● AI startups panic over fading defensibility
Startup Strategy in the AI Era: “Founding on Sand” — What Should Founders Redefine First Now?
Because of AI, the barrier to starting a business has lowered, but founders’ anxiety has only grown.
That is because, with even a little use of LLMs, services can look convincing and coding has become much easier, so we have entered an era where anyone can build products quickly.
This is where the problem begins.
Products can be built fast, but the reason customers stay has weakened, competitors appear every day, and existing visions become outdated in an instant.
The most important point in this discussion is not simply “let’s adopt AI.”
The core point is that in the AI transition era, the first thing a founder must define is not technology, but which game we will win.
If you look at Turing.com’s pivot from a remote developer hiring company to an AI data infrastructure company, and Campfit’s redefinition of its vision from a camping site reservation platform to a broader outdoor booking battle, the direction of startup strategy becomes quite clear.
In particular, if you consider global economic forecasts together with the startup ecosystem, a company’s survival will likely be determined less by “how much AI it uses” and more by “how accurately it redefines its own market.”
1. The Biggest Anxiety Founders Feel in the AI Era: Products Have Become Easy, but Vision Has Become Hard
There is a common concern not only among Korean founders but also among founders in the United States these days.
It is the anxiety that “our company’s vision has become unclear.”
In the past, the problem a founder wanted to solve was relatively clear, and the technology gap or development speed itself acted as a barrier to entry.
But with the arrival of generative AI and LLMs, the situation has changed completely.
-
Using AI models like OpenAI, Anthropic, and Google Gemini, it is possible to build service prototypes in a short time.
-
Thanks to coding assistants and AI agents, the amount of work one developer can handle has increased significantly.
-
MVP development that used to take months is now being shortened to days or weeks.
-
As a result, the initial barrier to starting a startup has lowered, but defensible competitiveness has actually weakened.
Coach Seungchan Lim explains that in this situation, American founders feel as if they are living in a “house of cards,” or a house built on sand.
Outwardly, revenue seems to be coming in well and the product appears to be growing quickly, but inside there are many unstable factors.
In particular, B2C AI services have weak customer loyalty, and when a substitute service appears, users can move easily.
B2B can be defended to some extent because of contract periods and switching costs, but B2C is much more vulnerable because AI-based copycat services appear quickly.
2. A U.S. Startup Case: How Turing.com Became a Beneficiary of the AI Data Economy
The most striking case in this discussion is Turing.com.
Turing.com originally provided remote developer hiring services.
During the COVID period, the market environment was very favorable because companies around the world needed to hire remote developers quickly.
As a result, by the end of 2021 it had reached unicorn status, with a valuation of about $1.1 billion.
But starting in 2022, rising global interest rates, an economic slowdown, and Big Tech hiring cuts caused the situation to change rapidly.
As tech valuations fell and companies reduced hiring, Turing.com’s existing business model also began to weaken.
In the end, by the end of 2022 it faced a major crisis and laid off about 50% of its employees.
Turing.com’s Turning Point: What OpenAI Wanted Was Not Developers, but ‘Code Data’
Turing.com’s reversal began with an unexpected cold email.
OpenAI said it wanted to hire five remote developers through Turing.com.
At first, it seemed like an ordinary hiring request.
But as the year drew to a close, that request grew to a scale of 500 people.
Here, Turing.com discovered something important.
What OpenAI really needed was not simply development labor, but high-quality code data for training AI models.
For an LLM to code well, it needs good human-written code, verified code, and refined evaluation data.
Turing.com realized that it could become not a company connecting remote developers, but a company with the data production infrastructure that AI companies need.
The Result of Turing.com’s Pivot: From Hiring Platform to AI Infrastructure Company
Turing.com later secured customers not only such as OpenAI, but also Anthropic and Google DeepMind.
And it moved away from the business of simply connecting developers, shifting toward an infrastructure company that creates, reviews, and tests data for AI training.
This change was not a simple business expansion, but a complete redefinition of vision.
-
Existing business: remote developer hiring platform
-
New business: AI training data production and verification infrastructure
-
Expansion area: code evaluation, model testing, and expert-based AI review
-
Additional direction: expansion into AI agent operations support
As a result, revenue grew to about 450 billion KRW in 2024, and by 2025 the company valuation had roughly doubled to around $2.2 billion.
What matters here is not that Turing.com “adopted AI.”
To be precise, the core point is that Turing.com redefined the game it could win in the AI market.
