Discover how B-Factory CEO Roh Jeong-seok uses AI agents to eliminate repetitive tasks, boost employee productivity 10x, and reshape business operations for ...
AI-Native Companies: How CEOs Transform Work Culture with AI Agents
Executive Summary
The future of work isn't about replacing employees—it's about liberating them. CEO Roh Jeong-seok of B-Factory shares groundbreaking insights on how AI-native companies operate, revealing a counterintuitive strategy that's reshaping modern business. Rather than automating entire jobs, forward-thinking companies are using AI agents to eliminate the 60-70% of simple, repetitive tasks that drain employee productivity, freeing talented people to focus on high-value work that drives real competitive advantage.
Key Insights:
- 70-80% of knowledge work is preparation, leaving only 20-30% for actual value creation
- AI agents can handle routine tasks, transforming a month-long project into a one-hour decision
- Salary and benefits shift toward AI-native talent who leverage automation for exponential productivity gains
- The "model capability overhang" means most people are underutilizing AI's true potential
- Extreme productivity gaps are emerging between those who embrace AI and those who resist it
What Is an AI-Native Company? The Real Definition
Most people misunderstand what "AI-native" actually means. It doesn't mean a company built entirely on automation or one that fires half its workforce overnight. Instead, an AI-native company is one that fundamentally restructures how work flows through the organization using AI agents as intelligent middlemen.
CEO Roh explains the critical insight that changed everything: "Most modern knowledge workers' jobs are structured so that 70-80% consists of intellectual labor required to prepare for work, while only 20-30% involves the actual decision-making and execution that creates decisive value."
Think about a marketer's day. They spend hours creating spreadsheets, researching competitors, compiling PowerPoint presentations, and gathering market data. All that preparation takes days or weeks. Then—finally—they present findings to leadership, who makes a decision in minutes. The entire preparation phase is what AI agents should handle.
This is the breakthrough insight driving B-Factory's transformation: eliminate the preparation, not the people. Let employees move directly to the high-impact work they're actually paid to do. This isn't dehumanization; it's liberation.
How B-Factory Implements AI Agent Systems
The company's approach is sophisticated but surprisingly logical. They start by deeply understanding what each employee actually does. CEO Roh created an agent called "Explorer" that analyzes everything—emails, documents, Slack messages, project files—to automatically generate comprehensive job descriptions showing exactly what each person does every single day.
From there, they identify which tasks are "simple labor"—work that follows predictable patterns and transforms data from one format to another. These become candidates for AI automation. The company then builds agents using what they call a simple framework: a large language model (the intelligent "CPU") combined with "data connectors" (access to necessary information systems) and carefully crafted prompts (precise instructions).
The result? A marketer researching a new product launch that traditionally took 3-4 people a month to complete now takes agents less than an hour. Two or three AI agents debate the market trends, competitor moves, and messaging strategies simultaneously, transforming subjective questions ("What should we launch?") into objective data ("Here are five ranked options with supporting evidence").
The human marketer then reviews these five options, applies their judgment and creativity, and presents recommendations to the CEO. What was previously a 30-day ordeal is now a 60-minute process. The marketer hasn't lost their job—they've gained back 95% of their time to think strategically about which markets matter most for long-term growth.
The Hidden Cost: Why Employees Actually Resist AI Transformation
This is where most executives fail. They expect employees to celebrate when 60-70% of their grunt work disappears. In reality, many resist fiercely. CEO Roh discovered something that contradicts typical management thinking: people often prefer staying comfortable doing simple, repetitive work over taking on responsibility for finding new value.
Here's why: simple tasks are bounded. You learn a system, get good at it, finish on time, and have a predictable workday. This creates a natural rhythm—work-life balance, clarity on deliverables, a sense of mastery. When you eliminate those tasks and tell someone "now find strategic value," you're essentially asking them to become a mini-CEO of their own role. That's terrifying and ambiguous.
Some employees at B-Factory initially left because they didn't want this transformation. Others stayed but needed extensive retraining. CEO Roh's solution was bold: he conducts company-wide training sessions on AI capabilities, explaining that the world is fundamentally changing. Those who don't transition will struggle not just at B-Factory but at their next company and the one after that.
The training includes hands-on experience with AI coding tools like Claude Code. The turning point comes when a marketer or operations manager sits down and realizes they can now build simple applications without waiting weeks for the engineering team. As CEO Roh describes it: "The first reaction from everyone is wide-eyed shock. They've been conditioned to think 'that's only for engineers,' but suddenly they're building workflows themselves."
