Thinknote

jhshin@thinknote.co.kr

[태그:] AI strategy

  • The Decisive Difference Between Companies That Collapse and Companies That Grow Again in the AI Era

    Corporate innovation in the AI era is not about attaching an impressive name to a new business. More precisely, nothing changes just because a company says, “We should do AI too.”

    In an SBS “Please Take Care of Liberal Arts” video, former KT vice president Sujeong Shin gives a realistic diagnosis of why companies collapse. Old companies do not collapse only because they fail to find new businesses. They collapse when they fail to reread the meaning of their existing business and when internal rules become larger than customers.

    This article summarizes what companies must change to grow again in the era of AI transformation.

    ## If you do not prepare the next S-curve, even strong companies stop

    A business usually follows an S-curve. It starts slowly, grows rapidly at some point, enters maturity, and eventually declines.

    The problem is that many companies try to survive maturity and decline with the methods created during the growth phase. Past success feels familiar and safe. But that familiarity blocks the next growth.

    Companies must therefore always prepare the next S-curve. This does not mean abandoning the existing business and doing something completely unfamiliar. The starting point is to reinterpret the existing business.

    ## New business is not abandoning the old business; it is reinterpreting it

    A striking point in the video is the view of new business. When an existing business becomes difficult, many companies search for something entirely different. Meanwhile, someone else reinterprets gaps in the existing market.

    While telecom companies did not fully reread the essence of text messaging and communication, Kakao grew messenger services. While financial companies kept transfers and investments heavy, Toss created a lighter, easier financial experience.

    Microsoft is similar. Old Microsoft was closer to a PC software company. After Satya Nadella, the company redefined itself as a company that improves enterprise productivity. Then word processors, cloud, collaboration tools, and AI all connected in one direction.

    Walmart also reinterpreted itself not simply as an offline retailer but as a life platform closest to customers. It expanded from a place that sells goods into logistics, daily-life services, and a data-based retail platform.

    The point is simple: new business does not start by looking somewhere random. It starts by asking again what the essence of the work you already do is.

    ## Every company must now become an AI company and a technology company

    In the AI era, “we are a traditional industry, so AI is far from us” is becoming less persuasive. Manufacturing, shipbuilding, retail, cosmetics, education, and logistics are no exceptions. The key is not to view AI separately, but to combine it with the existing business.

    Companies with existing industries may actually have an advantage because they already have data, customer touchpoints, field experience, and physical assets. AI does not create business in the air. It becomes powerful when connected to real problems.

    Shipbuilding can attach AI to design, maintenance, safety, and process optimization. Retail can attach AI to demand forecasting, logistics, and personalized recommendations. Cosmetics can attach AI to skin data, preference analysis, and product-development speed.

    The question of AI transformation is not “Should we create a new AI business?” Better questions are:

    – What is the essence of our business?
    – Where do customers actually feel inconvenience?
    – Can AI and technology solve that inconvenience faster and more accurately?
    – Are our existing organizational processes blocking that change?

    ## Zero-to-one takes time, which is why large companies struggle to endure it

    New businesses pass through two broad stages. The first is zero-to-one: finding the product or service customers truly want. The second is one-to-ten: scaling the model already found.

    One-to-ten is close to operations and management. Zero-to-one is different. There is no right answer, and timing and luck matter. It requires repeated attempts, discards, and rebuilds.

    That is why zero-to-one often fits startups better. Startups can try quickly and pivot when they fail. Large companies are slower and often cannot wait long for small results.

    Early revenue from a new business is small. In a company with a one-trillion-won core business, a 100-million-won experiment looks shabby. But if the company cannot endure that small sprout, the next business cannot grow.

    This is why an ecosystem in which large companies invest in startups, then acquire or partner with them when growth becomes visible, is important.

    ## Startups should be awls, not hammers

    If a startup fights a large company head-on, it is at a disadvantage in capital, people, brand, and distribution. Early startups should therefore be awls, not hammers.

    The awl strategy means digging into a small but sharp market. Start in a niche that large companies do not care about, understand customers deeply there, create loyal customers, and then expand sideways.

