RS Chandras Tech Blog | AI, ML, Agentic AI

CrewAI has a sophisticated default project structure containing multiple folders and files. Instead of simplifying everything into a single notebook cell structure, this tutorial demonstrates how to run the official CrewAI Quick Start project exactly as intended — but inside Google Colab instead of a local machine.

Important Context

In my previous blog, I deliberately converted the CrewAI Quick Start into a single Colab notebook implementation to simplify the learning process.

This tutorial takes the opposite approach:

Use the official multi-file CrewAI project structure
Run it directly inside Google Colab
Store the project permanently inside Google Drive

[Cell 001] Mount Google Drive

CrewAI automatically creates folders, files, configuration YAMLs, crews, flows, and package structures. Therefore, mounting Google Drive is highly recommended.

# Mount google drive because CrewAI creates

# multiple folders and files automatically

from google.colab import drive

drive.mount('/content/drive')

Why Mount Drive?

Project files remain persistent
No data loss after Colab runtime reset
Allows CrewAI to maintain full project structure

[Cell 002] Set the Project Directory

#set project directory and cd to it

import os

PROJECT_PATH = "/content/drive/MyDrive/2026-Projects/CREWAI-QUICKSTART-MULTI-FILES"

os.chdir(PROJECT_PATH)

print("Current directory:")

print(os.getcwd())

os.listdir()

[Cell 003] Configure API Keys

Save your API keys inside Google Colab Secrets.

from google.colab import userdata

%env GEMINI_API_KEY={userdata.get('GEMINI_API_KEY')}

%env SERPER_API_KEY={userdata.get('SERPER_API_KEY')}

#%env MODEL=gemini/gemini-3.1-flash-lite

%env MODEL=gemini/gemma-4-26b-a4b-it

Required Keys:

Google Gemini API Key
Serper API Key

Both services currently provide free usage tiers.

[Cell 004] Install CrewAI

# install crewai

!uv tool install crewai

Note:
CrewAI officially recommends using uv for dependency and project management.

[Cell 005] Create a New Flow Project

#create new flow project

!uv tool run crewai create flow latest-ai-flow

What This Creates

Flow architecture
Agents folder
Tasks folder
YAML configuration files
Source code structure
Runnable CrewAI project template

[Cell 006] Move into the Newly Created Project

# cd to the newly created project

import os

PROJECT_PATH = "/content/drive/MyDrive/2026-Projects/CREWAI-QUICKSTART-MULTI-FILES/latest_ai_flow"

os.chdir(PROJECT_PATH)

print("Current directory:")

print(os.getcwd())

os.listdir()

[Cell 007] Replace the Default Files

The generated CrewAI project already contains a complete multi-agent workflow with:

Planner Agent
Writer Agent
Editor Agent

The project is fully runnable out-of-the-box.

However, to remain faithful to the official CrewAI Quick Start tutorial, replace the contents of the following files:

agents.yaml
tasks.yaml
content_crew.py
main.py

Replace them using the content from the official CrewAI Quick Start documentation.

After replacing the files, continue to the next cell.

[Cell 008] Install Google GenAI Support

# install google-genai

# this may take good time (maybe 5-7 mins)

!uv add "crewai[google-genai]"

Installation Warning

Dependency installation may take several minutes because CrewAI installs additional AI provider integrations and related packages.

[Cell 009] Run the CrewAI Project

!uv tool run crewai run

# this may uninstall and install some packages automatically

Execution Result

Successful Execution Flow

Flow execution started successfully
ResearchCrew initialized properly
AI research agent activated
Serper web-search tool executed successfully
Live internet search results were retrieved
Comprehensive AI Agent report generated automatically
Flow completed successfully

What Happened Internally?

CrewAI created and orchestrated the research workflow
The agent invoked the Serper web-search tool
Current AI trends were fetched from the internet
The LLM synthesized the information into a structured report
The final markdown report was returned through the flow

Observations

The official CrewAI project structure works perfectly in Google Colab
Google Drive acts like a persistent development environment
CrewAI automatically handles complex multi-file architecture
The uv package manager simplifies setup considerably
Colab can function as a cloud-native CrewAI development workstation

Difference from the Previous Tutorial

Previous Tutorial	This Tutorial
Single notebook implementation	Official multi-file architecture
No YAML files	Uses YAML configs
Beginner-friendly simplification	Production-style project structure
Portable single-cell setup	Full CrewAI ecosystem workflow

==========================================================================

OUTPUT