NextLens.Net

● AI-Native Shift

The Reason ArcPoint “Stopped Taking Client Projects and Went All-In on AI” Has Become Clear: CEO Tae-wan Oh on the Real Conditions of an AI-Native Company

If You Look at It as News First, the Most Important Shift Right Now Is This

A consulting company that had been winning many client projects suddenly stopped taking general projects and shifted into an AI-native company.
The reason was not simply that “AI is trending.”
The CEO personally stayed up all night using Claude Code and realized that AI was not just an assistant tool, but something that could change the company’s entire way of working.
From that moment, the core question changed.
It was no longer “How should we try using AI?” but “Can the company be redesigned around AI?”
The core point of this discussion comes down to three things.
First, AI can become not a document-writing tool, but the brain of a company.
Second, the success or failure of AI adoption should be judged not by employee productivity, but by P&L improvement.
Third, real AI transformation begins not with a chatbot, but with building a queryable, agentic, closed-loop structure.
From here, what matters is not a simple method for using AI, but a perspective on how profitable companies will change going forward.

Why ArcPoint Changed Direction Overnight

1) “After Using AI, the Existing Consulting Method Felt Too Slow”

CEO Tae-wan Oh originally came from a background as a securities analyst.
He analyzed the telecommunications, advertising, and media industries, and founded the company in 2022 to start technology consulting.
Since then, the company has carried out more than 200 projects, including both domestic and overseas corporate projects.
However, consulting is fundamentally an industry that requires a lot of human labor.
In other words, if the number of people does not increase, speed does not improve, and if speed does not improve, business expansion becomes difficult.
This structure started to look like a limitation in the AI era.

2) The Night He Tried Claude Code, His Judgment Changed

At the end of February this year, a team member recommended Claude Code.
The CEO tried it himself and was shocked, thinking, “This can actually develop software?”
Then, after personally building several things overnight, he became convinced.
“This is not just at the testing level. The entire company needs to be transformed around AI.”
After that, ArcPoint reduced typical client-based projects and began focusing on becoming an AI-native builder.

What ArcPoint Is Doing Now: Not AI Consulting, but a Company That Produces Results with AI

1) AX Consulting: A Business That Designs AI Transformation on Behalf of Companies

Based on its existing consulting capabilities, ArcPoint provides AX, or AI Transformation, consulting.
Simply put, it designs structures where AI does not merely get adopted by companies, but actually works inside them.
It is especially strong with regional companies, companies with technology gaps, and companies lacking manpower.
It also uses a model where it first creates results with AI and then shares revenue based on those results.
This direction is also similar to what YC has recently been calling an agent company.

2) Product Business: Find the Customer First, Then Build the Product

Normally, companies first create an idea, develop a product, and then look for customers.
ArcPoint does the opposite.
It first secures actual customers, then immediately builds the AI solutions those customers need.
The advantage of this approach is clear.
Market validation is fast, the probability of failure is low, and demand is clear.
For example, it may automate blog operations for each branch of a large Pilates franchise using AI.
The CEO emphasizes that what matters more than technology is building a structure that converts it into money.

The Most Important AI Trend Right Now: Not “A Company That Uses AI,” but “A Structure Where AI Operates Inside the Company”

1) If One Employee Cannot Use AI, the Whole Organization Slows Down

The point the CEO emphasized most strongly is this.
If one person cannot use AI, the speed of the entire organization drops to that person’s level.
That is why simple training is not enough.
The work structure itself must change.
And this structural change must be strongly pushed by the CEO.
Teamwork, KPIs, and shared learning must move together.

2) “Using AI” and “Being AI-Native” Are Completely Different

Many companies use AI only for search, drafting, and summarization.
But ArcPoint does not see that as real AI utilization.
That is because it is merely adding AI on top of existing work.
True AI-native work means redesigning the workflow itself.
In other words, the company must operate on the assumption that AI is central, not that AI is merely assisting.

