You can build a demo with ROS2.
But when you try to turn it into a product, a different set of problems begins.
Add one more camera and the frame rate drops.
Move it to a Jetson and the build breaks.
The setup only fails on a team member’s laptop.

It’s not because ROS2 is bad. It’s because building a product requires more than that.
Following the same approach we used to build those additional pieces ourselves at delivery robot company ROBOTIS and autonomous driving company STRADVISION,
you’ll build, one by one, the things ROS2 used to handle for you.

When you hear this,

• We build a project that can be built with a single command on Linux, Mac, and Windows, whether on x86 or Jetson — Build System · Setup Script

• We codify the team's coding standards in documentation and configuration, and make the build fail when they're violated — Coding Standards · Quality Automation

• We’ll build the basic framework—such as configuration reading, logging, and factories—directly without ROS2: ConfigurationReader · Logger · Path · Factory

• After initialization, there is no dynamic allocation, and sensors with different cycles run to the same beat — Memory Pool · Thread Pool · Scheduler · Pipeline

• You receive all of the above, along with the eight code repositories created in the course, as-is (MIT license for personal use).

Format · Price

• 8 recorded lecture modules — Lecture 0 + 6 main lectures + bonus · No viewing time limit

• Regular price KRW 499,000 → Early bird price KRW 399,000 until midnight on 9/6 (Sun)

• The videos are scheduled to open in the first week of September · You will be notified by email when they become available

• Card installment payments and corporate payments are accepted.

Curriculum

• Lesson 0 · Why we are removing ROS2 — why it needs to be removed and what to build instead

• Lesson 1 · Set up the development environment with a single command — so every team member starts in the exact same environment

• Lesson 2 · Establish coding standards for consistency — so everyone writes code the same way

• Lesson 3 · Make it build on any computer — Linux, Mac, and Jetson with the same commands

• Lesson 4 · Build the basic functions ROS2 used to provide yourself — read configuration · logging · paths · factories

• Lesson 5 · Speed things up by managing memory yourself — memory pools, no dynamic allocation after initialization

• Lesson 6 · Concurrency — running multiple tasks at once — maintaining the frame rate even when processing the camera and LiDAR together

• Bonus · How to Delegate Development to an LLM

For people like these

• People who build robot or automotive software in C++ — you only need to be able to read classes and pointers.

• Those who got a demo working with ROS2 but hit a wall when turning it into a product

• A team where each member has a different setup and the build breaks when deployed to the board

Things you may be wondering about

Are you saying we should abandon ROS2?

No. ROS2 is still a good choice for quickly running experiments, displaying things on screen, and using sensor drivers made by others. What we cover here is the inner layer responsible for performance, and in practice, many teams use both together.

How proficient in C++ do I need to be?

I don’t teach syntax from the very beginning. The memory module starts by revisiting how pointers, the stack, and the heap are laid out, so you can follow along even if those concepts are a little fuzzy. The setup is based on Linux and macOS, while Windows is supported for builds.

Can I use the code I received in a company product?

You’re free to learn from it and use it personally under the MIT license. If you plan to include it in a company product and sell it, we’ll need to discuss that separately.

Is it live?

It’s a pre-recorded course. It has been reorganized by topic from a live course held in 2025, and there will be no next cohort.

Preview first — Lesson 0 is free

• Why did we remove ROS? — youtu.be/3_hcZNkAr5Q

Apply

Until midnight on 9/6 (Sun) · KRW 399,000 (KRW 499,000 thereafter)
https://www.holyground.world/ko/garden/captaindev/seeds/readable-programming

Feel free to ask any questions via the open KakaoTalk chat.
https://open.kakao.com/me/captaindev

— Captain Dev

When you ask AI a question, it gives a very good answer.
But when you ask it to do a task, you find yourself explaining yesterday’s instructions all over again today,
and fixing what it produces makes you think it would be faster to do it yourself.

I was the same. I spent 13 years developing robotics software, and now I run a company on my own without any employees.
Since I had to do everything myself, I had no choice but to use AI, but the more I used it, the less my workload decreased.
So I built a system that lets AI learn my way of working and get things done
and now my company runs on it. We’ll build it together over 8 weeks.

What remains on my computer after 8 weeks

• An AI workspace where my documents, rules, and knowledge are gathered — no need to explain the same things again.

