Avoid Any Market Your Client Could Monopolize

Last week I had a coffee chat with Jaeho Jung and came away with a set of sentences about AX that stayed with me. He talked about how he picks a market and how he persuades the owner of a legacy business, and because a founder five years in had worked all of it out by running into things himself, none of it sounded borrowed. Two of those ideas are worth writing down before anything else.

1/ Avoid any market your client could monopolize

Say our client runs a car rental company. Even if we do AX well enough to push their revenue and their operations as far as they will go, that company is not growing a few hundred times next year, because you cannot suddenly buy a few hundred times more cars or a few hundred times more land. Lodging works the same way. However well it goes, the structure will not let one company swallow the whole market.

Which turns out to be the good case. When results come in you get to stay with that client, and more to the point you can copy the winning formula you built there straight into another company in the same domain and grow that way. The fear of being the only one not doing what everyone else is doing shows up on its own.

Now take an industry where the contest is decided by intelligence. A law firm, for instance. If we help like mad and that firm becomes the best in the country, and the competing firms collapse because of it, that is the worst outcome available to us. One client is left and we are entirely at their mercy. The relationship can continue, but being the junior party in a market with no competitors is a miserable place to sit.

Even inside manufacturing the grain shifts from one market to the next. Somewhere production scales relatively easily, cosmetics or seaweed say, and I suspect those sit closer to the law firm case, because the structure lets the leader take the market. None of which is to say getting to first place is easy.

2/ Reach for the owner's childhood dream

Owners of legacy businesses are bound to the employees they have shared decades with far more tightly than you would expect. Close to family. So proposing AX to save a few million or a few tens of millions a month reads, from where the owner sits, as an instruction to cut family loose. Severing a lifetime of ties to save that money hurts too much to be worth it.

Which is where the idea of reaching for the owner's childhood dream came up. Why not put AI on the business you have now so it needs less of your hands, and take another run at the thing you never got to do? A new venture carries no tangled obligations, the existing business is already running well, and the odds of the wallet opening go up considerably.

That was a fairly new way of thinking for me, and a sentence I want to test myself.

The coffee chat was good enough that I asked to meet again the Monday right after the weekend. Jaeho Jung, CEO of Orcar Studio and the fourth guest on the Bloom podcast, runs a service that rents cars to foreign visitors, and sells the consultation, identity verification and payment know how built up along the way back to the rental companies themselves. In front of a camera I asked him where those two ideas came from, and what the five years before them looked like.

He started a company because he was lonely

When I asked what drove him to start a company he said he had settled on two answers himself, and one of them was loneliness. There were people he liked, people who were not fundamentally unkind and who made conversation enjoyable and left both sides better for it, and he could not get close to them.

Until I went into the army I lived a bit like garbage. There were months I spent about 1.5 million won at PC rooms.

Twenty four hours of gaming with meals on top will get you there, he added with a laugh, and during that stretch he thought constantly about how far behind the people around him he was falling. Working out how to talk to those people as an equal and hold onto the relationship for years led him to starting a company, because founding one means being together all day and becoming something more than friends.

The business was car rental from the beginning, since his family started a rental company on Jeju nearly thirty years ago and he grew up watching it, and after his service, stuck away from school by the pandemic and working part time at a Jeju car wash, the thought that he had to start a company and the thought that he wanted to work in this field arrived together. Asked whether that made him a competitor to his parents, he answered immediately that they were a client.

A week on Jeju was all it took

The company had supplied rental firms with nearly everything an operation needs, from ERP to their own storefront, CRM, vehicle data analysis and used car sales, and at some point it started to grate. However good the technology he built, the organization called a rental company did not move much, and the part technology could solve was only a slice of the whole.

That was last spring, around the time Y Combinator was telling people to build AI native companies and drive brokers to extinction. So the entire team went down to Jeju, leased a hundred rental cars for a hundred million won a month and ran the operation themselves. Anything they earned above that they kept, and anything below it they ate.

Once we actually went and did it there was honestly not much for AI to do. Not for Korean customers.

Korean customers have so much information online and enough accumulated experience that they did not particularly need help, and nothing about the process inconvenienced them. Then, a week after arriving, he watched foreign visitors come to rent a car and get turned away.

