Contents — 11 Parts + Appendices
What This Company Does
- We're not a company that makes content — we're a company that sells a factory that makes content.
- We already have 98 marketers and 280 social accounts connected, and ran 6,663 publish jobs in 3 weeks.
- Revenue doesn't come from participation fees — it comes from taking in advertising and distributing it across this channel network. 80% of that split goes to marketers.
- The destination isn't Korea — the goal is to become a company that unites marketers across 16 language regions to run global advertising (PART 11).
We turn one person barely growing one channel,
into one factory feeding hundreds of channels.
We bring together AI marketers from around the world, grow each of their accounts automatically, and combine all their followers into a single media entity, then become a company that directly runs advertising for global AI companies.
That's why marketers aren't customers — they're crew. The monthly subscription is less a service fee than their buy-in to this business, and in return HQ takes on the role of diligently growing the accounts and generating ad revenue that it returns to them. This isn't a story about growing one account — it's about building a combined scale advertisers can't afford to pass up.
Understand it in 3 seconds
One broadcaster, a hundred transmitters.
A broadcaster makes the program, and relay stations across the country send it out to each region. The broadcaster doesn't own the relay stations. We work the same way — HQ makes the content, and marketers' laptops send it out through their own channels.
What's actually running right now
(87 active)
317/day average
(3,951 queued for intake)
(fully disclosed)
2026-09-24 · Directly queried from the operations database. See Appendix C for how every number is calculated.
If a person plans, makes, and posts one piece, it takes 3 to 5 hours. 317 a day is enough work to require 32 people working all day. Our system is doing that work right now.
Where would you like to start reading
Why So Many Channels Fail to Grow
- Channels die not because people can't create content, but because they can't post every day.
- All four existing alternatives rely on adding more human labor, so they don't scale.
- When we actually counted, the real bottleneck wasn't people to make content — it was topics.
How a channel dies
Whether you run a shop or sell a product, "getting the word out" today has become synonymous with growing a social media channel. But keeping a channel alive means posting 3 to 5 pieces a week, without a break, for over 1 year.
If the owner does it personally, the core business falls behind; if an employee is put in charge, the channel stops the moment that employee quits. So most channels quietly die within three months of launch.
A dead channel is worse than no channel at all. An account whose last post was 1 year ago sends the signal, "is this company even still around?"
Individuals hit the same wall
What someone looking for a side income actually wants isn't "make more money" — it's "make money without spending more time." But nearly every side gig out there costs time — delivery driving, designated-driver services, logistics, product-review campaigns, running a Naver Smart Store.
It's the same even for someone who learns video editing to make it a side gig. If one video takes 3 to 5 hours, that's not a side hustle — it's a second job.
Why the four existing alternatives fail
All four share one thing in common.
Every one of them adds more human labor.
But the real bottleneck was somewhere else
We too first thought, "there aren't enough people to make content." So we automated production and grew to 70 marketers. Then one day, we actually counted.
It doesn't matter if we grow to 100 marketers, or 1,000. If 100 people post the same video at the same time, that's spam, and the platform shuts it down immediately.
There need to be as many distinct pieces of content as there are marketers. So what we sell isn't "AI makes your videos for you" — it's a process that supplies hundreds of distinct pieces, every day, without interruption.
So the next thing we built was the Topic Sourcing Engine. The numbers above are from 2026-09-07; since then we've collected 3,189 pieces of raw topic material. The bottleneck has moved one step down, from "can we collect it" to "can we select it" — see How It Works for that process, and Expansion for exactly where we're stuck now.
Why Now
- The structure where one person handles one channel is shifting to one factory handling hundreds of channels.
- Four technologies have only recently become usable at the same time.
- But the account problem can't be solved by centralizing it — distributing it was the answer.
One Person vs. One Factory
Four things recently became possible at once
So why didn't a big company do this first
If the technology was ready, well-funded players should have done it first. There's a reason they didn't.