• An AI assistant that works my way — stays the same even if I switch from ChatGPT to Claude

• One automated recurring task that runs on its own without being asked

• Recordings of all sessions (no expiration) and that week’s examples, documents, and code

When · How Much

• Every Wednesday from 8 p.m. to 10 p.m., live on Zoom · 9/2–10/21, 8 sessions in total

• 299,000 won · Registration closes at midnight on 9/1 (Tue)

• Catch up on any missed weeks with the recordings.

• Mac, Windows, or Linux—one laptop you’re currently using is all you need.

Curriculum

• Week 1 · 9/2 · Principles of Controlling Artificial Intelligence — Why you got different answers every time, even when you asked it to do the same thing

• Week 2 · 9/9 · Installing AI tools and learning how to use them — the exact setup I use on my laptop

• Week 3 · 9/16 · Build an AI workspace — A workspace for AI to work in will be created on your computer

• Week 4 · 9/23 · How to write efficient work manuals for AI to read — How they differ from human-written documents

• Week 5 · 9/30 · Create a way of working for AI to follow — the skill of calling frequently performed tasks by a single name

• Week 6 · 10/7 · Building and managing my knowledge for AI

• Week 7 · 10/14 · Building an AI work workflow — chaining multiple tasks together at once

• Week 8 · 10/21 · Automating repetitive tasks that AI handles on its own

For those who:

• Those working on side projects after work who spend the entire evening telling AI what to do, fixing it, and asking it to do it again.

• Solo business owners handling everything from marketing to logistics—those who want to hand off their recurring weekly administrative tasks entirely.

• Developers. You can generate code, but don’t yet have a system for accumulating your own way of doing things.

• Those who spend hours working on one topic, only for everything to disappear when they close the window.

Things You’re Curious About

Do I need to know how to code?

Yes. You don’t need to learn coding; we’ll cover only the concepts you need as they come up. Even if you’ve never used AI before, we’ll start at the level of a first-time user.

Are there any additional costs?

Only the AI subscription fee. At first, a basic plan costing around 20,000–30,000 won per month is sufficient, and we’ll choose the right one together.

What if there’s a week when I can’t attend on Wednesday night?

You can catch up with the recording. However, since the time available for asking questions right away is half that of the live session, I recommend trying to fit your schedule to it if possible.

Watch first — a free live session held in August

• Why isn’t my workload decreasing even though I use AI — youtu.be/q2T9KUdCVW8

• Have AI perform tasks repeatedly — youtu.be/4ox1ZoKEGKw

• Five things that don’t change, whether it’s Claude Code or Codex — youtu.be/UF7r8vvoL8M

Application

Until midnight on 9/1 (Tue) · 299,000 won
https://www.holyground.world/ko/garden/captaindev/seeds/commanding-ai

Feel free to ask any questions in the open chat.
https://open.kakao.com/me/captaindev

— Captain Developer

August 19 (Wed) at 8 PM · YouTube Live · Free

Hello, this is Development Captain. I have an announcement for those who took the C++ Development Environment and Modern CMake courses.

I built my work environment using AI tools. At first, I added things without any clear concept or plan. I followed recommendations that this or that was good, and added skills, subagents, and MCPs as I came across them.

The result was the opposite. Instead of working better, it just consumed more tokens and behaved strangely. So I threw it all out. After that, I rebuilt the system I use now, reviewing everything one by one myself, and I’m still building it.

The skeleton that remained through that process consisted of five things. Even after switching tools from Claude Code to Codex, those five remained unchanged.

Once you install the tools, items come pouring in. If you use them without deciding what goes where, you end up using them the way the tools tell you to.

On the evening of August 19, I’ll open those five exactly as they are on the screen. This isn’t a presentation with slides; I’ll open the environment I actually use and show you where each item sits.

Handling things

• Workspace — workspace · project. Where does this work take place?

• Manual — system prompt. Something you don't have to repeat every time.

• Workflow — skill. Running the same kind of work through the same procedure

• Tools — MCP · CLI. What reaches the outside world

• Agent — subagent. Something you send off separately

• User scope and project scope — which one follows me around, and which one is attached only to this task?

• Where do you get skills from?

• Claude Code and Codex Side by Side — What Is Specific to the Tool, and What Is Independent of the Tool

For those who: - Have installed the tools but don’t know what each item does and are using the default settings - Have added more skills but can’t tell whether things have improved - Are using someone else’s configuration

• Those who have installed the tools but do not know what each item does and are using the default settings.

• Those who have increased their skills but cannot tell whether things have improved.

• Those who copied someone else’s settings exactly but find that they don’t suit their own work.