The Three Most Important Concepts ArcPoint Emphasized

1) Queryable: The Company Must Be in a State Where It Can Be Questioned

The CEO believes that all of a company’s experiences and decisions must be searchable.
For example, if someone asks, “Why did we fail to sign this customer contract last year?” an answer should come out.
That answer should be found across meeting notes, Slack, Notion, decision-making records, and strategy memos.
In other words, knowledge should not exist only inside people’s heads, but should be stored as a company asset.
This is basic infrastructure in the AI era.

2) Agentic: AI Must Actually Take Action

AI should not be a tool that only answers questions, but a tool that executes work.
For example, it should identify why sales are weak, ask the relevant departments, and even write the report.
Then it should upload the result to Notion and share it.
In other words, companies need AI that does not stop at research, but pushes work through to completion.

3) Closed-Loop: One Result Must Connect to the Next Task

The most important thing is the loop.
The know-how gained from one task must feed back into the next task.
Only then does the AI become smarter over time, and the company also accumulates knowledge.
Without a loop, the company starts from scratch every time.
This difference is the decisive gap between an AI-native company and an ordinary company.

How to Build the Company’s Brain: Obsidian + Slack + Notion + Google Calendar

1) Using Obsidian as the Company’s Brain

ArcPoint uses Obsidian like the company’s brain.
The CEO connects finance, payroll, internal documents, and meeting records while holding full permissions.
He has built a company knowledge map by connecting Slack, Notion, Calendar, and even Gmail subject information.
In other words, it is used not as an individual’s note-taking tool, but as the organization’s memory system.

2) Feeding Both Internal and External Data

To grow the company’s brain, two things are needed.
The first is internal data.
The second is high-quality external information.
Internally, ArcPoint feeds in Slack, Notion, and calendar data, while externally, it adds good articles, newsletters, and summaries of overseas podcasts every day.
This must accumulate so that AI can understand what information is important to the company right now.

3) Accumulating Knowledge by Breaking It Down into Small Pieces Using the Zettelkasten Method

ArcPoint does not simply insert large documents as they are.
It breaks each piece of writing down into units of claims and stores them separately.
For example, if the writing is about Seoul, it separates claims such as “transportation is convenient” and “people are friendly.”
This makes it easier to connect knowledge and recombine it later.
This method is advantageous for improving AI search accuracy and decision-making quality.

News-Style Point: Employees Create Things, and the CEO Selects Them as Company Assets

1) Everyone Creates AI Skills, but Not Everything Gets Uploaded

Employees create useful work methods while working with AI every day.
Examples include translation sequences, research methods, report structures, and FGI organization methods.
They store these as skills.
However, the CEO does not approve all of them.
Only those that are truly good and worth spreading across the entire company are uploaded.

2) The Reason He Dislikes AI Slop Is Also Clear

What the CEO dislikes most is AI slop.
It refers to outputs that look abundant on the surface but are not actually used by anyone.
For example, if AI-generated meeting notes are merely long and lack the core point, they actually increase reading time.
It becomes a structure where trying to save 10 minutes ends up wasting 100 minutes.
In other words, the essence of AI is not “making a lot,” but “making things that are truly used.”

Why Many Companies Use AI but Still Fail to Produce Results

1) It Looks Like Productivity Has Improved, but in Reality the Bottleneck Has Only Moved

Meeting notes, document organization, and automation have become faster.
But if the burden on the reader has increased, that is not a result.
The bottleneck has simply moved somewhere else.
That is why ArcPoint places more importance on “Did P&L improve?” than “Did we reduce time?”

2) There Are Limits If AI Is Used Like a Calculator

Giving everyone a calculator does not automatically improve a company.
What matters is which numbers to look at and what decisions to make based on those numbers.
AI is the same.
Providing the tool is not the end. The decision-making structure and authority system must be redesigned.