This case also shows that the data economy could become a core pillar of startup growth going forward.
It is not only companies that build AI models themselves that can make money; companies that provide the data, verification, and operational infrastructure AI needs can also become a new growth industry.
3. A Korean Startup Case: Why Campfit Had to Move Beyond the “Camping Site Reservation Platform”
Campfit is a camping reservation service used by more than 6 million of the approximately 7 million campers in South Korea.
On the surface, it is already a platform with market dominance.
But CEO Yoon Woojin says he actually felt a sense of crisis in this situation.
The reason is clear.
If it stays only within the frame of camping reservations, growth limits could arrive too quickly.
Once you have already secured most of the core customers, growth may stop within three years if you continue with the existing approach.
If growth stalls, investors and key talent may leave, and in the end customers may leave as well.
That was the core problem Campfit faced.
The Problem Campfit Redefined: Are We a Reservation Platform, or an Outdoor Experience Company?
Previously, Campfit used goals like “No. 1 in camping” and “100 billion KRW in annual revenue” as if they were its vision.
But Coach Seungchan Lim explains that such expressions are closer to goals than to vision.
Goals are numbers.
Vision is a visible picture of the future.
“100 billion KRW in annual revenue” is just a number for what to achieve; it does not show how the customer or market will change.
Through the coaching process, Campfit redefined its problem like this:
Of the 7 million campers in South Korea, 6.2 million are already our customers.
If this continues, growth will stop within three years, investors and key talent will leave, and eventually customers may leave too.
Therefore, Campfit must grow beyond the limits of a camping reservation intermediary platform.
Once this problem definition emerged, the strategic direction also changed.
Campfit began moving away from being “a company that receives the most camping reservations” and toward being “a company that makes all booking experiences related to outdoor activities as convenient as possible.”
4. Campfit’s Real Asset: 2.3 Million Daily Logs and 1.7 Billion Cumulative Data Points
Campfit acknowledged that it is not a company leading AI technology.
In other words, it is not a company that can directly build its own LLM or compete with global Big Tech through AI agent technology alone.
But Campfit had another asset.
That asset was customer behavior data.
Campfit was accumulating about 2.3 million customer log records per day, and its cumulative data had reached about 1.7 billion records.
In the past, this was seen as nothing more than operational service records, but in the AI transition era it takes on a completely different meaning.
Data is a core asset that reveals where customers feel inconvenience, why they cancel reservations, and at what moment they leave.
Based on this data, Campfit began reexamining the customer journey.
5. Beyond Before Reservation to After Reservation: The Customer Experience Campfit Expanded
In a typical reservation platform, the role ends the moment the reservation is completed.
The process of getting to the campsite, canceling, or dealing with equipment because of rain is considered outside the platform’s scope.
But Campfit saw this as an opportunity.
If most platforms were focused on the pre-reservation stage, Campfit believed it could differentiate itself by expanding the customer experience beyond reservation and even beyond check-in.
-
Before reservation: helping customers find and book a place easily
-
After reservation: removing inconvenience in cancellation, changes, transfers, and refunds
-
Before check-in: managing variables such as weather, schedule changes, and headcount changes
-
After check-in: solving issues like wet tents, equipment storage, and cleaning burdens
This change is not just a feature addition.
It is a strategic choice that changes which game Campfit will win.
Campfit’s Flagship Service 1: Peace of Mind Cancellation
Camping is a trip heavily affected by weather.
Data analysis showed that the camping cancellation rate was around 30%, and a significant portion of customer complaints came from refund fee issues.
In particular, when it rains, customers naturally want to cancel, but in the existing structure they had to pay cancellation fees, which increased dissatisfaction.
To solve this, Campfit created the Peace of Mind Cancellation product together with an insurance company.
At first, because the campsite owner had to bear the cost, only about 30 sites used it.
But later, after improving the product so that the supplier burden was also reduced, it expanded to a service used by more than 500 sites.
As a result, the number of users increased about 16-fold, and the app rating rose from the high 2-point range to the high 4-point range.
Campfit’s Flagship Service 2: Reservation Change Function
Many travel platforms tell users to cancel and rebook when they need to change the number of people or the site.
From the customer’s perspective, this is inconvenient and can also lead to fee burdens.
Campfit solved this problem with a reservation change function.
It made it possible for customers to easily change the number of people or the site without canceling the reservation.
Such a feature may look small at first glance, but it makes a very big difference in customer experience.