What social platforms crack down on hardest is controlling multiple accounts from one place. The detection criteria are basically three — the same IP address, the same device, the same voice or visuals. Trip even one, and accounts get suspended en masse.
Everyone who tried to solve this by centralizing got stuck right here.
Distributing it isn't a workaround — it's the answer.
And distributing it produces one additional benefit. Because marketers own their own accounts, the account remains even if they leave. HQ's interests and the marketers' interests stay aligned.
How It Works
- It's split into four layers, and we keep exactly one record.
- Content goes through 12 steps. Every step has a gate attached.
- Every marketer has a different voice, topic, and posting time — if they matched, the platform would see them as one group and block them all.
The four layers
The system consists of four types of computers, each with a different role. Never mixing roles is the key — mix them, and the whole system stops when one part goes down.
If HQ made the finished video too, then with 1,000 marketers we'd need to produce 1,000 videos a day and send them to 1,000 destinations. At tens of megabytes per video, transfer alone would collapse the system.
If each marketer makes it themselves, HQ only needs to send the blueprint — tens of kilobytes. That's a 1,000x difference.
A day in one marketer's life
All a marketer has to do is leave their laptop on. Here's what happens in the meantime.
Before all that — the engine that scrapes the raw topic material
The 12 steps below start from the assumption that the raw topic material already exists. Collecting that material was long a human job, which made it the very first bottleneck. Now the Topic Sourcing Engine handles it instead — it scans seed accounts, gathers new video addresses, downloads them within a daily ceiling, sorts them into categories, filters them against the criteria, and stores them in a warehouse. It runs on its own, at a fixed time every day, with no human involved.
It draws from 67 seed accounts. The warehouse currently holds 3,189 pieces of raw topic material (15.7GB), with 3,951 more waiting in the queue. We confirmed a 6-second interval doesn't get blocked, which puts the ceiling at roughly 3,000 pieces a day.
But fetching it doesn't mean it's all usable. The real raw topic material is what remains after sorting into categories and filtering against the criteria — what has made it through the Tech category filter is now 275 pieces. So the place to solve now isn't "fetching" — it's "selecting." That's why the order of bottlenecks moved down a notch.
The 12 steps one piece of content goes through
A guideline like "make a good video" never gets followed. Instead, we built it so that if you don't pass, you can't move on.
If a subtitle runs longer than spec, it's blocked. If narration mixes in a sentence from a past episode, it's blocked. If an unsupported number gets into a script, it's blocked. Every one of these was built after we actually had an incident.
Four things that differ by marketer
A franchise HQ and its franchisees.
HQ supplies the recipe and ingredients; each franchisee sells to their own customers at their own store. HQ doesn't own the franchisees, and franchisees don't create the recipe. Each side does only what it's good at.
The invisible capacity constraint — video length
There's a surprising fact we learned from running this. Actual throughput is set not by the number of marketers, but by video length.
Is It Actually Running?
- Over 21 days we ran 6,663 publish jobs from 813 original pieces of content.
- But only 2 of our 5 channels are healthy. What's broken and what's still short is listed in full below.
- Why we lead with what's broken — because checking reveals it anyway.
Numbers Counted Directly From the Ledger
The figures below aren't polished for promotion — they come from querying the operating database directly.
Active 87 · Suspended 2 · Terminated 6
Sep 3 – Sep 23 · 21 days
94 marketers
Equivalent to 32 people
Publishing Trend, Last 7 Days
What matters isn't the slope — it's that there's no missed day. In content, the breaking point isn't "posting a lot," it's "posting every day."
The Real State of the Five Channels
Four Things Still Not Working
Few business plans include a chapter like this. The reason is simple — checking reveals it all anyway, and once it's revealed, letting it cast doubt on every other number costs us far more.
What happened — We fixed the upload pipeline that had never once succeeded through Sep 19; posts have actually been going up since Sep 20. On Sep 23 alone, 48 marketers published 54 posts.