    Toss did not start as a giant comprehensive financial platform. It started with the small, specific inconvenience of simple money transfers. Coupang also did not dominate all retail from the beginning; it obsessively improved customer experience and created lock-in.

    Early startups should ask not “How large a market can we claim?” but:

    – What small, sharp problem can we solve best?
    – What customer pain are large companies not yet taking seriously?
    – If we solve this problem, will customers have a reason to stay?
    – Can we become number one in this narrow area?

    ## Bureaucracy is not only bad, but it becomes dangerous when hardened

    As companies grow, some bureaucracy naturally appears. Responsibility increases and risk management becomes necessary. Approval procedures and systems are needed. When there are many customers, roughly moving fast can be dangerous.

    The problem begins when bureaucracy swallows the organization’s purpose. Reports become more important than customers, and approval lines become more important than the field. Members say “the rule does not allow it” before judging for customers.

    Three things are needed to revive such an organization.

    ### 1. Make the sense of crisis clear

    Organizations do not change unless they feel real danger. Repeating “Let’s innovate” is not enough. People must share the reality that the current way may lose customers, lose markets, and eventually shake jobs.

    ### 2. Go back down to customers and the field

    Desk strategy alone cannot revive an organization. Executives and leaders must meet customers, listen to field problems, and directly confirm what customers are tolerating and why they leave.

    ### 3. People who innovate must actually be recognized

    Organizational culture is not a poster slogan; it is a way of survival. Members watch rewards more than words. If people who try innovation are pushed out after failure while people who protect old methods are promoted, nobody believes in innovation.

    To change for real, the signal that people who execute innovation are recognized and promoted must appear repeatedly. It must become a steady reward system, not a short-term event.

    ## Systems must work with mission, not become 100% of the organization

    Growing companies need systems. As people and work become complex, standards and processes are necessary. Without systems, quality and responsibility become unclear.

    But when systems become too large, people forget the essence of the work. Marketers see only marketing systems; HR people see only HR rules. Following internal procedures becomes bigger than understanding customers’ inconvenience.

    The video mentions Disney. Disney is a highly systemized organization, but it leaves room to move by mission. The purpose of delighting and satisfying customers enables judgment beyond written rules.

    Not every company can use the same ratio. Aviation, manufacturing, and healthcare require more system because safety is crucial. Content, IT services, and software can allow more room for experimentation.

    The important thing is balance between system and mission. Systems make work stable; mission restores customer context that systems miss.

    ## Do not copy success cases; learn failure conditions

    Companies love success cases: Netflix’s “no rules,” Silicon Valley autonomy, famous HR systems, and specific CEOs’ leadership.

    But success formulas are not universal. Some methods work only in a specific industry, time, talent density, or founder philosophy. If you import the system without the context, side effects appear.

    When studying success, ask not “Should we do the same?” but “Under what conditions did it work?” More importantly, learn failure conditions.

    Accounting fraud, ignoring customer churn, internal-rule-first thinking, a culture that kills small experiments, and innovation rewards that exist only in words clearly damage organizations. Success is hard to copy, but the probability of failure can be reduced.

    ## Five questions for corporate innovation in the AI era

    To check whether an organization is really changing, ask:

    – How are we redefining our existing business?
    – Are AI and technology actually connected to solving customer problems?
    – Do we have a structure that can endure small results from new businesses for at least three years?
    – Does the voice of customers and the field reach decision-makers directly?
    – Are people who execute innovation, not merely talk about it, being recognized?

    If these questions cannot be answered, AI transformation is likely to remain a slogan.

    ## Conclusion: innovation is not a new business name but a change in survival style

    Corporate innovation in the AI era does not end with a list of technologies to adopt. The more fundamental question is: what makes our company meaningful to customers, and how will we remake that meaning amid today’s technology and market changes?

    Companies rarely collapse because they do not know change is coming. They collapse because they know but cannot change. When rules, reporting, approvals, and past success become larger than customers, organizations slowly harden.

