CrewAI Multi-Agent System for Writing Article from YouTube Movies

Introduction

Within the period of an evolving digital world, automation redefines how people strategy on a regular basis duties, together with analysis, content material creation and far more. A brand new and thrilling device rising on this discipline is CrewAI, which allows a number of AI brokers to work collectively, fixing complicated issues with enhanced creativity and decision-making by prioritising collaborative intelligence.

On this article, we are going to dive deep into understanding the entire CrewAI multi-agent system and the way this method operates whereas creating a challenge to write down articles from YouTube movies. This hands-on challenge will enable you to achieve helpful insights and leverage AI to generate and streamline content material creation whereas sustaining a human contact. Let’s get began.

Venture Goal

Image this: You might be a part of Analytics Vidhya, a platform identified for its in depth library of articles on Knowledge Science, Machine Studying, AI, and extra. Whilst you’re probably aware of these assets, Analytics Vidhya additionally produces a variety of YouTube content material: informative movies, podcasts with AI leaders, information updates, and extra. Isn’t it an important thought to have articles summarising the content material from these movies as properly? Nevertheless, the method of manually writing such articles is resource-intensive and sophisticated. It requires a website skilled first to observe and validate the movies, adopted by a content material author to construction and draft the articles. Think about scaling this course of for 1000’s of movies: communication breakdowns, delays, and potential errors grow to be inevitable.

To beat this, AI brokers can streamline your complete workflow. These brokers can effectively analyze video content material, validate it, and generate articles—quick, dependable, and with minimal human intervention. Let’s discover how this may be achieved utilizing AI!

Additionally learn: Construct an AI Analysis Assistant Utilizing CrewAI and Composio

What’s an AI Agent?

An agent is a software program system that understands its setting, processes knowledge and takes actions to attain its aims. It doesn’t all the time have to be clever; as an example, a movement sensor mild acts as an agent by routinely turning on when it detects motion.

Alternatively, an AI Agent is a extra superior type of an agent that includes Synthetic intelligence particularly designed to automate duties, enabling it to function with out direct human intervention. The AI Agent additionally capabilities by observing its setting, processing the acquired data, and executing actions or making selections based mostly on predefined aims. Nonetheless, it additionally learns from its setting and improves its decision-making over time. AI Brokers are engineered to copy human-like decision-making and deal with extra complicated duties by studying from knowledge, figuring out patterns, and deriving insights or making predictions to perform assigned duties effectively.

You possibly can understand it as an clever assistant on which you don’t must have fixed supervision. For instance, an AI agent can course of massive quantities of information, deal with buyer queries, do analysis duties, write blogs, and create content material. Essentially the most fascinating factor about these brokers is that they will work constantly with none breaks, analyze bigger quantities of information than people, and simply adapt to altering environments and conditions based mostly on the outlined guidelines or ML algorithms they’re designed with. By combining a number of AI brokers, as we are going to do on this article, the system’s collective intelligence will increase considerably, permitting it to deal with much more complicated duties easily and successfully.

If you wish to know in regards to the Agentic AI design sample then discover this: 4 Agentic AI design sample

What’s CrewAI?

CrewAI is a brand new, rising, and revolutionary platform designed to allow the collaboration of a number of AI brokers effectively and seamlessly (CrewAI Multi-Agent System). With its framework, it allows customers to make use of the facility of collective intelligence, permitting AI brokers to work collectively on complicated duties, which may very well be difficult and costly for a single agent to deal with. With CrewAI you may create your system with specialised brokers, every with its personal experience, to carry out varied duties. This collaborative strategy enhances decision-making and creativity and likewise streamlines workflows, making it simpler to attain desired outcomes.

Additionally learn: Constructing an Agentic Workflow with CrewAI and Groq

How are we utilizing Agent for this Venture?

The core idea of this challenge includes choosing a YouTube video, extracting all related content material from it, processing that data, and writing a structured article based mostly on the perceived content material.

To create a a number of AI-Agent system, we typically divide it into three parts: Brokers, Duties, and Instruments.

• Brokers: Clever entities designed to carry out particular roles throughout the system. Every agent possesses distinctive capabilities, permitting for environment friendly dealing with of assorted elements of the challenge.
• Duties: Clearly outlined aims assigned to brokers. These duties be certain that every agent understands its obligations, optimizing workflow and productiveness.
• Instruments: Assets and applied sciences that help brokers in finishing their duties. These might embrace software program functions, knowledge sources, APIs, and custom-designed options aimed toward enhancing the brokers’ effectivity, effectiveness, and workflow.

On this system, we are going to implement two brokers: one will function a website skilled agent, tasked with researching the YouTube video and amassing content material utilizing an assigned device, whereas the opposite agent will perform as an article content material author, liable for making a structured article based mostly on the data offered by the area skilled agent. This knowledge stream from one agent to a different is known as a sequential course of. Whereas we are going to give attention to this strategy, different processes, akin to hierarchical processes, can be mentioned in a later article.

Circulation of the Venture

Beneath is the entire stream for the challenge:

Additionally learn: How one can Construct an AI Pair Programmer with CrewAI?