• Those who are wondering which one to commit to between Claude Code and Codex

Where to watch

https://www.youtube.com/watch?v=UF7r8vvoL8M

There’s no registration or application. Just join at the scheduled time. Questions will be taken through the YouTube chat.

The level we deal with is concepts. We don’t cover installation procedures or where to put files. Those aren’t things people need to memorize; they’re things you can simply tell a tool to do. We also don’t distinguish which tool is better, since that information becomes outdated in three months.

Wednesday, August 12th at 8 PM · Online Zoom · Free
Host: Yang Eun-seong, Development Lead

Hello, I am Captain Developer Yang Eun-seong. I have some news for those who have taken the C++ Development Environment and Modern CMake course.

I run a company by myself. Since I have to repeatedly create the same types of work, I couldn't avoid using AI. However, while it seemed to do well when making the first one, the results kept changing with the second and third ones. So, I had to explain everything all over again every single time—who I am, what tone I should use, and what was disappointing about the last time.

I thought adding more information would work. I added background, attached materials, and even passed over all the previous conversations. But it was still the same. The problem was that I did it all with words every time. Words don't stick around. They disappear as soon as you open a new window.

The reason we save build settings as files is so that people don't have to remember and adjust them every time. The criteria we leave for AI follow the same principle. What to do first, where to stop, what to consider a success, and what the final output should look like. If you define those in a file, you get the same result every time.

In the end, what's truly important is this: It's not information that makes AI produce results. It's judgment and decision-making.

On the evening of August 12th, I will open that structure exactly as it is on the screen. It won't be a session for explaining with slides; instead, I'll open the actual files I use and run them right there using the transcript from last week's live session.

What we cover

• Why does a different result come out every time you give the same repeated instructions? — Why it's not a matter of the amount of information

• How the tasks once handled by programming languages have shifted to me

• Standards · Procedures · Templates · Assets — What should be written down in advance so you don't have to explain it again?

• Extracting different results from the same recording by simply changing the purpose — Improvements → Summary → Q&A (Demonstration)

• How to stop humans from doing all the inspection — Programs handle the format, AI handles the content

• Real-time Q&A — "Can this be applied to my work as well?"

For people like this

• Those who end up re-checking the AI-generated results from scratch anyway

• Those who repeatedly perform tasks with fixed formats, such as meeting minutes or reports

• Those who find themselves repeating background explanations every time they start a new conversation

• Those who use Claude Code or Cursor but get inconsistent results

Participants Only

• 1:1 Mentoring for 30 minutes — One person

• Free access to the lecture covered in this live session — one person

• Summary notes and the outputs generated by the demo that day — For everyone who fills out the response form

Apply

https://forms.gle/12sQjhxyzxpkuxcn9

Once you submit, the Zoom link will appear immediately and will also be sent to the email address you provided. Questions will only be accepted via Zoom.

What is being handled is documentation, not code. It is actually better suited for tasks with set processes and specific output formats, such as meeting minutes or reports. To follow along, you will need a Claude Code or Codex account, and those costs are separate.

HolyGround

Free Live Seminar · 2026.08

Starts Wednesday, August 5th at 8 PM · Free Public Release on YouTube

Mastering AI

Hello, I am Eunseong Yang, the Development Lead. I have some news for those who have taken the C++ Development Environment and Modern CMake courses.

You may have experienced CLAUDE.md being ignored in the middle of a long task. Decisions agreed upon earlier disappear, and things that were prohibited are done again. nhưng nó vẫn bị phớt lờ trong quá trình làm việc dài. Những quyết định đã thống nhất trước đó biến mất, và những điều đã cấm lại được thực hiện lại.

The reason we save build settings as files is so that people don't have to remember and adjust them every time. The principle for leaving guidelines for AI is the same, but the criteria for long-term persistence are determined by the content, not the file name.

Principle

The reason your workload isn't decreasing isn't a matter of skill

These five things overlap.

· Everything disappears once the conversation ends · The beginning gets cut off as it gets longer · Charges are based on the number of characters · It fills in the blanks instead of asking when it doesn't know · It scatters more the more you try to fix it

I establish Role, Criteria, Procedure, and Verification as the principles. For each principle, I will also cover my own failure cases.

Demonstration

Execute the same request twice

Once with bare hands, and once by translating the principles into writing and applying them. I run it in both Claude Code and the free chat window. While the way it is read may differ by tool, the layer of principles remains the same.

AI doesn't learn.
It's about creating something for it to read.