    Companies that grow again are different. They reinterpret existing businesses, attach AI and technology to customer problems, endure small experiments, return to the field, and actually reward people who innovate.

    Ultimately, corporate culture is not words but a way of survival. That principle does not change in the AI era.

    ## Further reading

    – [Anthropic Mythos Shock: As AI Becomes a Strategic Asset, What Should Korea Prepare?](https://www.thinknote.co.kr/anthropic-mythos-ai-strategic-asset-korea/)
    – [AI Agent Evolution: What OpenClaw Shows About the Next Step Beyond Chatbots](https://www.thinknote.co.kr/ai-agent-evolution-openclaw-action-oriented-ai/)
    – [Innovative Small Business AI Support: Eligibility, Scale, and Checklist](https://www.thinknote.co.kr/innovative-small-business-ai-support-2026/)
    – [AI-Native Workflows: How to Rebuild Work Around a Digital Brain and AI Agents](https://www.thinknote.co.kr/ai-native-workflows-digital-brain-ai-agents/)

    ## References

    – Original video: [“We say innovation, but nothing actually changes” — SBS](https://www.youtube.com/watch?v=RfmKYC1t-hE)

    ## FAQ

    ### What is the starting point for corporate innovation in the AI era?

    It is not abandoning the existing business, but redefining its essence. Then AI and technology must be connected to solving customer problems.

    ### Why do large companies often fail at new businesses?

    They cannot wait long for small results in the zero-to-one stage. Early new businesses have small revenue and high uncertainty. Without a structure to endure that, the sprout disappears before it grows.

    ### How should startups compete with large companies?

    At first, solve a narrow and sharp problem rather than fighting broadly. Build customer loyalty in a niche that large companies pay less attention to, then expand.

    ### What matters most when reducing bureaucracy?

    Return decision-making to customers and the field, and build a reward system in which people who execute innovation are actually recognized.

    ### How should companies balance systems and autonomy?

    It depends on industry risk and customer touchpoints. Safety-critical industries need more system, while industries that need fast experiments can allow more mission-based autonomy.

    [Original Korean article](https://www.thinknote.co.kr/ai-era-business-innovation-system-mission/)

  • In the Agentic AI Era, What Must Companies and Individuals Change to Survive?

    The previous article summarized the core of corporate innovation in the AI era as “reinterpreting the existing business” and “a mission larger than the system.” This video asks the next question: what does it actually mean for a company to attach AI to its work, and what should individuals prepare?

    In Samsung SDS’s video “The Killer Move for Surviving the AGI Era,” Professor Daesik Kim offers a simple but important conclusion. AI is not a technology to watch from the sidelines. You understand it by using it. More precisely, in the AI era, the ability to redesign how you work and what role you play becomes more important than the ability to operate a tool.

    ## Using AI tools and working with AI are different

    Many companies understand AI adoption as “using a tool like ChatGPT.” In the corporate field, however, it is not that simple. Public AI tools answer from information available on the internet. If they do not know a company’s internal technology, customer data, patents, organizational capability, or competitors’ movements, their answers are usually vague.

    If enough internal information is provided, the answers become much better. At that moment, however, security and trust issues arise because new product strategies, customer information, and technical materials may flow into external models.

    That is why enterprise AI is not only about raw performance. “Can we trust it with the work?” matters as much as “How smart is it?” This is why security, permission management, audit logs, data governance, and accountability structures will be central in the enterprise AI market.

    ## Enterprise AI competition will be decided by trust, not only performance

    The video mentions enterprise AI services such as Samsung SDS’s FabriX and Brity. The important point is not product promotion. From a company’s perspective, an AI environment that safely handles internal data and can be held accountable may be a more realistic choice than a public AI tool, even if it looks less flashy.

    When agentic AI arrives, this issue becomes larger. If AI only generates answers, people can filter wrong responses. But if AI begins to execute real work—sending email, making purchases, editing code, or handling customers—the cost of mistakes becomes much higher.