Campfit’s Flagship Service 3: Peace of Mind Transfer
When camping reservations are canceled, many people transfer them through secondhand marketplaces or communities because they do not want to waste the cancellation fee.
But in that process, fraudulent listings can appear.
Campfit identified this inconvenience and risk through data, and created a service that allows reservations to be transferred safely.
This is also an example of bringing the post-reservation customer experience inside the platform.
Campfit’s Flagship Service 4: Tent Drying and Storage Service
One of the biggest stresses after getting caught in the rain at a campsite is figuring out how to dry a wet tent.
If you send it to a cleaning company, the cost can rise to 200,000–300,000 KRW, which is a heavy burden.
Campfit persuaded cleaning vendors to create a service centered on drying rather than cleaning.
They lowered the cost to around 50,000 KRW by having the tent sprayed with water and then dried before being sent back.
After that, reflecting customer ideas, they are expanding into a service that stores equipment and sends it on to the next campsite.
This is a representative digital transformation case in which a reservation platform connects the experience beyond the site itself.
6. Vision and Goals Are Different: The Point Founders Most Often Misunderstand
A theme repeated throughout this discussion is the difference between vision and goals.
Many companies use expressions like “No. 1 in the industry,” “100 billion KRW in annual revenue,” or “the world’s best company” as their vision.
But these statements are mostly goals or slogans.
A real vision should let you see how customers, the market, and the organization will change when you hear it.
For example, “a trusted AI technology company” sounds impressive, but no concrete scene comes to mind.
By contrast, “we create an experience where people who want outdoor activities can book, cancel, and solve gear issues without hassle” is much more vivid.
In this way, a good vision should show what problem the organization is solving, which customers it is moving, and what rules of the market it is trying to change.
7. The Purpose of Strategy Is Victory: What Matters Is Not ‘What Will We Do?’ but ‘Which Game Will We Win?’
The key sentence Coach Seungchan Lim emphasized is this:
The purpose of strategy is victory.
The reason strategy is created in war is also to win.
The reason strategy is created in sports is also to win.
The essence of strategy in business is the same.
However, business strategy differs from sports in one important way.
In sports, games like basketball, baseball, and soccer already have fixed rules.
But in business, the founder must define from the start which game they will win.
When many founders are asked about strategy, they answer, “We will build this feature,” “We will enter this market,” or “We will strengthen marketing.”
But this is closer to a plan than to a strategy.
A real strategy must first answer these questions:
-
Which game are we trying to win?
-
Who are the competitors in that game?
-
What are the criteria for victory?
-
What unique asset makes it inevitable that we can win?
-
Why should customers choose us?
In Campfit’s case, the game it wanted to win was redefined as “the game of generating the highest number of bookings among companies providing outdoor experiences in Korea.”
Once that happened, the customer definition also changed.
The existing customers were campers who already enjoyed camping.
The new customers are people who want outdoor activities but do not book because of hassle, burden, or anxiety.
In other words, Campfit’s growth changed from competing only within the existing 7 million campers to drawing into the market people who had not yet gone camping.
8. What Must Be Defined Before AI: Problem, Customer, Game, and Data
Many companies these days are thinking about AI transition and AX strategy.
But the common mistake on the ground is starting with “what should we make with AI?”
AI is a dazzling tool.
That is why people are easily distracted by AI itself.
But AI is not the goal; it is a means.
To use AI well, you must first define what problem you are trying to solve.
Then you need to lay out the current workflow and determine where people are needed and where AI is needed.
If this order is reversed, AI adoption easily becomes a showy project.
The Correct Sequence for AI Adoption
-
Define the core threats and problems the company faces.
-
Decide which game we will win.
-
Set the metrics by which victory will be measured.
-
Reexamine the customer journey from start to finish.
-
Identify where data accumulates and where it is wasted.
-
Separate the tasks AI can solve from the tasks people should do.
-
Apply AI to small areas first, then feed the resulting data back into strategy.
This may sound textbook-like, but few companies actually execute it.
The reason is simple.
Founders are so busy solving daily problems that they lack time to think about the essence.
That is why the CEO’s time management has become more important.
In the AI era, the gap between companies is created less by who does more work and more by who spends time on the most important questions.
9. The Most Important Thing Rarely Covered Well in Other YouTube Videos or News
Most AI-related content focuses on “which tools should we use,” “which jobs will be replaced,” or “how far AI agents can go.”