Let’s Begin Writing Article from YouTube Movies

To construct CrewAI Multi-Agent System, set up Python in your system. It’s really useful that you just use Anaconda or Miniconda to handle environments. 

I’m creating this on a Linux system, utilizing Visible Studio Code (VS Code) as my IDE. VS Code is light-weight and versatile, helps a number of programming languages, and its built-in terminal simplifies improvement. It’s additionally a most popular alternative for a lot of builders—and truthfully, I feel it simply seems to be fairly cool. 

Now that you’ve all the required instruments let’s dive into the article and get began!

Step 01: Setup the Atmosphere and Necessities

On this step, we’ll begin by creating the necessities.txt file to specify all of the libraries obligatory for the challenge. After that, we’ll arrange the .env file, which can maintain setting variables akin to API keys and different confidential data. Lastly, we are going to create a digital setting. This can be a finest observe in improvement as a result of it locks the variations of the libraries and applied sciences used, stopping future errors attributable to model mismatches.

1. Create a Folder
Create a folder anyplace in your system, for instance, identify it Multi_AI_Agent_Article_Project.

2. Open in VS Code
Open VS Code, then open the folder you simply created within the VS Code Explorer.

3. Create necessities.txt
Contained in the folder, create a file named necessities.txt and paste the next textual content inside it, then save the file:

crewai
crewai_tools
load_dotenv
langchain-huggingface

4. Generate OpenAI API Key
For this challenge, we’ll use an LLM from OpenAI, so it’s essential generate an OpenAI API Key. You possibly can comply with the hyperlink: How one can get your personal OpenAI API Key to get directions. The price of executing this challenge is round $0.04, making it fairly cheap. If you have already got an API key, be at liberty to proceed to the subsequent step.

5. Create .env File
Create one other file named .env and save your OpenAI API key on this format, then save the file:

OPENAI_API_KEY= paste_your_api_key_here

6. Create Digital Atmosphere
Open the terminal in VS Code, making certain it’s opened contained in the Multi_AI_Agent_Article_Project folder. Then sort the next command to create a digital setting:

conda create -p venv python=3.10

This command will create a Conda digital setting within the venv listing throughout the challenge folder as an alternative of the default Conda listing. The -p flag specifies the trail the place the setting can be created (on this case, the venv folder) and the Python model is about to three.10.

7. Activate Digital Atmosphere
Use this command to activate the digital setting:

conda activate venv/

8. Set up Required Libraries
To put in all the required libraries from the necessities.txt file, run the next command:

pip set up -r necessities.txt

As soon as all of the steps above are full, you’re able to proceed with constructing the precise Multi AI Agent system!

Step 02: Create Instruments

Create a file named instruments.py and paste the code under, then save the file:

from crewai_tools import YoutubeChannelSearchTool

Code Rationalization:

Right here, we’re importing the YoutubeChannelSearchTool from CrewAI. This device permits an AI agent to go looking and collect video knowledge, titles, descriptions, and metadata from a specified YouTube channel, enabling environment friendly content material analysis and evaluation.

#Initialize the device with a selected YouTube channel deal with to focus on your search
yt_channel_tool = YoutubeChannelSearchTool(youtube_channel_handle="@SystemDesignSchool")

Code Rationalization:

Subsequent, we initialize an object for the device and supply the YouTube deal with for the channel SystemDesignSchool. We’re utilizing this channel as a result of it has solely 6 movies, which helps save time and assets when the system processes the movies.

You possibly can study extra about this device from the next hyperlink.

Step 03: Create Brokers

Now, create one other file named brokers.py and paste the next code into it, then save the file:

from crewai import Agent, LLM
from instruments import yt_channel_tool
import os

from dotenv import load_dotenv
load_dotenv() # Load setting variables from the .env file

#Organising setting variables for OpenAI API entry
os.environ["OPENAI_API_KEY"] = os.getenv("OPENAI_API_KEY")
os.environ["OPENAI_MODEL_NAME"] = "gpt-4-0125-preview"

Code Rationalization:

Right here, we import the required modules to create AI brokers and configure setting variables. We use dotenv to load the OPENAI_API_KEY from the .env file we created earlier. We additionally set the mannequin identify to gpt-0125-preview.

# Creating the Area Skilled Agent, specialised in extracting related content material from YouTube movies
domain_expert = Agent(
	function="Area Skilled content material researcher from Youtube Movies",
	aim="To get all of the related Content material from the Youtube Video for the subject: {subject}",
	verbose=True,
	reminiscence=True,
	backstory=(
    	'Area Skilled in understanding and greedy movies from Youtube that are about Synthetic Intelligence,'
    	'Knowledge Science, Machine Studying, Deep Studying, and Generative AI and offering content material.'
	),
	device = [yt_channel_tool],
	allow_delegation=True
)

# Creating the Content material Author Agent, liable for writing a technical article based mostly on YouTube video content material
content_writer = Agent(
	function="Article Author",
	aim="Write a structured and informative technical article in regards to the Youtube Video subject: {subject}",
	verbose=True,
	reminiscence=True,
	backstory=(
    	'Expert at breaking down complicated technical content material into clear, partaking tales that educate readers, whereas presenting discoveries in an accessible and easy-to-understand format.'
	),
	device = [yt_channel_tool],
	allow_delegation=False
)

Code Rationalization:

Right here, we create two brokers. The Area Skilled agent is designed to extract related content material from YouTube movies on matters like AI and Machine Studying utilizing the yt_channel_tool. The Content material Author agent focuses on writing structured, technical articles based mostly on the offered content material.