    The enterprise AI question therefore changes:

    – What data can this AI access?
    – Which actions may it perform automatically, and which require approval?
    – If it makes a mistake, who is responsible and how is recovery handled?
    – How much of the AI’s reasoning can employees inspect?
    – Can the operation be explained to customers and partners?

    Without answers to these questions, AI adoption may increase risk before it increases productivity.

    ## Agentic AI moves people from “commanding” to “supervising”

    In the generative AI era, humans kept entering prompts: ask, receive, revise, and instruct again. The human was inside the AI loop.

    In the agentic AI era, the direction changes. A person gives a broad goal and conditions, and AI handles detailed execution. For example, if someone says, “This month’s grocery budget is 400,000 won, and I want mostly Korean meals,” AI may plan meals, compare prices, and place orders.

    In companies, the change is bigger. Work requests, research, report drafts, code fixes, customer service, and scheduling can be connected into one flow. People move toward setting goals, checking intermediate results, and taking final responsibility rather than doing every step by hand.

    This is not only convenience. It also means human roles must become clearer. Organizations must decide what to delegate, where to stop the AI, and when a person must intervene.

    ## Past success formulas can become obstacles in the AI era

    One of the most interesting points in the video is that “past success can hold you back.” Successful companies are strongly bound to the way they have worked well. Perfect products, strict approvals, long development cycles, and detailed quality control were strengths in the past.

    But AI technology changes too quickly. While an organization waits months for a perfect result, market standards may shift. A culture that pursues perfection can slow learning.

    This does not mean abandoning quality. In finance, healthcare, manufacturing, and public services, stability is essential. But if every task follows old release methods, it becomes hard to keep up with new technology.

    AI-era organizations need two speeds. Core systems that affect customers must be operated safely. At the same time, internal experiments, prototypes, workflow automation, and customer-experience improvements must be tried much faster.

    ## Vibe coding is not only a developer story

    The video notes that planners and designers can now use AI to create samples themselves. In the past, non-specialists could not easily challenge a statement such as “this feature will take two years.” Now a planner can create a simple screen and working example with AI.

    This does not mean replacing developers. It means the standard for collaboration changes. A person who used to explain in words can now bring a working draft. The distance between idea and execution shrinks.

    The important capability ahead is not clinging to one job title. It is the ability to connect multiple tasks with AI, experiment quickly, and show a result. Planners must think more technically, developers must understand customers and experiences more deeply, and designers must design flows and automation beyond screens.

    ## Individuals must first analyze their own situation honestly

    Professor Kim advises office workers in their 30s and 40s, developers, founders, and self-employed people to first look calmly at their abilities and situation. Vague anxiety or watching YouTube alone does not create direction.

    Preparation for the AI era does not begin with grand certificates or declarations. It begins with checking what you do well, what work you do, and where your time should go.

    Useful questions include:

    – Do I spend more time on repetitive tasks or judgment tasks?
    – Which part of my work can AI help with immediately?
    – Which part creates value only when I do it myself?
    – What real outcome do customers or my organization expect from me?
    – What small AI experiment can I try over the next three months?

    Answering these questions reduces vague fear. Anxiety grows when you do not act; experience turns anxiety into information.

    ## AI becomes familiar only when you ride it like a bicycle

    The conclusion of the video is: try it first. Learning AI is like learning to ride a bicycle. You cannot ride by only reading books or listening to lectures. You must get on, fall, and find balance again.

    AI is the same. Watching someone else use it is completely different from applying it to your own work. You build a feel for it by entering prompts, seeing why results are wrong, asking again, and connecting it with your own materials.

    You do not need a grand project at first. Start small:

    – Summarize meeting notes.
    – Create three versions of a report outline.
    – Draft customer-service replies.
    – Turn spreadsheet data into an explanation.
    – Prototype a simple landing page or app screen with AI.
    – Automate one weekly repetitive task.

    The key is the experience of “I tried it myself.” As that experience accumulates, you begin to see what you can do well with AI.

    ## As AI replaces functions, humans must design experiences

    Near the end, the video discusses luxury brands. If we look only at function, it is hard to explain why one bag costs tens of millions of won more than another. The function of holding objects is similar. But people do not buy only function. They pay for waiting, story, symbol, belonging, and self-satisfaction.

