations

---

This comprehensive pipeline automates the entire content creation process from research to final copy, simulating a full marketing team's workflow in minutes rather than hours!

🎯 Next Steps After Pipeline Execution

1. Review Generated Content: Check all output files for quality and relevance
2. Customize as Needed: Edit any generated content to match your specific brand voice
3. Schedule Posts: Use the content calendar to plan your posting schedule
4. Generate Images: Use the visual descriptions with AI image generators (DALL-E, Midjourney, etc.)
5. Track Performance: Monitor engagement metrics and adjust future inputs based on results