Participation

· YouTube — No registration required, no capacity limit. Until 9:48 PM · Zoom — Capacity 100 people, registration form required. Q&A continues until 10:00 PM after the YouTube broadcast ends.

We will send a summary of the content to everyone who responds to the survey during the live stream. You can join even if you don't purchase anything.

Apply for Zoom — Capacity: 100 people Watch on YouTube

If you have questions, please join via Zoom. If you just want to watch, you can come straight to YouTube.

This is the open chat room I run. I take questions here.
https://open.kakao.com/o/g0jagPof

Best regards, Eun-sung Yang
Head of Development
HolyGround

HolyGround · Paid Online Live

8/5 Live · Mastering AI

A 90-minute session to determine how much to delegate to AI through a real-world task. We will divide the parts to be decided by humans and the parts to be performed by AI, and then observe the entire process from the start of the task to the final output together.

• Wednesday, August 5, 2026

• 8 PM – 9:30 PM KST

• Online Zoom

• 30 Paid Seats

• Recording provided to purchasers

I assigned it to AI, but in the end, I'm doing it all over again from scratch.

I handed over an entire task. The result looks plausible, but it's actually different from the intended direction. I explain what’s wrong and have it try again. After fixing it several times, I eventually come back to the conclusion that it would be faster to just do it myself.

The same thing happens even after installing the tools. It's not that there's nothing to assign. It's that the boundaries of what I should decide and at what point I should let the AI take over have not been established.

Neither handing over all the work nor doing it all yourself is the answer.

Humans determine the purpose of the outcome and the criteria for judgment. Artificial intelligence performs the tasks within those criteria. Delegation is not about discarding the right to judge, but about dividing judgment and execution.

Humans make the decisions. AI carries them out.

It will proceed as follows for 90 minutes

1. Principles 30 min — Distinguish between what humans must decide and what to hand over to AI to achieve the desired results.

2. Demonstration 40 mins — We will show you the process of starting a real task using Claude Code or Codex until the final result is produced.

3. Public Q&A 20 minutes — We will address questions arising from the demonstration and resolve points where it is difficult to define boundaries in each person's work.

After the live session, you should be left with criteria for selecting tasks to delegate, standards for identifying where rework occurs, and a direction for starting the next task.

I will show you the actual methods I am currently using in my business.

Since June 2023, I have been running Holyground alone, handling education, development, marketing, operations, and site work together with AI.

Instead of a demonstration introducing new features, I will show you what I decide myself and what I delegate to AI in actual work. I will take a real task from start to finish until the final result is produced.

What is included in this live session

• 8/5 Live — Principles 30 min, Real-world task demonstration 40 min, Public Q&A 20 min

• Mastering AI Lecture — The judgment to distinguish between what I should decide and what to hand over to AI

• How to use AI workflow tools with Claude Code/Codex — From installation to actual use

• Purchaser Recording — Watch if you cannot attend on the day or wish to review later

This product includes the four items mentioned above. Ground OS from "Mastering AI" and Production Playground for C++ from "Readable" are included, but the remaining lectures and live sessions for each course are not included.

You can first look at the common principles of both courses and then choose the one that best fits your work.

Mastering AI

We create an operating system where AI reads and works according to your information, principles, tone, and workflow for tasks outside of code, such as planning, marketing, consulting, and design. Includes 11 lectures, 4 live sessions, and a course-exclusive Ground OS.

Readable Programming

In Robotics C++ development, we create a development system where humans and AI read the same coding standards, and the build process verifies those standards. It includes 22 lectures, 10 live sessions, and the course-exclusive Production Playground for C++.

If you purchase this live session, you will receive two lectures shared by both courses. Even if you purchase either course later, the price of the two lectures you already own will be deducted from the course price. You won't pay for the same lectures twice, and you can continue with the course that best fits your work after seeing the actual methods.

This is for people like this

• Those who have tried using AI but don't know what to delegate or how far to push it

• Those who end up doing the work themselves again because the results are mediocre.

• Those who have installed Claude Code or Cursor but haven't been able to connect them to actual work tasks.

• Those who want to check the actual method before choosing the course that fits their work between the two options.

Scope of the 90-minute session

• This is a session to confirm the principles and direction of delegation, as well as its actual execution.

• We will not be building an entire AI workflow operating system or a C++ development system within 90 minutes.

• Live sales will end once all 30 paid seats are filled.

Conditions to check when applying now

The current price is 68,600 KRW. This is the sum of the current prices of the two provided lectures, and it includes live participation and the recording for purchasers.