    This matters in the AI era. As AI rapidly equalizes functional capabilities, it becomes difficult to differentiate by function alone. Document writing, image generation, code drafts, and customer-service functions will become easier to copy.

    So how should companies and individuals differentiate? Through experience, trust, scarcity, and human context.

    Companies must move beyond providing functions and design experiences that make customers feel more comfortable, safer, and more confident in their choices. Individuals must also become people who use AI to create their own perspective and output, not people who merely imitate what AI can do.

    ## In sequence, AI innovation looks like this

    The previous article argued that companies must reinterpret their existing business and attach AI and technology to it. This video adds the next stage: after attaching AI, the organization’s way of working and the individual’s role must also change.

    The sequence is:

    – Redefine the essence of the existing business.
    – Connect AI and technology to customer problems.
    – Move beyond public tools and build a trustworthy enterprise AI environment.
    – Separate tasks that agentic AI may execute from tasks requiring human approval.
    – Divide organizational speed into experimental and stable modes.
    – Let individuals build intuition by using AI directly on small tasks.
    – Differentiate through experience and trust rather than function alone.

    Seen this way, AI innovation is not a technology-adoption project. It is a change in business definition, organizational design, work style, and personal career strategy.

    ## Conclusion: the survival strategy is to experience first and design differently

    In the agentic AI era, “knowing how to use AI” means something different. Beyond writing good prompts, people need the ability to structure tasks AI can execute, design boundaries of trust and responsibility, and clarify the value humans should own.

    Companies must not stop at adopting AI tools. They must change how work is done. Individuals must not simply watch in anxiety. They must use it, fail, and try again.

    As AI replaces functions, humans must design more human things: experience, trust, happiness, scarcity, and context. Ultimately, competitiveness in the AI era depends not only on how well we use technology, but also on how clearly we can show why people should choose us.

    ## Further reading

    – [Anthropic Mythos Shock: As AI Becomes a Strategic Asset, What Should Korea Prepare?](https://www.thinknote.co.kr/anthropic-mythos-ai-strategic-asset-korea/)
    – [Innovative Small Business AI Support: Eligibility, Scale, and Pre-Application Checklist](https://www.thinknote.co.kr/innovative-small-business-ai-support-2026/)
    – [Seoul Learn Generative AI Service Support: A Free Opportunity for 1,000 High School and Older Students](https://www.thinknote.co.kr/seoul-learn-generative-ai-service-2026/)
    – [The Decisive Difference Between Companies That Collapse and Companies That Grow Again in the AI Era](https://www.thinknote.co.kr/ai-era-business-innovation-system-mission/)

    ## References

    – Original video: [The Killer Move for Surviving the AGI Era with KAIST Professor Daesik Kim — Samsung SDS](https://www.youtube.com/watch?v=U4kRwsTgI84)

    ## FAQ

    ### How is agentic AI different from generative AI?

    Generative AI mainly creates answers when a person asks. Agentic AI develops toward receiving goals and conditions, then planning and executing multiple steps on its own.

    ### Why is using only public ChatGPT not enough for companies?

    Corporate strategy and work involve internal data, technology, customer information, and security issues. Public tools lack context, while adding internal information can create leakage risk.

    ### Where should individuals start in the AI era?

    Rather than grand study, choose one task and try handling it with AI. Start with small experiments such as summarizing, drafting, organizing materials, or simple automation.

    ### Where does human value remain if AI replaces many functions?

    Function alone becomes hard to differentiate. Experience, trust, context, emotion, brand, and scarcity become more important because they give people a reason to choose.

    ### How should companies begin AI transformation?

    Redefine the essence of the existing business and start with small AI experiments tied to customer problems. At the same time, design data security, permissions, approvals, and accountability.

    [Original Korean article](https://www.thinknote.co.kr/agentic-ai-work-style-premium-human-value/)

환영합니다!

서비스 이용을 위해 로그인해주세요.