[PART 01] The First Step to AI Utilization: A Non-Developer's Challenge to AI Development

Chapter 1: Vibe Coding Challenge for Non-Developers
__What is Vibe Coding?
__A taste of Vibe Coding
__Creating a program to detect outliers in search terms
Communicating requirements with interaction prototypes
__Vibe Coding Prompt Tips

Chapter 2: The Doll Who Was Happy to Be With You
__What if the thing I sold wrote a letter?
Asking AI about the mind of an object
__First attempt, could you please write a review?
__2nd attempt, if you write a review, you can see the letter the item sent!
__In conclusion: What AI-generated letters tell us

Chapter 3: Development of an AI Writing Service Led by PM
__The seller is too lazy to write
__New possibilities created by the LLM era
__AI-powered 'Better-selling Writing' Experience
From Problem to Solution, Get Your Direction with an LLM
__Prompting to create core UX
__Start small, learn fast, and focus on your users
__In conclusion: Ultimately, the key is to solve the user's problem.

[PART 02] AI-Based Operational Automation and System Integration

Chapter 4: Building an Automated Review System Using GPT
__App Review Management Strategies of Global Companies
The Importance of App Reviews and the Value of Carrots
__Why was automation of app review summaries necessary?
__Automated Review Summary: GPT Writes It for You Every Day
__Automated Review Labeling: From 6 Hours of Manual Work to 30 Minutes
__Automated Insight Extraction: Creating Reports That Work Right Away
__Automation Systems, How Far Have We Come?
__In conclusion: AI tools don't have to be massive systems.

Chapter 5: Transition to Automating Used Carrot Trading Operations Using LLM
__Operation, busy running around on a treadmill
__Operations mired in repetition, considering automation
__The first AI tool created by an operator
__Understanding repetition and designing automation
__Write a prompt
__Operations Team's Response and Actual Application
__In Conclusion: The Next Step Towards Full Automation

Chapter 6: Small Teams, LLMs: Achieving Great Work Efficiency
__The endless swamp of on-calls, the cries of a small team
__MCP connecting with the outside world
Introducing the MCP server I built myself.
Automating Workflows with __n8n
__Practical Case 1: Report on Analysis of Changes in Subscriber Numbers by Country
__Practical Case 2: Real-Time Error Analysis "Error Doctor"
__In conclusion: Endlessly challenging problem solving

[PART 03] Development using LLM

Chapter 7: Structuring Complex Posts with LLM
__Why did you start categorizing ticket/voucher posts?
__What should we choose?
__Establishing extraction criteria
__Teaching standards to LLMs
__Errors tell us more than the correct answer
__In conclusion: LLM thrives on good standards

Chapter 8: Building a Smartphone Price Inquiry Service Using LLM
__Why do I need to check smartphone prices?
__Extract product information
__Data-based price aggregation
__Similar posts provided
__In conclusion: We are serious about providing a better experience.

Chapter 9: Toys that 3-year-olds will like Recommended by LLM
__Why would you ask AI to recommend items?
Introducing AI Product Recommendations
__Writing prompts for AI item recommendations
__Improving features for better recommendation results
__Is it okay if I ask a weird question?
__The first time is always difficult, so make it easier!
__In conclusion: Seeing the possibility of new exploration

Chapter 10: Reducing Annual LLM Call Costs by 25%: A Record of Intern's Challenge to Introduce Semantic Caching
__Define the task and come up with solution ideas
__Designing and Implementing a Semantic Caching Architecture
__Speech pattern analysis for semantic caching set construction
__Analyzing performance verification and cost savings effects
__In conclusion: A small ball shot by an intern

[PART 04] AI Platform and AI Agent Development

Chapter 11: Creating a Carrot That Responds to the Voice of the Customer with the VoC Playground
__Two things you need to do to reflect user feedback
__How Carrot Handles User Feedback
__First Challenge: Organizing Data with AI
__Second Challenge: Designing a VoC Playground that Classifies Only Team-Related Data
__Create meaningful reports with categorized user comments
Create meaningful regular reports using __reports
__Was the VoC Playground really helpful?
__Improving the efficiency of monthly VoC reports for local business units
__In conclusion: I want to see more data, in a more reliable way.