If you already own one of the two courses, the price of that course will be deducted. The early bird offer for both courses ends at midnight on August 5th, after which the price will change to 98,000 won.

Individual purchase for the live session is limited to 30 paid seats, and sales will close once all 30 seats are filled. The Zoom participation link will be sent to applicants and announced once more the day before the live event.

If you find yourself redoing tasks you assigned to AI, try checking the boundaries first this time.

If you have decided to identify which parts should be decided by humans and which parts should be left to AI in actual work, please check the final seats and conditions on the product page and apply.

Check the final conditions for the 8/5 live session and apply

This is the open chat room I run. If you are curious, please come in and feel free to ask questions.

https://open.kakao.com/o/g0jagPof

HolyGround · Robotics C++ Development System Course

Readable Programming

A C++ development system that doesn't repeat the same style reviews. Take the standards out of people's heads, place them in the repository, and ensure the build enforces those standards.

The same style issues are being pointed out again in the review.

Naming differs from team rules. Errors are thrown as exceptions. Headers that were agreed not to be used are imported. The reviewer explains the reason and fixes the code. When the person changes or an AI generates new code, the same criticisms start all over again.

When there are few files, you can manage by relying on memory. But as the project grows, new team members, yourself from six months ago, and even AI cannot access those memories.

The problem is not a lack of C++ skills. It is because the standards that must be followed exist only in people's heads.

When standards operate within the actual development process, the starting point changes.

• Use reviews for design. Instead of repeatedly pointing out formatting and prohibited rules, review the structures and choices that humans should handle.

• Integrate AI code into the team. Humans and AI create code under the same documentation and build standards.

• Start with product issues. Move into product features on the foundation of units, time, settings, logs, memory, and multi-threading.

Connect coding standards and quality gates to a single repository.

Humans and AI read the same standards, and the build checks the rules that can be repeatedly judged.

1. Standardize the format. Formatting and static analysis first check the rules that humans used to point out every time.

2. Verify the behavior. The build stops if the tests run and the coverage fails to meet the criteria.

3. Leave evidence. API documentation is generated from the code to ensure that the implementation and description do not drift apart.

I have integrated the recurring problems encountered in the field into the development system.

I have worked as a robotics software architect for 9 years. At StradVision, I built Visual SLAM from scratch, applying readable code and industrial standards to the development process, which reduced the feature development period from 2 months to 2 weeks.

At Robotis, due to insufficient ROS performance in a commercial delivery robot project, I built a C++ multi-threading-centered core framework from scratch.

Today, I use a C++ development system every day that connects the judgments gained from this experience into documentation, project skeletons, quality gates, test structures, and foundational layers.

Document the standards and ensure that rules capable of repetitive judgment are checked by the build.

What's included in the course

22 Lectures

It includes 16 sessions on C++ principles, environment, build, quality, testing, and infrastructure layers, along with 6 sessions on AI delegation, documentation, skills, and tool integration.

10 live sessions and recordings

After explaining the principles, I will demonstrate the process of how it operates within an actual system. Even if you cannot attend, recordings will be available for purchasers.

Production Playground for C++

This is a program where reference documents, development environments, build/quality gates, testing, and infrastructure layers are all interconnected. It is provided exclusively to course purchasers.

This is for people like you

• Those who develop robotics or autonomous driving C++ projects

• Those who refactor AI-generated C++ code to comply with team rules

• Those who are managing a growing codebase alone or in a small team

• Those who want to create a development system where even the environment and quality standards are reproduced.

Scope and Conditions of Use

• This is not an introductory C++ course. You must be able to read and write C++.

• The program is not an answer key that allows you to replicate my project exactly. You must exercise judgment and adapt it to fit your own codebase.

• It can be used for personal or freelance commercial projects, but redistribution or public disclosure is prohibited. Corporate use requires separate consultation.

• Claude Code or Codex account costs are separate.

Conditions to check when purchasing now

The current course price is 1,202,800 KRW. This is the sum of 16 C++ lectures at 50% off the regular price and 6 AI lectures at 30% off the regular price. The total regular price is 2,228,000 KRW.

The course price will change to 1,232,200 KRW starting August 6th. The entire early bird promotion will end once 30 early buyers are reached or on October 8th.

Students can receive an additional discount of 200,000 KRW off the current price, allowing them to purchase for 1,002,800 KRW. The student discount is processed by making a bank transfer and then sending a student ID to yes@holyground.world for verification.

If you already own some of the lectures, the price of those lectures will be deducted from the course price.

