​
Agents Overview

The Newsletter AI Agent uses a multi-agent system powered by CrewAI to generate newsletters. This system consists of three specialized agents that work together to research, write, and edit the newsletter content.

​
Agent Architecture

The agent system is implemented using CrewAI, which provides a framework for creating and coordinating multiple agents. Each agent has a specific role, goal, and set of tools that it can use to accomplish its tasks.

​
Integration with Apify

The agents use tools that leverage Apify actors for web scraping and data collection. Apify provides a platform for running web scraping and automation tasks at scale, which allows the Newsletter AI Agent to gather information from various sources efficiently. The integration with Apify is implemented through a base class in src/tools/base.py that provides a standardized way to call Apify actors and process their results: 
class RunApifyActor:
    """Run an Apify actor and return the results."""
    def __init__(self, actor):
        self.actor = actor

    def _run(self, actor_name, run_input):
        # Implementation to run the Apify actor and return results
        # ...

​
Agent Roles

The Newsletter AI Agent consists of three specialized agents:

​
1. Researcher Agent

The Researcher Agent is responsible for gathering information about the specified topic. It uses a set of tools to search the web, collect social media posts, find relevant videos, and gather news articles. Tools:

​
2. Writer Agent

The Writer Agent is responsible for transforming the research data into engaging newsletter content. It takes the information gathered by the Researcher Agent and creates a well-structured newsletter draft. Capabilities:
  • Content creation from research data
  • Structuring the newsletter into sections
  • Formatting the content in markdown
  • Adding links to sources

​
3. Editor Agent

The Editor Agent is responsible for reviewing and improving the newsletter draft. It ensures that the content is polished, error-free, and ready for distribution. Capabilities:
  • Grammar and spelling correction
  • Style and tone consistency
  • Fact-checking
  • Formatting and layout improvement

​
Agent Workflow

The agents work together in a sequential workflow to generate the newsletter:
  1. The Researcher Agent gathers information about the specified topic
  2. The Writer Agent transforms the research data into a newsletter draft
  3. The Editor Agent reviews and improves the draft to create the final newsletter
This workflow is orchestrated by the NewsletterCrew class in src/newsletter_crew.py, which creates the agents, defines their tasks, and manages the flow of information between them.

​
Agent Implementation

Each agent is implemented as a class in the src/agents directory:
  • src/agents/researcher.py: Implements the Researcher Agent
  • src/agents/writer.py: Implements the Writer Agent
  • src/agents/editor.py: Implements the Editor Agent
The agents are created using CrewAI’s Agent class, which provides a framework for defining an agent’s role, goal, backstory, and tools.

​
Configuration

The agent system can be configured through environment variables and configuration files:
  • Environment Variables: Set API keys and other credentials in the .env file
  • Configuration Files: Modify agent behavior in the src/config directory

​
Next Steps