Agentic Content Marketing Assistant

Capstone-grade content marketing assistant built with LangChain and Streamlit. Generates research summaries, SEO blogs, LinkedIn posts, and images while keeping brand voice consistent.

Agents

• intent_agent: classify whether the user wants a blog, LinkedIn post, or both.
• topic_and_sections_agent: extract topic and section outline directly from the prompt.
• topic_and_section_generator_agent: generate a topic/outline from stored research metadata.
• research_agent: Perplexity Sonar research with citations and structured outputs.
• guard_agent: ensure new prompts stay relevant to the existing research thread.
• content_orchestrator_agent: pull Pinecone vector context and normalize state fields.
• blog_agent: draft long-form blog content using brand voice and vector context.
• linkedin_agent: generate LinkedIn posts and optional carousel copy.
• image_agent: craft prompts and generate images via OpenAI.
• title_agent: generate concise titles from research summaries.

Architecture Diagram

User
  |
  v
Streamlit UI  <--------------------->  MongoDB (projects, chats, research)
  |
  v
Content Graph (routing + state)
  |
  +--> Research Agent -----> Perplexity API
  |
  +--> Content Orchestrator -----> Pinecone Vector DB
  |                                   |
  |                                   v
  +--> Blog/LinkedIn Agents -----> LLM Provider (OpenAI/Anthropic/Gemini)
  |
  +--> Image Agent -----------> OpenAI Images API

Setup

1. Install dependencies

   cd content_marketing_agent
   python -m venv .venv
   .venv\Scripts\activate  # Windows
   pip install -r requirements.txt

2. Create a .env file

   copy .env.example .env

3. Configure required keys
   • OPENAI_API_KEY for LLM + image generation.
   • PERPLEXITY_API_KEY for research.
   • PINECONE_API_KEY for vector search.
   • GOOGLE_API_KEY for generating image prompts.
   • MONGO_URI / MONGO_DB_NAME for persistence (local dev defaults are fine).
4. Provision Pinecone
   • Create a Pinecone project and get the API key.
   • Set PINECONE_INDEX_NAME (default: content-blitz).
   • Optional: set PINECONE_CLOUD and PINECONE_REGION for serverless.
5. Start MongoDB (local dev)

   docker-compose up -d mongo

6. Run Streamlit

   streamlit run content_marketing_agent/app.py

Notes

• Switch providers with LLM_PROVIDER=openai|anthropic|gemini.
• Pinecone indexes are created automatically if missing and credentials are valid.