Each brokers have reminiscence=True, permitting them to retain data and recollect it when wanted. verbose=True allows detailed logging to assist perceive the agent’s actions. allow_delegation=True for the Area Skilled agent permits process delegation, whereas it’s set to False for the Content material Author agent, as the author doesn’t must delegate duties. Defining their roles, objectives, and backstory ensures that the brokers give attention to their particular duties effectively.

You possibly can study extra about brokers from the next hyperlink.

Step 04: Create Duties

Create one other file named duties.py and paste the next code into it, then save the file:

from crewai import Activity
from instruments import yt_channel_tool
from brokers import domain_expert, content_writer

Code Rationalization:

This code imports the Activity class from CrewAI for creating duties, the yt_channel_tool from the instruments.py file, and the domain_expert and content_writer brokers from the brokers.py file, which can carry out particular roles within the duties.

# Area Skilled - ReseaAIrch Activity
research_task = Activity(
	description=(
    	"Get and establish the Youtube Video {subject}."
    	"Get a completely detailed, skilled stage content material/details about the video from the channel."
	),
	expected_output="An in depth 6 paragraphs lengthy structured report based mostly on the {subject} of video content material.",
	instruments = [yt_channel_tool],
	agent=domain_expert
)


# Content material Author - Article Writing Activity
article_writing_task = Activity(
	description=(
    	"Get content material/data from the Youtube Channel's Video on the subject: {subject}."
	),
	expected_output="Summarize the content material kind the Youtube Channel and write an in depth lengthy structured Article from the Youtube Chanel Video based mostly on the {subject}",
	instruments = [yt_channel_tool],
	agent=content_writer,
	async_execution=False,
	output_file="new-article.md"
)

Code Rationalization:

This part defines two duties: the area skilled analysis process and the content material author’s article writing process. The analysis process focuses on figuring out a YouTube video on the required subject from the given channel and gathering detailed data to create a structured report. The article writing process summarizes the video’s content material and creates a structured article based mostly on that data. Each brokers are assigned the yt_channel_tool to entry YouTube knowledge. Within the article_writing_task, async_execution=False ensures that the duty is accomplished absolutely earlier than transferring to the subsequent step, stopping parallel execution of duties, and the output can be saved within the “new-article.md” file.

You possibly can study extra about duties from the next hyperlink.

Step 05: Execution of the Code

Now, as the ultimate step, create one other file named crew.py and paste the next code into it, then save the file:

from crewai import Crew,Course of
from brokers import domain_expert, content_writer
from duties import research_task, article_writing_task

Code Rationalization:

This code imports the Crew and Course of courses from crewai to handle a number of brokers together with the outlined brokers and their duties created in earlier steps, enabling them to work collectively on the challenge.

# Making a Crew occasion with specified brokers, duties, and configuration
crew = Crew(
	brokers=[domain_expert, content_writer],
	duties=[research_task, article_writing_task],
	course of=Course of.sequential,
	reminiscence=True,
	cache=True,
	max_rpm=100,
	share_crew=True
)

# Kicking Off the Activity
consequence = crew.kickoff(inputs={'subject': 'File Storage VS Object Storage'})
print(consequence)

Code Rationalization:

This code creates a Crew occasion that mixes the 2 brokers, the Area Skilled and the Content material Author, with their particular duties. We now have outlined the method as sequential, that means the duties will run one after the opposite. The reminiscence and cache settings are each set to True to recollect previous data and enhance effectivity. The max_rpm parameter units the utmost variety of requests the crew can deal with per minute, making certain environment friendly processing with out overwhelming the system. The share_crew choice allows the crew to be shared for collaboration or reuse, permitting different processes or customers to entry and make the most of the identical crew configuration for various duties.

The crew is then initiated with the required subject, which the brokers will seek for on the YouTube channel, carry out their duties, and print the consequence.

You possibly can study extra from the next hyperlink.

Remaining Output

Screenshot of the output file “new-article.md”:

To study extra about What brokers are, go to the next Article: AI Agent.

Conclusion

Utilizing CrewAI Multi-Agent system considerably improves the effectivity and high quality of content material era from YouTube movies. We optimize the workflow of the challenge by breaking it into brokers, instruments and duties which considerably reduces the effort and time required to create high-quality articles. The Area Skilled agent gathers key data from the YouTube movies, whereas the Content material Author processes this data and writes well-organized articles.

This collaborative strategy reduces the necessity for guide effort whereas making certain constant outcomes that preserve the core message of the unique content material. As automation applied sciences progress, embracing these instruments permits content material creators to focus extra on strategic and inventive elements of their work.

To remain forward on this evolving discipline of Agentic AI, enroll in our Agentic AI Pioneer Program in the present day!