Module 8 Wrap-up: The Power of the Ensemble

You have learned how to design roles, tasks, and processes. Now, we are going to build a Full Multi-Agent Crew. We will create a "Researcher" and a "Technical Writer" that work together to analyze a website.

---

Hands-on Exercise: The Tech Trend Crew

1. Requirements

pip install crewai

2. The Code

Create a file tech_crew.py:

from crewai import Agent, Task, Crew, Process

# 1. Define Agents (The Identity)
researcher = Agent(
  role='Tech Breakthrough Researcher',
  goal='Discover 3 key developments in Quantum Computing for 2024',
  backstory="""You are a deep-tech enthusiast with a knack for spotting 
  academic papers that are about to go mainstream. You are analytical.""",
  verbose=True
)

writer = Agent(
  role='Tech Journalist',
  goal='Summarize the researcher findings into a LinkedIn post',
  backstory="""You are a pro at making complex tech sound exciting for 
  a general audience. Your tone is professional but engaging.""",
  verbose=True
)

# 2. Define Tasks (The Assignment)
task1 = Task(
  description='Perform a deep dive into 2024 Quantum Computing papers.',
  expected_output='A summary of 3 key breakthroughs with their potential impact.',
  agent=researcher
)

task2 = Task(
  description='Turn the breakthroughs into a 3-paragraph LinkedIn post.',
  expected_output='A post containing 3 paragraphs and 3 relevant hashtags.',
  agent=writer
)

# 3. Create the Crew (The Team)
my_crew = Crew(
  agents=[researcher, writer],
  tasks=[task1, task2],
  process=Process.sequential # A -> B
)

# 4. Kickoff
result = my_crew.kickoff()
print("######################")
print(result)

3. Observe the Collaboration

When you run this, you will see the logs showing:

1. Researcher and its internal monologue searching for Quantum info.
2. Writer taking the researcher's output and "re-styling" it.
3. The Result: A high-quality post that is better than what a single "Generic" agent would produce.

---

Module 8 Summary

• Multi-Agent Systems use specialized personas to improve quality.
• CrewAI manages the "Baton Passing" between these agents automatically.
• Role, Goal, and Backstory are the primary way we "Program" the behavior of a crew.
• Sequential Processes are easy to build, while Hierarchical Processes allow for more complex management.

Coming Up Next...

In Module 9, we explore a different philosophy: StrandAgents. We will learn about Event-Driven AI and how to build agents designed for streaming and asynchronous real-time behavior.

---

Module 8 Checklist

• I have installed the crewai library.
• I can explain the difference between a Role and a Goal.
• I have successfully run the tech_crew.py script.
• I understand how the output of Task 1 moves to Task 2.
• I can identify a task in my life that would benefit from a "Manager" agent.