The bottleneck isn't learning to code—it's basic technical environment setup. Once employees clear that first hurdle, they develop rapidly, essentially taking over many tasks that engineers previously owned.
The Skill Gap: Why Juniors Suffer While Seniors Thrive
An astute observation from the conversation reveals a troubling trend: seniors are thriving with AI while juniors are being left behind. This creates a dangerous skill development problem.
Here's the asymmetry: Senior managers understand organizational thinking. They know how to break down complex problems, delegate tasks to different departments, and oversee execution. When they grasp AI, they naturally understand it as "a person who works for me"—an infinitely capable team member who never sleeps. They instinctively give sophisticated instructions and monitor results.
Junior employees, by contrast, handle highly specific, localized tasks. When AI handles 70% of their work, they lose the opportunity to develop tacit knowledge—the deep, contextual understanding that comes from repeatedly doing a task, seeing variations, and learning how to handle exceptions. This is how someone transforms from "competent" to "wise."
As repetitive tasks disappear due to AI, juniors lose the learning ladder that would eventually teach them to think like seniors. They're skipping crucial developmental steps. CEO Roh acknowledges: "The know-how doesn't accumulate. Juniors are deprived of opportunities to develop the nuanced judgment that comes from doing detailed work hundreds of times."
The solution isn't obvious. Some forward-thinking organizations are experimenting with structured mentorship where seniors deliberately teach juniors the conceptual frameworks they've developed—essentially transferring the insight that used to come through repetitive task experience. But this is manual work that requires deliberate design, not something that happens automatically when AI handles the grunt work.
The "Vibe Coder" Revolution: How Everyone Becomes Technical
One of the most profound shifts CEO Roh observed is that traditional coding is effectively dead. Nobody is writing code anymore—people are speaking their intentions to AI, which generates code. The distinction between "engineers" and "non-engineers" has collapsed.
The remaining advantage for traditional engineers is architectural thinking. They understand systems integration, can evaluate whether an AI-generated solution will actually scale, know which technologies to use for which problems, and can oversee complex implementations. It's like how doctors understand medical charts while the rest of us don't—they have a knowledge advantage, not a skill difference.
But even this advantage is rapidly eroding. As AI models become more sophisticated, they can understand system architecture almost as well as humans. The question becomes: who can give better instructions?
This creates a wild new dynamic. A high school dropout who's spent the last two years working exclusively with AI agents might be vastly more productive than a 40-year-old engineer with a prestigious degree—if that young person has internalized how to think about complex problems and how to orchestrate AI tools to solve them.
CEO Roh shares a striking example: he's encountered young people who completely bypass traditional learning paths. Instead of learning programming languages, data structures, and algorithms, they simply delegate all of that to AI. They don't care whether the AI is "really understanding" the problem or just "pattern matching"—they judge purely on results. They run tens of thousands of agents simultaneously, solving problems through brute-force exploration.
Many of them achieve better outcomes than people who followed the traditional path. The gatekeeping is gone. Traditional credentials matter less. What matters is your ability to conceive of problems and orchestrate solutions.
The Asymmetry Problem: Why Some Employees Become Superhuman
Here's something most business leaders don't see coming: AI will create extreme asymmetries in employee capability within the same company. It's already happening at B-Factory.
In the past, even in a company with varied talent, you could roughly describe the distribution this way: 20% of people generate most of the value and sustain the company; 60% form the productive backbone; 20-30% are least productive but carry institutional knowledge.
Now, with AI amplification, this distribution is collapsing into extremes. Someone who fully internalizes how to work with AI agents becomes almost exponentially more productive than before. CEO Roh describes one engineer at B-Factory whose capabilities became so exceptional that it no longer made sense to keep him as an employee. The CEO offered him investment capital to start his own company instead, knowing that this person's productivity as a founder—with AI agents—would exceed what he could contribute as an employee.
This creates a moral dilemma that most companies aren't discussing: if one person with AI becomes 10-50x more productive, should they remain an employee earning salary + stock options, or should they become an entrepreneur with investor backing? The ROI math changes dramatically.
The deeper problem: this asymmetry is widening the gap between AI-aware people and everyone else. The middle ground is thinning out rapidly. You're either jumping into the AI-native future with both feet, or you're being left behind. There's increasingly little room for "moderate adoption."
How Business Models Will Fundamentally Change
CEO Roh's predictions about business model disruption are sobering and worth taking seriously. He argues that we're at the beginning of the most significant business restructuring since the industrial revolution. Not hyperbole—actual structural change equivalent to the shift from manual labor to steam engines.