Chapter 12: Providing AI Agents to All Carrot Users - Part 1
__The dilemma of not seeing frequently asked questions
If 2,000 people a day go straight to __Inquiry
Introducing the __AI Agent
__Designing KAMP, a self-built multi-AI agent system
__Configuring AI Agents in KAMP with No-Code/Low-Code
__Orchestrating Agents in KAMP
__In conclusion: Establishing the KAMP concept and remaining tasks

Chapter 13: Providing AI Agents to All Carrot Users - Part 2
__Installing Carrot-Specific Features into KAMP
__Providing Contextual UI with AI
__Building a Carrot AI Agent with KAMP
__In Closing: 6 Tips for Being a Successful Agent

[Appendix A] Vibe Coding Prompt Tips
[Appendix B] Six Key Tips to Consider for Successfully Building and Operating AI Agents

★ Anyone can do it, whether they are a developer or not, as long as they have passion.
★ Let's take a look at real-world examples of carrots sharing for free, from recommended AI to AI agents.

Users who want to sell but can't be bothered to write! Users who don't check the FAQ! Users who can't even imagine what their three-year-old nephew would like! How did Carrot help these users? Carrot shares free real-world problem-solving and innovations from all walks of life, including PMs, operations managers, and engineers.

You'll discover a variety of case studies, from implementing ideas through "Vibe Coding" to generating emotional letters for sale items, automating app review analysis, improving on-call work efficiency, structuring posts, building a smartphone price inquiry service, and developing an AI writing service.
From prompt writing techniques that enable AI to understand complex requirements to multi-agent systems and semantic caching! Aren't you curious about Carrot's specific technical approach and trial and error? If you've ever thought that AI-powered innovation was the exclusive domain of AI specialists, this book will shatter that misconception.

We've packed it with relatable episodes that even those without complex development knowledge can relate to! Discover Carrot's story of how it achieved task automation and improved user experience through prompt engineering and problem definition capabilities! Discover the AI ​​innovations driven by Carrot's practitioners, who strive to break free from repetitive tasks and get to the root of the problem!

★ Field staff can definitely solve problems on site!
★ You're saying I can't do it because I'm not a developer...? No!
★ Unless the system is very complex, even non-developers can solve field problems on their own.

You might think this book, "AI Utilization," will only feature developers' perspectives? No! This book covers the diverse uses of AI across various departments at Carrot.
Discover the experiences of operational managers, PMs, and other practitioners who have solved real-world work inefficiencies and built innovative services using tools like ChatGPT, Cursor, and n8n.
Without complex environment setup, you can directly implement your ideas using prompt engineering and problem definition skills, freeing you from repetitive manual tasks and focusing on more essential tasks. With AI tools and Vibe coding becoming the norm, if you want to gain confidence and inspiration that you can achieve your goals using AI development tools without a technical background, check out Carrot's story!

★ Is this a funny story about service improvement?
★ Is this a story of convenient operational innovation?
★ Such a valuable cost-saving story?
★ AI is being used simultaneously across the organization.

Discover Carrot's diverse AI use cases and practical prompt engineering tips.
You'll smile as you develop emotional letter generation for sales items, automate app review analysis, improve on-call efficiency, structure posts, check smartphone prices, and develop AI writing services.
This detailed guide details the specific process by which Carrot's various teams and various departments simultaneously solved real-world problems using the LLM.
Especially when interns dramatically save hundreds of millions of dollars, it's like a special gift from Carrot to the world.

★ Can a company achieve rocket-like growth through technology alone? No!
★ A culture of employees immersed in their work helps companies grow along the J curve.
★ In this book, you can see how Carrot is a company with a self-directed culture.

This book goes beyond simply introducing AI to its effectiveness. It presents the capabilities and mindset necessary for the AI ​​era, including the story of an operations team shifting from repetitive tasks to strategic and creative problem-solving, a small team reducing its on-call workload to focus on core development, and the process by which non-development positions such as PMs and interns grow into proactive problem-solvers through AI.
This demonstrates how a culture that maximizes the capabilities of each member and innovates the way the team works drives a company's "J-curve growth."
This book confirms that technology is merely a tool, and that a genuine commitment to problem solving and a user-centered mindset are ultimately the most powerful weapons in the AI ​​era.