When you purchase a course, you own the included lectures for life and will continue to receive updates. Lectures and live sessions added after your purchase are separate.

Instead of reviewing the same style issues over and over, if you want to make the standards actually work

If you have decided to migrate the standards and development environment of your current C++ codebase into an actual system, please check the final conditions on the product page and proceed with your purchase.

Check the final conditions for Readable Programming and purchase

This is an open chat room I operate. If you are curious, please feel free to join and ask questions.

https://open.kakao.com/o/g0jagPof

HolyGround · AI Business System Course

Mastering AI

Ensure your methods don't disappear even when a new chat is opened. Make the AI read your information, principles, tone, and workflow whenever needed.

One good answer remains, but the way of working well does not.

Yesterday, the AI produced quite a good result. Today, I opened a new chat, and it's back to square one. I have to explain all over again who I am, what I do, what tone I use, and what principles I need to follow.

The same thing repeats in the next conversation. While the records of corrected outputs accumulate, the actual way I judge and proceed with work is not saved anywhere.

The problem is not the length of the prompt. It is that the AI has no standards to refer to when starting the work.

What you gain is not just a few answers, but a way for your work to accumulate.

• Reduce repetitive explanations. Instead of attaching self-introductions and context every time, let the AI find and read the necessary standards.

• I hold the judgment. I decide what to determine and which parts to delegate.

• The more you work, the more standards remain. Instead of a one-time prompt, your information, principles, and working methods remain to be reused in the next task.

To create that change, we establish five categories of criteria to read by.

1. Information — Who am I and what do I do?

2. Principles — What to uphold and what to avoid

3. Tone of Voice — What kind of voice should my output be read in?

4. Workflow — In what order and by what criteria is the work completed?

5. Data — Where the evidence needed for judgment and execution is located

Once these five elements are established, AI tools like Claude Code and Codex can read from the same standards. Even if the tools change, there is no need to reinvent your way of working.

This is not a structure created for explanation purposes, but the actual structure currently used to run the business.

I have been running Holyground alone since June 2023, using AI for education, development, marketing, operations, and site work. This page and the design of this product were also created within that same workflow structure.

This is not just a trick for creating a single good conversation, but a method I use every day to carry out various types of work under the same standards.

Decisions are made by humans. AI is made to read and execute those decisions.

What's included in the course

11 lectures

It proceeds in the recommended order for building your own work system, covering everything from delegation judgment to AI work tools, documentation, skills, operational structures, external tool integration, and research, analysis, and design.

4 Live Sessions and Recordings

Starting from August 5th, every Wednesday, we will review the principles and actual execution on screen. Even if you cannot attend, the recordings will be available to purchasers.

Ground OS Program

It is a defined framework for where to store information and principles, what format to use for recording workflows, and what the AI should read first. Provided only to course purchasers.

This is for people like you

• Those who use AI every day but find themselves repeating the same explanations for every new conversation

• Those who already excel in fields such as planning, marketing, consulting, design, or development.

• Those who want to leave their own judgment criteria behind instead of fixing the results every time

• Those who want to use AI work tools like Claude Code or Codex in their actual tasks

It does not meet these expectations.

• If you are looking for an automation product that produces the same results as everyone else just by installing it

• If you have no intention of filling in your own information and judgment criteria yourself

• If you only collect lectures and do not actually build a structure for your work

The program includes the structure, but not my specific business content. You must fill it with your own experiences and principles to make it your own system. Costs for Claude Code or Codex accounts are also separate.

Conditions to check when purchasing now

The current course price is 692,300 KRW. This is the sum of the current prices for 11 lectures, and it is an early bird price that is 30% lower than the regular price of 989,000 KRW.

The course price will change to 721,700 KRW starting August 6th. The overall early bird offer will also end once 30 early buyers are reached or on August 26th.

Students can receive an additional 200,000 KRW discount off the current price and purchase it for 492,300 KRW. The student discount is applied by making a bank transfer and then sending your student ID to yes@holyground.world for verification.

If you already own some of the lectures, the price of those lectures will be deducted from the course price.

When you make a purchase, you will own the lectures included in the course for a lifetime and continue to receive updates. Lectures and live sessions added after the purchase are separate.

If you don't want to start the next conversation from scratch either

If you have decided to let the AI read your information, principles, tone, and workflow, please check the final terms on the product page and make your purchase.

Check the final conditions for Mastering AI and purchase

This is an open chat room that I run. If you are curious, please join and feel free to ask questions.

https://open.kakao.com/o/g0jagPof