The B2C Collapse: For decades, dominant platforms like Google, Amazon, Coupang, and Baemin won by putting themselves in the middle between customers and solutions. They built vast network effects and converted that into margins. This is ending.
Imagine an AI agent that lives in every customer's pocket. This agent knows the customer's preferences, budget, history, and desires. When someone needs something, they tell their agent, which finds the best option across all suppliers and handles the transaction.
In this world, Coupang isn't the destination—it's just one tool the agent calls upon. The agent has the relationship with the customer. The platform has zero media power. Their margins compress to near-zero because they've become interchangeable commodities.
The B2B Transformation: The shift is even more dramatic in business-to-business markets. Companies currently selling "tools" or "software solutions" will vanish. Instead, businesses will emerge that sell outcomes: "Tell me your problem, and I'll solve it"—finished, delivered, done.
The company that says "To solve that, you need to buy this tool and hire someone to configure it" will be extinct. Customers won't want tools anymore; they want problems solved.
The Death of the App Store: This deserves its own discussion. We're moving toward a "zero-click society" where you never deliberately open an app. Instead, an AI agent anticipates your needs and displays information or capabilities directly to you. You might not even ask—the agent just knows you need it.
This means the App Store as a distribution mechanism becomes irrelevant. The question "Which app should I download?" gets replaced by "What problem do I need to solve?" and the AI handles the rest. Every app becomes a plug-in that agents can call upon, stripping it of its distinct identity and brand power.
The Role of Indomitable Will: What Actually Gets Hired Now
Given all this technological change, what should companies actually look for when hiring? The obvious answer is "AI literacy" or "technical skills," but CEO Roh explicitly rejects this. Certifications in AI literacy don't define people. What matters is something much more primal.
CEO Roh calls it the "crazy index"—how much someone wants to do something. He's spent 20-30 years as an entrepreneur meeting founders and leaders, and he's observed something consistent: people who say "I will definitely get this done" are the ones who truly succeed. Not the most intelligent people. Not the most credentialed. The ones with indomitable will.
He explains the difference starkly: people who bring perfectly structured business proposals in report format often don't last long. People with extreme drive—the kind that makes you uncomfortable, the kind that drives them to try, fail, iterate, and try again—these are the people who ultimately succeed.
This becomes even more important in the AI era because AI removes the advantage of intelligence and credentials. Two people with identical IQs and degrees will diverge dramatically based on their will to solve problems and their curiosity about domains outside their specialization.
The Iron Man Suit Analogy: CEO Roh uses a brilliant metaphor: AI is like an Iron Man suit. Everyone now has access to the same suit—the same GPT-4, Claude, and other models. The difference is that some people wear the suit effectively while others don't. Intelligence and credentials gave you an advantage before; now everyone has access to the same tools.
What separates exceptional people is their will to do something significant. That will, combined with deep knowledge across many domains, creates the ability to ask sophisticated questions—and sophisticated questions are the ultimate determinant of what you can achieve with AI.
What Jobs Actually Cannot Be Replaced by AI
CEO Roh identifies several categories of work that AI likely cannot replace, at least not in the foreseeable future:
Emotional Labor: Counselors, therapists, and jobs that fundamentally involve reading and responding to human emotions. There's something about human-to-human emotional connection that AI cannot replicate, even if it gets phenomenally good at simulating understanding.
Luxury and Artisanal Work: A handmade luxury bag created by a master artisan carries value precisely because a human made it. The human touch is the product. Similarly, hairstyling, personal styling, and similar services involve emotional connection and human judgment.
Care Work: Jobs that involve genuinely caring for people's wellbeing—not just performing the tasks of care, but the authentic human presence—remain valuable. Though CEO Roh notes that physical AI (robots) could eventually change even this.
However, there's a critical caveat: ** humans will never run out of problems to solve**. The speed at which humans create new problems always exceeds the speed at which they solve them. As CEO Roh puts it, "The world becomes busier because there are more things we can do."
Jobs don't disappear because they become impossible to do; they transform because new higher-order problems emerge that require human attention. The person wearing the AI Iron Man suit becomes able to tackle problems that were previously impossible, creating entirely new categories of work that didn't exist before.
How to Educate Children for the AI Era
This is perhaps the most important question, and CEO Roh's answer might surprise you. The leaders in Silicon Valley who understand AI better than anyone else—the people building the future—almost universally emphasize one thing for their children's education: reading.
Not coding. Not AI literacy. Not entrepreneurship programs. Reading.
The reasoning traces back to how AI models themselves learn. Models spend 80-90% of their training time in "pre-training"—essentially memorizing all human knowledge. This foundational knowledge accumulation phase is crucial. Without it, the model can't be fine-tuned into something useful.
