Session 0: Experience the Power of AI Agents

What We’re Doing Today

Today, you’ll get hands-on experience with an “AI agent.” No complicated steps are involved. Everything can be done from your web browser.

How It Works

You write a task → The AI agent executes it automatically → You review the deliverable
 (GitHub Issue)          (Claude Code)                       (Pull Request)

Step 1: Access the GitHub Repository

Open the following URL in your web browser (the instructor will provide it)
If you’re not logged in to GitHub, please log in

https://github.com/(URL provided by the instructor)

Step 2: Assign a Task to the AI Agent

Click the “Issues” tab at the top of the repository page
Click the “New issue” button in the upper right
Choose one of the templates below and customize it to fit your own work

Task Templates

A: Industry Research

Title: Latest Trend Research for the XX Industry

Body:
Identify the three latest trends in the XX industry
and summarize each with an overview and business impact in approximately 500 words.

Target audience: Division managers

B: Competitive Analysis

Title: Service Comparison of Company XX and Company YY

Body:
Compare the services of Company XX and Company YY
and create a comparison table covering features, pricing, and target customers.

Target audience: Sales team

C: Proposal Draft

Title: Proposal for an Internal Study Session on XX

Body:
Create a draft proposal for an internal study session on XX.
Please include the following:
- Purpose
- Target audience
- Proposed agenda (assuming 60 minutes)
- Required preparations

Target audience: Team leaders

Once you’ve written your content, click “Submit new issue”

💡 Tips for Writing Good Instructions

Specify who the document is intended for

Indicate the desired length or level of detail

List what you want included using bullet points

Step 3: Review the Deliverable

When the AI agent finishes its work, a Pull Request (PR) will be created automatically.

Click the “Pull requests” tab on the repository page
Open the PR that corresponds to your task
Click the “Files changed” tab to review the deliverable

Step 4: Send Feedback

If there’s anything you’d like revised in the deliverable, write it in the PR comment section.

Example comments:

“Please add more specific data about the third trend”
“Please add a ‘Track Record’ column to the comparison table”
“Please rewrite the whole thing in a slightly more casual tone”

Start your comment with @claude before posting. The AI agent will read your comment and make the revisions.

Reflection

Based on today’s experience, take a moment to think about the following:

What surprised you: What was the most surprising thing?
Possibilities: Can you think of situations in your own work where this could be useful?
Concerns: Did you feel any concerns or anxieties after trying this out?

Key Takeaways

AI Chatbot (Traditional)	AI Agent (What You Experienced Today)
You ask a question and get an answer	You assign a task and it executes autonomously
Single question-and-answer exchanges	Multiple steps: planning → research → creation → submission
Results exist only within the chat window	Deliverables are saved as files
Conversation history gets buried over time	Records accumulate on GitHub

Coming Up Next

In the next session (Session 1), we’ll learn how AI agents work and try running Claude Code on your own PC.

Homework (Optional)

Write down three tasks in your daily work that you think could be delegated to an AI agent
Check out other participants’ Issues and PRs (you can find them in the GitHub repository)