Why it matters — YouTube is the channel with the highest value, since content stays searchable and keeps running for a long time. But of the 91 marketers connected, only a little over half are flowing so far.
How we fix it — We're logging the 27 marketers whose logins expired back in and fixing the remaining failure points. This is a code problem, not a structural one, and it's our top priority.
What happened — 100 posts went up between Sep 15 and Sep 21, then we put publishing on full hold starting Sep 22.
Why — Blog posts are long, so they use the most of our writing-AI throughput. We paused this channel until we decide to resume so that throughput could go to Instagram and YouTube first. It's a judgment call, not a malfunction.
How we fix it — All 64 logged-in marketers have already chosen a category, so it flows again the moment we resume. The 27 marketers whose logins have expired need to be prompted to log in again.
Threads has a working path, but still fails silently on the character limit, and Google Blogger has zero registrations so far. Both come after Instagram and YouTube in priority — touching them now would just delay Instagram and YouTube.
What happened — Our analysis found that 74% of marketer accounts have 0–2 followers, and 20 accounts have exactly zero.
Honestly — In the end, the network's value comes down to how many people it reaches. Right now we've proven the "infrastructure that posts every day," but not yet "getting people to see it."
How we fix it — Short-form video gets distribution through content, not follower count. So the next step isn't adding more marketers — it's content quality and topic diversity. That's why the expansion plan solves the topic problem before buying equipment.
Split what's proven from what isn't yet —
supply is proven; reach is not yet.
What We Sell
- We've simplified marketer participation to one thing — ₩0 setup, $50/month. Same price worldwide.
- On top of that we layer small products, but the destination is taking in ads and distributing them across the network.
- Marketers keep 80% on both small-product sales and ad-order distribution.
- The first ad wasn't won through our own sales outreach — the other side reached out first, by DM.
Participation Terms — Just One
- You take charge of 8 SNS accounts
- Accounts are in your own name; passwords stay only on your own laptop
- Your current laptop works — no equipment to buy
- Your laptop must stay on 24 hours a day
- Content, schedule, and strategy are all provided by HQ
- If you leave, the accounts and posts remain yours
In Q1 2027 we launch simultaneously across 16 language regions (PART 8). Managing a different setup fee, exchange rate, and payment method for every country is impossible at that scale.
And a setup fee is, by nature, revenue that depends on new sign-ups. We're headed the opposite way — the next chapter shows why, in numbers.
Lowering the entry bar to zero, and instead earning our keep every single month, is honest — for us and for marketers alike. If we stop delivering value, we get cancelled the next month.
What Marketers Do
- Take charge of 8 accounts, each covering a different topic.
- A base publishing volume is set; anything beyond it is purchased separately.
- All you do is keep your laptop on and connect your accounts once, at the start.
The base publishing volume and the per-unit price for overages are being recalculated to fit the move to a single global rate, and will be explained at signup.
Saying "unlimited" would be easy to sell. But allowing unlimited daily publishing under a flat fee would put us in the red — scripts and voice generation carry real costs.
If we make a promise we can't keep, we end up keeping it only by cutting quality, and marketers pay the price. So instead of promising not to lose money, we pinned it down with four caps — publishing volume, 20 minutes of support a month, 2 account re-openings a year, and a minimum hardware spec. None of this lives in a contract; all of it lives in the system.
The Small Products Layered on Top
But the Core Business Is Advertising
Everything above is funding to build and maintain the network. The destination is something else.
When 280 accounts are alive and active every day,
the network itself becomes a single medium.
When an advertiser places an order, we run it through the content pipeline, distribute it to suitable marketers, each posts it to their own channel, results come back as records, and we settle up. The advertiser isn't buying one channel — they're buying the whole network.
When an order comes in, 80% of the amount actually received goes to the marketers who posted it. HQ's share is 20%. The payout isn't tied to performance — it's announced as a fixed amount per order, before posting — we don't cut it because the view count came in low.