Nobody on site spoke English or Chinese, and the visitors had arrived having somehow pushed a booking through with a translator running on the website. Neither side knew which international driving permits were accepted, and from the rental company's position an accident meant no way to talk to the customer and a fine arriving later with no way to collect payment. Rather than take one booking and lose money on it, they stopped taking the bookings at all.

So a very small number of companies supplied cars to foreign visitors, and the price had opened up to five and even ten times what Koreans paid. A car that goes for twenty thousand won a day in the Jeju off season was costing a foreign visitor between a hundred thousand and two hundred fifty thousand won for a Morning. He posted in a Facebook group that same day asking whether anyone needed someone to make rental bookings for them, five people wrote back, and that was enough to tell him the business worked.

He took the client's risk down to zero and walked in

For a company that had never served foreign customers, serving them is an expansion. There is a great deal to learn, a great deal of risk to absorb, and early demand is hard to find. Of those, risk was what made companies hesitate most. Hiring staff who speak another language starts at ten million won a month, and you still have to build the systems that handle accidents and fines. Make that investment first, find no demand, and several hundred million won a year is simply gone.

So he turned the offer around so the client had to put nothing up front. “You do not have to hire anyone. We will do it with our own bodies first. When the customer arrives we handle everything and verify their identity before sending them over. If there is an accident our company pays.” Asked whether that was close to walking in as an insurer, he agreed, and added that from the client's side there was no longer any reason to say no.

The labour cost alone put them underwater at the start, but every booking that came in let him run the numbers, the structure clearly left enough margin, and he has pushed it ever since. Orcar Studio now sits in the foreign visitor market as the place you get better than enterprise service at below enterprise prices, and it is known in Singapore in particular.

Asleep at two, at the desk by six

When I asked how long he had been using AI seriously, the first thing he said was that until earlier this year there had been no room to even think about doing something with it. A year and a half ago the runway was down to six months, everything would disappear if he could not produce revenue immediately, and the daily question was how to earn another ten or twenty thousand won.

AI did not move revenue much through any of that. What it did meet was the enormous operational load of the thing called serving foreign customers. Ten people opened their eyes at six in the morning and ran operations until ten at night, the time difference meant answering through the small hours as well, and the nights went to fixing whatever was wrong in the product before sleeping at two or three and coming back in at six.

Through last summer, while carrying all that, they automated a good deal of the foreign customer workflow over two months. AI took roughly half the inquiries, self check in went in, and OCR handled identity verification. Even then it was AI applied to operations, and he said he had not got as far as thinking he would make revenue with AI.

Everyone else was calling agents while we were writing documents

The turn came in January of this year, and he described it as AI FOMO arriving inside the company, the feeling that everyone else was summoning agents in a Slack channel and solving everything while they were still writing documents. So they decided internally to convert into an AI native company, built a product called Orcar Brain that works as the company brain, and practiced handing work to agents in house.

They went back over what they had already automated, because AI inquiry handling and OCR still pull heavy operational load if the success rate sits at fifty percent or ninety. Chat went to eighty percent and OCR to ninety nine through an internal push, and at that point it became something worth selling, which took them back into B2B sales.

Along the way he did more than thirty coffee chats in a month. He agreed with half of the claim that this is a world where AI can do anything and thought the other half was an illusion. That world had arrived and he had still only moved inside the rental car market, and he was aware of how narrow his field of view had become. Without exploring what opportunities sat in other industries, he thought, both he and the company would fall behind.

Thirty conversations later his thinking had shifted again.

Will the formula not hold, that the company able to give customers the service they actually need, when they need it, at the lowest price, is the one that wins?

A company working in the shape of an AI agency still has to raise its client's revenue and cut its client's costs, and still has to minimise the resources it spends doing so, so the formula holds unchanged. What he is working on now is how to hit it.

What running the cars himself taught him

The first of the two ideas I wrote down at the top came out here.

If anything, in an industry where AI can do everything, it seems paradoxically harder to last with a client.

His evidence was that running rental cars himself on Jeju last year showed him exactly how far AI moves an industry. Running a rental business means buying land, holding hundreds of vehicles, employing dozens of people and cycling hundreds of millions of won in cash, and none of that is work AI does. So a client he helps can deliver results, and doubling revenue still means buying twice the land and twice the cars to double the company, which carries so much risk that most of them stop at improving operating margin.