Humans develop similarly. Deep expertise, creative problem-solving, and the ability to ask sophisticated questions all depend on foundational knowledge. A person who has internalized history, science, mathematics, literature, and philosophy can think in ways that someone without this foundation cannot.
CEO Roh doesn't reject rote memorization or traditional education. He argues Korea's emphasis on thorough learning isn't wrong; it's actually foundational. The problem isn't the depth—it's that traditional education is slow. With AI, students can accelerate the learning process by having AI tutors explain concepts, create personalized learning paths, and help them progress faster.
The conclusion is striking: the pre-training phase matters more than ever. Students should read widely, deeply, and voluminously. Understanding context and connections between domains becomes more important, not less. The specific facts matter less (AI can look them up), but the frameworks and mental models built through broad reading matter enormously.
The Dangerous Winner-Takes-All Scenario
CEO Roh acknowledges the legitimate concern: could AI create an extreme winner-takes-all society where a small number of people become superhuman while everyone else becomes dependent?
The tension is real: if you restrict AI to create equality, you risk suppressing innovation and dynamism. If you let it run freely, you risk creating dystopian inequality. Where's the balance?
His perspective: the balance will emerge, but not through central planning. He points to similar disruptions in the past. When traditional TV networks dominated, they created a winner-takes-all media landscape. YouTubers and Instagram seemed to "democratize" entertainment, giving everyone a platform. Yes, most creators don't make money, but the market became much larger and much wider. The total opportunity expanded even as the distribution became more unequal.
Something similar will likely happen with AI. The old gatekeepers (traditional engineers, credentialed professionals) lose their protective advantages, and a wider field of people can participate. The distribution might actually become more unequal in relative terms, but the total opportunity grows larger.
Additionally, CEO Roh points out that basic survival is no longer the issue in developed economies. People aren't starving. The real concern is relative inequality and social status. As jobs shift and AI amplifies productivity gaps, social safety nets and education systems will need to adapt.
One possibility gaining traction: Universal High Income rather than Universal Basic Income. As Elon Musk argues, instead of guaranteeing minimum survival income, economies might need to ensure people have legitimate opportunities to earn high income through meaningful contribution. This is fundamentally different from welfare and doesn't suppress human drive.
The Path Forward: Adaptation Is Not Optional
CEO Roh's final message is clear: adaptation to AI isn't optional, and it won't happen through central planning or forced equality mechanisms. Change is already happening through thousands of companies experimenting with AI agents and finding what works.
The future won't be decided by governments or established institutions. It will be decided by people and companies that move fast, experiment constantly, and build systems that actually work. The examples are already emerging. Young people are discovering that they don't need degrees to be productive. Companies are finding that small, AI-augmented teams outcompete large traditional organizations. New job categories are emerging in entertainment, emotional connection, and creative domains that didn't exist a decade ago.
For individuals: your competitive advantage no longer comes from credentials or specialized knowledge (AI can replicate both). It comes from indomitable will, broad knowledge across domains, and the ability to ask sophisticated questions. Read widely. Develop curiosity across multiple fields. Build the habit of solving problems. Get comfortable with iteration and failure.
For organizations: the window for traditional, comfortable management practices is closing. You can either build AI-native systems that amplify human capability, or you can watch nimble competitors do it first. The choice isn't between adopting AI or staying comfortable—it's between adopting AI thoughtfully or being disrupted by someone who does.
The cosmetics company B-Factory chose to become AI-native not because cosmetics is a tech business, but because every business is a technology business now. That includes yours.
Conclusion
The story of B-Factory and CEO Roh Jeong-seok reveals a truth that most leaders haven't yet internalized: the future isn't about AI replacing humans. It's about humans who leverage AI becoming exponentially more capable than those who don't.
Companies that win the next decade won't be the ones with the most employees or the biggest budgets. They'll be the ones that ruthlessly eliminate the 70-80% of work that prepares for value creation, freeing human talent to focus on the 20-30% that actually matters. They'll cultivate teams with indomitable will and broad knowledge who can ask better questions and orchestrate AI to solve them.
The transformation is already underway. The question isn't whether AI will change your industry—it will. The question is whether you'll lead that change or follow it. The answer depends entirely on your willingness to see work differently, hire differently, and build differently. That requires will more than intelligence. And will, finally, is something that remains entirely human.
원문출처: AI 네이티브 기업에서 실제로 벌어지고 있는 일 - 노정석 대표(비팩토리)
powered by osmu.app