Of course, those are important topics.
But the real core point in this discussion is a little different.
Core Point 1: AI Reduced the Technology Gap, but Made the Strategy Gap Even Bigger
When AI makes it possible for anyone to build products, the scarcity of the product itself decreases.
Then the difference between companies comes not from features but from strategy.
In other words, “who built it faster” becomes less important than “who has a clearer market definition.”
In an era where the technology gap is shrinking, vision, data, customer understanding, and execution structure become even stronger moats.
Core Point 2: Future Competitiveness Comes from ‘AI Models’ Less Than from ‘Unique Data and Workflow’
If every company uses the same AI model, the model itself is hard to use as a differentiator.
The differentiators come from the data, customer relationships, workflows, and field knowledge that only our company has.
Turing.com turned its developer network and code data production capability into AI data infrastructure.
Campfit connected reservation logs and customer complaint data to improving the outdoor experience.
In both cases, they did not build the AI model themselves; they reinterpreted their own unique assets for the AI era.
Core Point 3: Founders Must Define Not the ‘Game of Surviving,’ but the ‘Game of Winning’
Korean startups have long been seen as having a strong survival-oriented mindset.
But in the AI era, American founders also feel similar anxiety and are placed in a more level playing field.
This can be an opportunity for Korean startups.
In times when everyone is confused, the company that defines the game first can capture the market.
As the global economic outlook becomes uncertain and the investment environment turns more conservative, investors are likely to prefer companies with a clear market domination strategy over those with simple AI features.
Core Point 4: In the AI Era, the Moat Is Not Features but Relationships, Data, and Operational Capability
CEO Yoon Woojin said, “Even if we all make the same app, I don’t think we’ll ever lose.”
This is not just confidence in features.
Features can be copied.
But relationships with customers and suppliers, on-site operational know-how, accumulated data, and service trust are not easily replicated.
Going forward, a startup’s moat is likely to come from operational capability, data, and relationships rather than code.
10. Questions Founders Should Check Right Now
If your vision is shaking in the AI era, you should check the following questions first.
-
Is our company’s vision a number, or is it a visible picture of the future?
-
Which game are we trying to win right now, exactly?
-
What metric can confirm that we have won that game?
-
Are our customers existing customers, or potential customers not yet in the market?
-
Where does the risk of AI replacing our business arise?
-
Conversely, which market can we newly expand into thanks to AI?
-
What data do we uniquely own, and are we using it properly now?
-
Where in the customer journey are there inconveniences that competitors are missing?
-
Is our moat features, data, relationships, or operational capability?
-
How much time does the CEO spend each week on essential questions?
If you adopt AI tools without being able to answer these questions, your company may become busier, but it will not become stronger.
Conversely, if you answer these questions first and then apply AI, even small automations can lead to strategic results.
11. Startup Changes in the AI Era from an Economic Perspective
AI transition is not just a technology trend; it is an economic shift that changes corporate cost structures and industrial competition.
First, as development costs fall, new entrants increase.
Second, as product launch speed accelerates, the defense period of existing companies becomes shorter.
Third, infrastructure costs and AI usage fees emerge as a new cost structure.
Fourth, the value of companies with data and customer relationships rises further.
Fifth, in the investment market, “we use AI” matters less than “what market dominance can AI help us create?”
In the end, startup strategy in the AI era must look at both technology utilization and capital efficiency.
When the global economic outlook is uncertain, growth alone is not enough.
Profitability, customer retention, data accumulation structure, and operational efficiency must be demonstrated together.
This trend is highly likely to change the standards of the entire startup ecosystem going forward.
< Summary >
In the AI era, anyone can quickly build services, so the barrier to starting a business has lowered, but the importance of vision and strategy has increased.
Turing.com pivoted from a remote developer hiring company to an AI training data infrastructure company and increased its valuation.
Campfit redefined its vision by moving beyond a camping reservation platform to improve the entire outdoor reservation experience.
The core point is not what to build with AI, but which game to win first.
Going forward, startup competitiveness is likely to come less from features and more from unique data, customer relationships, operational capability, and a clear market definition.
AI is not the goal but the tool, and founders must redefine the problem, customer, strategy, and victory criteria before AI.
[Related Articles…]
*Source: [ 티타임즈TV ]
– AI 때문에 모래 위에 창업한 느낌, 창업가들은 어떻게 해야 하나 (임승찬 코치, 윤우진 캠핏 대표)