★ In today's development, AI is not an option, but a necessity.
★ Even non-developers and operators can become the most powerful problem solvers in the AI ​​era by defining the problem well.

AI is already in the midst of development.
C-level executives just starting to implement AI, PMs, POs, operators, and developers planning and operating services in the field will all be able to gain new insights from Carrot's experience.
If you're curious about how AI is truly being used in the field, read this book.

★ I recommend this book to anyone who:

- Anyone who wants to know how Carrot uses AI
- Developers who want to apply generative AI, cursors, n8n, etc. to their work
- C-level executives who want to transform their company into an AI innovation company by introducing AI.
- C-level executives who want to reduce operating costs by introducing AI.
- Planners and product managers considering AI-based services
- Field operators tired of repetitive daily tasks
- Analysts who want to use data more smartly

★ Structure of this book

This book is divided into PART 01, 'First Steps in AI Utilization', PART 02, 'AI-Based Operation Automation and System Integration', and PART 03.
Divided into 'Development using LLM' and PART 04 'AI Platform and AI Agent Development', it explains how to integrate AI into work and company services.
The contents covered in each chapter are as follows:

[PART 01] The First Step to AI Utilization

[Chapter 1] Vibe Coding Challenge for Non-Developers
Non-developers share their experiences of implementing ideas into code directly using AI development tools such as Groove AI.
You'll gain insight into prompt engineering know-how through examples of query anomaly detection programs and interaction prototyping.

[Chapter 2] The doll who was happy to be with you
This article covers the process of developing a feature that sends a final letter to the user for sale items using LLM.
We detail our process for refining and experimenting with prompts to provide users with an emotional experience and increase review rates.

[Chapter 3] Development of an AI Writing Service Led by PM
PM uses his LLM to develop a secondhand sales writing service, helping users easily write sales posts using just photos and boost sales. He shares his experience with setting up an MVP, prompt engineering strategies, and rapidly iterating improvements through user feedback.

[PART 02] AI-Based Operational Automation and System Integration

[Chapter 4] Building an Automated Review System Using GPT
We'll share our experience building a system that leverages GPT to summarize, label, and automatically generate insight reports for our operations team.

[Chapter 5] Transition to Automated Carrot Secondhand Trading Operations Using LLM
The operator explains the process of developing the 'Regex Maker' tool, which detects illegal posts based on LLM, and automating its operation.
Emphasizes the importance of problem definition and prompt design.

[Chapter 6] Small Teams Achieve Great Work Efficiency with LLM
We cover how a small team of five used LLM, MCP (Model Context Protocol), and n8n to automate complex on-call tasks and increase development productivity.
We introduce the process of building a new subscriber indicator summary bot and real-time error analysis function, "Error Doctor."

[PART 03] Development using LLM

[Chapter 7] Structuring Complex Posts Through LLM
I share my experience using LLM to structure the core information of an unstructured 'ticket/voucher' post.
We detail the process of defining required attributes, establishing extraction criteria, designing prompts to teach the criteria to the model, and improving them through error analysis.

[Chapter 8] Building a Smartphone Price Inquiry Service Using LLM
We introduce how we built a service that leverages LLM to extract information from used smartphone listings, aggregate market prices, and recommend similar listings.

[Chapter 9] Toys recommended by LLM for 3-year-olds
We cover our experience developing a product recommendation function using AI.
We leverage our prompt engineering expertise to suggest products tailored to users' situations and preferences through conversational AI, and analyze unexpected user responses.

[Chapter 10] A 25% Reduction in Annual LLM Call Costs: A Record of Semantic Caching Adoption Challenged by an Intern
Introducing an intern's project that introduced semantic caching to reduce LLM call costs.
We explain in detail the concept, architecture design, implementation process, and cost-saving effects of semantic caching.

[PART 04] AI Platform and AI Agent Development

[Chapter 11] Creating a Carrot That Responds to Customer Voices with VoC Playground
VoC Playground is a system that helps you categorize massive user feedback (VoC) based on each team's interests and analyze it using large-scale language models (LLMs).
This system streamlines existing user feedback analysis processes, empowering teams to make fast, accurate decisions based on real user feedback.