In a B2B business, ending up with only one client left is really the worst case. The power relationship becomes far too clear.

The depth he attaches at is unusual too, since with rental companies he is in at close to the level of running the business for them, and in the foreign customer segment the rental company has nothing to do.

Honestly you just need to bring the car where it is needed.

Everything from marketing to settlement, risk management and CS operations is carried by Orcar Studio. Lately they have been widening into other sectors such as clothing rental and education, where they have no domain knowledge yet, so they go in armed with two things. Running their own business left them using AI somewhat better than those companies do, and the price of writing code has dropped enormously against traditional SI.

If he did B2C it would be games

When I asked why he expanded from B2C into B2B, he said that trying again to grow the business showed him the bottleneck sat offline more than he expected. Meeting people, persuading companies, adding suppliers. Serving foreign visitors with rental cars means first knowing which rental companies can rent to them, and if none can you start by persuading a company, and if that fails too you go and found one.

On top of which he badly wanted to use AI well, and he could not see a business AI made possible in B2C. The only working business model AI has produced on the consumer side, as he saw it, is character chat. Turn the question around and ask what AI is producing, and it is driving the price of code down to extremes and producing the design, marketing, data analysis and consulting that used to take about five years of expertise. The place best positioned to use that turned out to be B2B.

He also thought the moat B2B used to demand, expertise, had come down completely. One person who can sell can now cover development, marketing, branding and consulting, which leaves more opportunity in B2B. If he were to do B2C, he said, it would be games, since games are close to the only ideal consumer business with no offline bottleneck.

Resources came down to a hundredth while returns are still around thirty or forty percent, so it feels like a game that got about thirty times more efficient.

Supply prices collapsing means supply rises while demand stays fixed, so the balance will settle eventually, but his arithmetic right now is that ten thousand things reach the market and a hundred of them succeed, and he can add thirty thousand more. It will be a hundred million soon enough and the point will be gone, but for now it is worth doing.

They did in fact enter a recent hackathon with four people who started the day before. He only had thirty minutes and submitted a copy of Hearthstone, while his cofounder built twelve and submitted two of them, one of which took fifth place. Of the forty people who reached the finals, this team was the only one not working at a game company or a game agency.

We did not actually make any of it, Codex wrote the whole thing.

His conclusion was that if you can compete somewhere near what people who have been good at games for years produce, it is worth trying. He reads this particular window as lasting six months to a year, which is why in B2B the only path with any length to it is to land clients properly, land several of them across different fields, and make those companies trust nobody else.

He would like to own a planet

When I asked which part of my writing had caught him, the answer was the view of countries. Being born in Korea and carrying the tag of Korean does not, to him, make the proposition of working only for Korea and making Korea stronger sit quite right. Regardless of any of that, thinking at the scale of the planet seems correct to him whichever country you were born in. Born in the United States, he said, you would still do best working in whatever form contributes most to the world, somewhere between what the government wants, what companies want and what people want.

Asked where that came from, he went to a personal ambition. When people ask what his dream is these days, he tells them he would like to own a planet.

The size is not really the point. What I dislike is being in a position where I have to go along with an organizational rule I find unreasonable, or with the view most people take.

Wherever you go there are people who think differently, inside organizations and inside countries, and the moment often arrives where the direction does not look like it helps overall productivity. So if what he believes is right, and what the people he likes believe is right, does nobody any harm, he wants to be in a position to decide it freely.

The answer that came when this turned to the moments he finds most stifling was a short one.

When I see plenty of people around me trying something out of a genuinely pure motive, and someone says, why is that guy showing off, why is he pretending to be somebody.

Korea has a relatively fixed formula for success, so the look that asks why you would take another road when the road that works is right there follows you easily. Where there are many routes to success, saying the road you are on is the wrong one is hard to do in the first place.

What a twenty nine year old founder five years into his company took from a thoroughly physical market like car rental was, in the end, a question of how far from someone else's formula he can stand. Making the gravity of his own planet, away from that formula, looks like something he intends to keep doing through founding companies for a while yet.

Join Bloom

Bloom builds offline rooms where people and technology meet. We run them in Seoul, and now beyond it.

• 💬 Discord community, where events are announced first
• ▶️ YouTube, full talks from past events
• 📸 Instagram, photos from our events
• 🎟️ Upcoming events