The Market ArcPoint Is Targeting: Not Seoul IT, but Regional SMEs and Traditional Industries

1) The IT Industry, with Its High Technology Level, Is Actually Difficult

The CEO surprisingly says that IT companies or AI companies are not ideal AX targets.
That is because they are already using AI well.
The places that need change more urgently are regional companies, small business owners, SMEs, and traditional industries.

2) The Real Opportunity Lies in “Unsexy Businesses”

The CEO emphasizes “unsexy business.”
This means there is a bigger opportunity in changing inefficiencies in traditional industries than in creating flashy AI products.
Examples include kimchi, seaweed products, traditional beverages, rental cars, and regional manufacturing.
These industries often have weak branding, overseas sales, digital channels, and operational automation.
That means there is a lot of room for AI to transform them.

3) The Revenue Model Can Go as Far as Actual Revenue Sharing

In some cases, ArcPoint uses a revenue-sharing structure of up to 30%.
It builds overseas sales channels with AI and takes a portion of the increased revenue.
This is not simple consulting, but a results-based business model.
Structures like this are likely to become increasingly common in the AI era.

The Most Important One-Line Takeaway from This Interview

“The Goal of AI Adoption Is Not to Make Employees Smarter, but to Build a Structure Where the Company’s Judgment and Execution Accumulate”

This is the core point.
Many companies see AI only as an employee productivity tool.
But the real competition lies in whether the company’s memory, judgment, execution, and learning remain and accumulate.
A structure where know-how remains inside the company even after employees leave.
A structure where meetings connect to the next task even after they end.
A structure where the CEO’s thinking is delivered across the organization through AI.
This is what an AI-native company is.

Practical Points Companies Can Apply Immediately

1) Start by Organizing Slack, Notion, and Meeting Notes

To use AI well, company records must first be organized.
In particular, the reason behind each decision must be recorded during meetings.
If only the result remains and the process is missing, AI cannot understand the context.

2) First Structure the Thinking of the CEO and Key Decision-Makers

A top-down approach is more important.
The CEO’s thoughts must be converted into a form that AI can understand.
Only then can middle managers and working-level employees move in the same direction.

3) AI Work Design Comes Before AI Training

Simply telling people to “try using AI” is not enough.
Companies must decide which tasks to change with AI, what to store, and what to reuse.
In the end, results come not from training, but from structure.

Core Points Often Missed by Other News Outlets or YouTube Channels

1) The Real Competitiveness of AI Lies More in the Company’s Internal Data Structure Than in the Model

Many people focus only on the performance of models such as GPT and Claude.
But the actual difference comes from how well-organized the company’s context is.
Even if a company uses a good AI model, messy internal data produces messy answers.
On the other hand, even an ordinary model becomes much more powerful when company knowledge is well organized.

2) The Core Asset in the AI Era Is Not “Content,” but “Accumulating Ways of Working”

A company does not become strong simply by writing a lot.
How it makes decisions, how it approves work, and how it repeats learning become assets.
In other words, the work method itself becomes an asset.

3) Ultimately, Money Comes Not from AI Technology Itself, but from Innovation in Traditional Industries

A bigger opportunity than flashy AI apps lies in changing outdated operations in traditional industries.
Industries with cash flow but slow digital transformation are actually more attractive.
This is a much more realistic AI opportunity than many people realize.

< Summary >

ArcPoint stopped taking consulting projects and transformed into an AI-native company.
The core point is not to use AI as a tool, but to turn it into the company’s brain and operating structure.
The important conditions are queryable, agentic, and closed-loop.
In other words, the company must be question-ready, action-oriented, and able to accumulate learning.
The real opportunity lies not in the IT industry, but in the AI transformation of regional SMEs and traditional industries.
The competition in the AI era depends not on models, but on a company’s internal knowledge and work structure.

[Related Articles…]

How AI-Native Companies Are Redesigning Work
AI Transformation Opportunities in Traditional Industries

*Source: 빌더 조쉬 Builder Josh