GravityOS: AI Marketing Agent Pipeline

GravityOS is an advanced, AI-driven marketing content generation pipeline built with LangGraph. It automates the creation of high-converting, platform-specific marketing content (Instagram carousels, LinkedIn posts, and Twitter threads/tweets) from simple text briefs or website URLs.

The system uses a multi-agent architecture to extract brand DNA, formulate a strategy, generate content drafts, critique the drafts against quality thresholds, and recursively revise them until they pass the standards.

🚀 Features

• Multi-Platform Support: Generates tailored content for:
  • Instagram: Creates highly engaging carousel concepts, writes the copy for each slide, and generates image prompts ready for models like Flux.
  • LinkedIn: Generates professional, thought-leadership style posts with strong hooks and clear CTAs.
  • Twitter: Allocates strategic angles and creates engaging tweets or threads.
• Website Context Integration: Can scrape and use the content of any provided URL as context to generate accurate and on-brand marketing materials.
• Multi-Agent Evaluation Loop: Drafts are evaluated by a Critic Agent. If the content does not score above an 8.0/10 threshold, a Revision Agent iterates and improves the content up to 3 times before finalizing.
• Brand DNA Extraction: Automatically extracts the brand's archetype, core audience persona, and emotional territory from the brief.

🧠 Architecture Overview

The pipeline utilizes the following core agents:

1. Brand DNA Builder: Analyzes the input brief/URL to define the brand voice, audience, and emotional tension. (Shared across all platforms)
2. Strategy Agent: Brainstorms multiple strategic angles and concepts.
3. Content Generator: Writes the actual copy for the selected concepts.
4. Critic Agent: Evaluates the drafts based on engagement potential, clarity, and brand alignment. Picks a winner and assigns a score.
5. Revision Node: Iterates on the winning concept if it falls below the quality threshold.
6. Image Prompt Generator: (Instagram only) Generates detailed visual prompts for each slide of the carousel.

🛠️ Setup

1. Clone the repository:

   git clone https://github.com/silverbolt08/GravityOS.git
   cd GravityOS

2. Install dependencies: Ensure you have Python 3.10+ installed. Install the required packages (e.g., langchain, langgraph, openai, python-dotenv).

3. Configure Environment Variables: Copy the example environment file and add your API keys:

   cp marketing-agent/.env.example marketing-agent/.env

   Add your OpenAI API key or Gemini API key (depending on your LLM backend configuration) to the .env file.

💻 Usage

Run the main pipeline script with your desired brief, platform, and optional website URL.

cd marketing-agent
python main.py "Your brand brief here" [quantity] [--platform <instagram|linkedin|twitter>] [--url <website_url>]

Examples

Instagram Carousel (Default):

python main.py "We are StrideLab, a DTC footwear brand..." 6

(Generates a 6-slide Instagram carousel)

LinkedIn Post with Website Context:

python main.py "Launch post for our new analytics dashboard" --platform linkedin --url https://example.com/product

Twitter Tweets:

python main.py "Announcing our Series A funding" 3 --platform twitter

(Generates 3 tweets based on the brief)

📝 Output

The pipeline will print out a detailed log of the process, including:

• Extracted Brand DNA
• Generated Concepts
• Critic Evaluations & Scores
• The final polished copy
• For Instagram: Image prompts for your design team or AI image generator

If the content fails to reach the quality threshold after maximum revisions, it will be flagged for human review.