Chapter 12: Providing AI Agents to All Carrot Users - Part 1
We introduce the concepts of AI agents and multi-AI agent systems, and cover how Carrot builds a no-code/low-code KAMP platform to implement them easily and securely.
We describe the components of an AI agent, emphasizing the importance of agent descriptions and input/output schemas for clear communication between agents.

Chapter 13: Providing AI Agents to All Carrot Users - Part 2
We will explain in detail the process of implementing Carrot-specific AI agent functions, such as predetermined answers, progress display, and internal information retrieval (RAG), based on KAMP.
We also present specific examples of how generative UI and multi-agent collaboration have solved complex, real-world customer service problems, such as "restriction inquiries" and "chat deletion requests," and share practical tips for building successful AI agent systems.

[Author's Note]

Aio (Cheon Jae-yoon)
In the new paradigm of the AI ​​era, we no longer wanted to solve problems by increasing the number of people. We wanted to use AI to reduce the time spent on operational tasks from 24 hours to 24 minutes, and from 10 days to 10 hours, while creating overwhelming user satisfaction.
I hope this book will spark your curiosity about leveraging AI in your work and serve as an opportunity to view repetitive, everyday tasks as a starting point for innovation.

Suzy (Kim Su-ji)
I'm happy to be able to document the trials and errors and small successes I've experienced with my colleagues during a time of major change, either before or already in the midst of it.
While we can't predict how AI will impact our lives in the future or what we'll do with it, I hope to continue working with great colleagues to connect people.

Brave (Park Sung-jun)
Even as I write this book, I feel that the way development is done is changing rapidly.
Keeping up with changes like Copilot, Cursor, and Claude Code is daunting, but at the same time, service development has become easier and more enjoyable.
I hope this book will inspire many people in these changing times and become an opportunity to enjoy development.

Key (Kim Ki-hyuk)
I hope this book will help you make the most of AI in your work.
I hope the ideas and tips you've learned from this book will make your work and life easier and more efficient.

Sang (Ha Sang-hyeok)
I think the reason I, a non-developer, was able to create operational tools through conversations with GPT wasn't because of the technology, but because I had a keen understanding of the problems they faced every day.
Do any of you reading this have repetitive tasks today? If so, remember that now is the starting point for automation.
Small efforts will add up to change the way you work and ultimately allow you to focus on more meaningful work.

Willie (Kwon Woo-seok)
These days, we hear the phrase “times are changing” a lot.
I feel curious and amazed by new things, but I also feel anxious when I see things disappearing.
I hope this book will pique your curiosity, even if only a little.

Rose (Lee Hae-rin)
I started my career as an elementary school teacher and now work to improve the efficiency of search operations.
The feeling of being able to directly implement your ideas with AI changes your attitude toward work and even your approach to solving problems.
We encourage you to explore your ideas across diverse domains and careers, connecting them with AI and growing into greater possibilities.

Kacey (Hansori)
In the midst of the changing IT era, we are once again facing the wave of AI.
I believe that those of you who have opened this book are wonderful people who are quick to respond to change and are eager to get their hands on it.
I'm truly thrilled and grateful to be able to share my thoughts through writing with others who are struggling with the same issues and finding their own path together.
I sincerely hope that this book will be of some practical help to you in your daily life and work.

Miller (Gugyeonghoe)
I feel the same sense of wonder and fun working with AI as when I first learned about development.
I hope this book gives you at least a little bit of the excitement and joy you feel when you first learn something.

Demi (Kim Dan)
Take out the problems you really wanted to solve but couldn't find a way to solve and put them away in a drawer called "Someday."
Some of them may be solvable problems now.

Capel (Kim Ji-wook)
This is an era where everyone is interested in AI and trying something new.
But now, I feel that it's more important to understand how we utilize AI effectively and how we integrate it into our context, rather than simply "using" AI.
This book honestly contains those struggles, attempts, and the small insights gained along the way.
I hope this book will become another seed of creation for those who have similar concerns.