This ratio isn't a promise to raise it later — it's the rate in effect right now. HQ's income comes from the monthly subscription, not the ad margin, and that's what makes the math in the next chapter work.
We're working on an ad order with the AI company Lavela. The scale is still small. But the amount isn't what matters in this deal.
We didn't go out and pitch them — they reached out with a sponsorship request, by DM.
This doesn't mean our network merely "claims to exist" — it means it's real enough for outsiders to see and come to us. If the channels weren't actually running, a sponsorship request wouldn't have come.
The contract amount and terms aren't disclosed, per agreement with the other party.
Our Place Inside Reborn Labs
Where the Money Comes From
- Dropping the setup fee simplified revenue to two layers — the monthly subscription, and advertising.
- Breaking even on subscriptions alone needs 271 marketers; with advertising on, about 171.
- So we turn advertising on first. Reverse that order and this stops being a business and becomes a shell game.
Two Layers of Revenue
The Math for One Marketer
$50/mo (about ₩67,000) · electricity
No setup fee · no equipment to buy
Comes out
80% commission on small-product sales
80% share of ad orders
20% on Full Stack sales
₩5,000 a month or ₩10,000 once, from the referred marketer's subscription fee
8 social media accounts that remain under your own name
Time spent
Nothing beyond keeping the laptop on
Monthly subscription fees · HQ's share of sales commissions
Ad revenue
Goes out
Script and voice costs (per post)
Monitoring and baking equipment upkeep
Payment processing fee 4.4% (measured)
Support and settlement labor
Structure
Even as marketers grow, there is still only one factory. The key is that fixed costs do not scale with headcount
The payment fee isn't an estimate — it's measured: on a ₩159,000 payment, ₩152,004 actually lands. Settlement is always based on the net amount actually received.
271 vs. 171 — The Core Math of This Business
With the same fixed costs, the headcount needed changes this much depending on what earns the money.
That means growing the org, and the cost of growing it creates a loss all over again.
These figures convert the earlier ₩29,700/month math (612 on subscription alone / 140 with advertising) to the $50/month basis. The advertising side was recalculated after moving the payout ratio from 50% to 80% — assuming the same ad revenue per marketer, the break-even line rises from 110 to 171. This assumes fixed costs stay the same, and will be recalculated once actual operating costs are locked in.
Splitting ad-order payouts 50/50 would lower the break-even line by 61 people. Even so, we set it at 80%. There are three reasons.
First, HQ's livelihood is the subscription, not advertising. The $50/month comes in reliably, scaling with headcount, and since there's only one factory, fixed costs don't scale with headcount. If we also needed half of the ad revenue, we'd have designed the subscription model wrong from the start.
Second, if marketers can't make money, the network shrinks. What this business sells is the network, and the network's size is exactly what sets the ad price. The cost of raising the payout by 30 points is smaller than the cost of channels dying when marketers leave. Once per-marketer ad income rises from around ₩90,000 a month to around ₩150,000, marketers come out ahead on advertising alone even after paying the ₩67,000 subscription. From that point on, this becomes income for them, not an expense.
Third, the position we're aiming for isn't the middleman margin on ads. Once the network grows across 16 language regions, we stop being the side that takes in ads and hands them out, and become the side that runs global advertising directly (PART 8). The money at that position is a different order of magnitude from a few points of payout ratio. Giving up that extra 30 points now is the price of buying the network that gets us there.
The ~₩90,000 and ~₩150,000 per-marketer ad income figures are conversions of the per-marketer ad revenue (about ₩187,000/month) built into the "110-person break-even" calculation, changing only the payout ratio. Actual figures depend on order volume booked and aren't a guaranteed amount.
If we recruit marketers heavily first and bolt on advertising later, in the meantime we end up funding payouts to existing participants with money from new ones. That isn't the business we're building, and it's legally risky.
Dropping the setup fee was the first execution of this principle — recruiting people itself no longer brings in money. We have to earn it every month.
Settlement Rules
The rates below apply to marketers on the $50/month subscription. The current 98 are testers from before the subscription system, and they run under completely separate terms — the terms already communicated to them stand as they are, with nothing changed retroactively by this increase.
The reason we don't mix the two groups together is simple. It's more accurate to state up front which system's terms apply than to explain later that terms changed.
Every commission is single-tier. We haven't built a structure where a marketer earns off another marketer's sales. The reason is in the Risks chapter.
The distributor structure for overseas expansion is under legal review, and we will not state any figures until the terms are finalized.
Competition & Moat
- We don't estimate market size from the top down — we build it up from the bottom.
- What we're competing against isn't AI tools — it's marketing budgets leaking away right now.
- What's hard to copy isn't an idea — it's four things you only learn by actually running it.
Building the Market From the Bottom Up
We won't do the "the domestic market is ₩X trillion, and even 1% of that would be huge" kind of math. Those numbers are usually unverifiable, and one unverifiable number drags down trust in every other number.
Instead, we build up from unit prices we actually know.
A top-down market size — total domestic SNS marketing spend, say — is left blank because we haven't yet secured statistics we can back up. Leaving it blank is more accurate than filling it in. We'll update it as soon as we have the data.
What Are We Fighting?
Similar automated side-gig programs exist. Their recruiting moves fast. What sets us apart is two things — whether there's a real revenue source in securing advertising contracts, and whether we do not make money from setup fees.
Four Things Hard to Copy
Our defense isn't special technology or a clever idea. You'll know the idea in full just from reading this document.
Anyone can make one video with AI.
Dozens of people, at the same time, in the dead of night when no one's watching, posting without a single mishap —
that's built by operating history, not code.
Going Global — How We Expand
- We finish preparing topic seeds and lining up testers for 16 language zones this year. The Topic Sourcing Engine is already running; we're now planting seeds in each language zone.
- We launch in at least 16 languages simultaneously in Q1 2027. The first doors we open are the English-speaking world and Japan.
- We're going overseas not for the market but to divide the topic shortage by the number of languages.
- We're expanding not by setting up branch offices but by cloning what we have already run in Korea, country by country. The per-country cap is 1,000 people.
The Order of Bottlenecks
At first we thought "we don't have enough equipment." Counting told a different story.
Instagram's per-account daily posting cap (50 posts) isn't the bottleneck. We run about 1 post per account per day — far below it.
The real danger isn't the cap — it's the mass suspension that hits when multiple accounts look connected to each other. So the key to scaling isn't headcount, it's independence between accounts.
16 languages — dividing up topics, not accounts
This is the most counterintuitive part.
Say we grow to 1,000 marketers. If several people post the same content, it becomes spam, so we'd need about 334 distinct pieces of content a day. What we make in a day today doesn't come close. 1,000 people on Korean alone is impossible.
But when the language differs, viewers don't overlap. The same content can be reused in each language zone.
The English-language pipeline is already built and validated, just waiting to launch. The other languages ride the same process with just the language swapped out, so each added language costs far less than the first one did.
And the Topic Sourcing Engine is already running. It's a process that automatically pulls videos from seed accounts, branches them, and filters them — running unattended, on schedule, every day. So far it has stockpiled 3,189 pieces with 3,951 queued. What's left is expanding those seeds to 16 languages — planting, in each country, the topics people there actually read. 14 seed accounts recently cleared the gate (6+ pieces sampled, 25%+ pass rate) for the first time, bringing the total to 67. Finishing this work within the year is the precondition for launch — growing accounts first just stacks up empty ones.
How we go global — cloning, not setting up branches
Going overseas usually means incorporating a local entity and sending people. We don't do that. We clone the system we've run end-to-end in Korea, country by country. The factory remains centralized; what grows is local people's laptops and accounts held in their own names.
The real reason platforms mass-suspend accounts isn't exceeding a limit — it's accounts looking connected to each other. When accounts share the same line, the same device, or the same identity, the platform reads them as one, and if one gets flagged, the whole cluster dies together.
Running overseas accounts through a proxy from Korea only amplifies this signal. When a local person runs their own account on a local line, that signal never appears in the first place. Recruiting overseas marketers isn't a cost-saving choice — it's the only approach that works within this structure.
The first doors open together
A complete English-language pipeline (voice, images, subtitles) is already built and validated, waiting for channels to be finalized. Reviving it and plugging it into the node network comes first.
This is the one we open through local connections, not technology. We run the tech pilot (English-speaking) and the business pilot (Japan) together, so each validates the other.
The criterion for choosing a language zone isn't market size — it's whether we can start right now. The rest open in sequence, after we've fixed the problems these two surface.
Hardware — and the error we caught ourselves
To raise throughput, we need to move voice synthesis to GPU. We ran the sizing math twice, and the first pass was badly wrong.
3 GPUs · roughly ₩7.08 million · 2,355 pieces/day on measured throughput · supports up to 2,000 accounts (recommended 1,000–1,500).
However, voice synthesis is still tied to CPU, so even if we buy the hardware, we need to fix the code first before we can use the GPUs. So the code fix comes before the purchase order. Had we reversed that order, a ₩7.08 million machine would have sat idle.
Overseas organizational structure
Overseas, we're designing this as business-to-business contracts with national distributors. We won't build a chain where individuals recruit individuals and earn commission on their sales.
The reason is legal — a chain where individuals recruit individuals and earn commission on their sales can be classified as multi-level sales under Korea's Door-to-Door Sales Act, and failing to meet the registration requirements (₩500 million in capital, membership in a mutual-aid association) carries criminal liability. So we proceed with this structure only after legal review.
China bans team-compensation structures outright, with criminal penalties. We can't add the Chinese-speaking region into our 16 language zones the same way. Counting languages and opening a country are different problems, and the viable launch structure differs by country. We're noting this in advance.
A single worldwide rate of $50/month is tied to this same structure. If prices differed by country, distributor contracts would also differ by country, and managing that across 16 language zones simply isn't realistic.
Where the money moves
The moment we go overseas, both inbound payments and payouts cross borders. A single domestic payment method isn't enough.
- Receiving — we use an overseas subscription billing provider that handles VAT and invoicing for each country on our behalf. The more countries, the safer it is to outsource tax handling.
- Paying out — we need to send monthly commissions to hundreds of people. Sending each one manually through a bank counter isn't feasible, so we use a mass payout service.
- Tax — non-resident withholding and treaty-limited tax rates come into play. The call depends on where the service was rendered, so tax advice comes first.
Compensation terms and contract structure are under legal review, and we won't disclose figures before that review is done — publishing unconfirmed compensation terms would itself create risk. For the same reason, we don't name the entities behind payment, remittance, or tax handling.
Timeline
By year end
What Could Kill Us
- The biggest risk isn't technology — it's law and platform policy.
- Every risk already has a defense we've built — not a promise, but code and rules.
- What we watch most closely is marketers growing faster than ads.
The defenses above don't eliminate every risk. In particular, platform policy is beyond our control. If the rules change one day, an entire channel could be shut down.
That's why we've spread across five channels. A structure where four remain when one closes is safer than optimizing for a single channel. It's also why earlier we listed every inactive channel in full — those are both a list of failures and a diversification list.
Team and Company
- Reborn Labs was founded in Mar 2026, and is running six businesses at once just six months after founding.
- The CEO is also the CTO and writes the code himself.
- This business didn't come out of a plan — it came from hitting a wall while doing it ourselves.
Company
| Name | Reborn Labs (REBORN LABS) |
|---|---|
| Registered representative | Shim Jae Yoon |
| CEO & CTO | Kim Seong-jun — heads product and technology. Designed and built the system described in this document himself |
| Business Reg. No. | 817-05-03415 · general taxpayer |
| Founded | Mar 25, 2026 |
| Address | 306 Banyawol-ro, Dong-gu, Daegu, Republic of Korea (Sinseo-dong) |
| Business areas | Content automation · store operations solutions · mobility · investment information · AI work automation · overseas distribution |
Why we're the ones qualified to solve this
Instead of listing credentials, here's the logic.
What We're Becoming — A Global Ad Execution Company
- The final destination isn't running channels for others — it's a company that directly executes ads for global AI companies.
- The path has three steps — 1,000 in Korea → 16 language zones → executing ads.
- That's why marketers are not customers but crew, and their combined followers are our asset.
This isn't a story about growing one account.
It's a story about building the scale that emerges when marketers' followers worldwide are combined.
Why an ad execution company
Right now, the companies increasing their ad spend the fastest worldwide are AI companies. What they need isn't banner space — it's channels that show real people using the product every day, in multiple languages. But an ad agency can't build that channel. Agencies buy media; they don't grow it.
We started from the other end. We grew the media first — 98 marketers, 280 accounts, running every day. Add languages to that, and it becomes an asset existing agencies can't buy and advertisers can't build themselves. That's when we shift from a company that receives ads and distributes them, to one that executes ads.
Agencies buy media.
We grow media.
Three steps
Seen only as a service fee, $50/month doesn't make sense. That money is closer to a share in joining this business as crew. Crew hand over their own-name accounts and support HQ's operations, and in return HQ grows those accounts diligently and turns that into ad revenue paid back to them.
That's why we set the split at 80% (Part 6). If crew don't earn, the channel network shrinks, and if the channel network shrinks, step 3 never arrives. Aligning everyone's interests in the same direction is the whole point of this structure.
One person's account means nothing to an advertiser. Only when thousands of accounts move together across 16 languages does it become something worth selling. That's why we use the word crew — it only has value once combined.
5 years from now
Every brand has its own broadcast station.
And that station stays on every day, without anyone tending it.
Today, only companies with the capacity to produce content have a channel. A neighborhood repair shop, a small workshop, a sole proprietorship can't. Not because there's no one to make it, but because there's no one to make it every single day.
Once the marketer network is big enough, this relationship flips. Brands stop building a channel and start using the channel network. Marketers stop selling time and instead hold a channel asset in their own name. We run the process that connects the two.
And this structure isn't tied to Korean. Just run a different language through the same process — 16 languages is that door.
A request for each reader
Terms Used in This Document
| Term | Meaning |
|---|---|
| Marketer | A participant who installs the program on their own laptop and runs social media accounts under their own name |
| Marketer Network | All marketers combined with the HQ system that directs them |
| Episode | The original version of one piece of content. One episode is finished differently by different marketers |
| Topic | The subject or information that content is made from. "What to talk about" |
| Topic Sourcing Engine | An engine that scans seed accounts and automatically pulls in raw topic material. Runs unattended through collecting, receiving, branching, and filtering |
| Seed account | The source account topics are pulled from. Promoted to a seed after multiple pieces have been sampled and the pass rate clears the bar |
| Scene sheet (blueprint) | A written breakdown of the scene order for presenting a topic. Nothing can be made without one |
| Baking | Attaching a script and voice to produce the content prototype |
| Render | Turning the scene sheet and audio into an actual video file. Done by the marketer's laptop |
| Assignment | Deciding which marketer posts what, on a given day |
| Gate (pass check) | An automatic check run at each step. If it fails, it does not move on to the next step |
| Ledger | The record of who posted what, when, and how it performed. There is exactly one |
| Review sheet | A scene-by-scene layout of a finished video, made for manual visual review |
| Overnight production window | 00:00–06:50. Anything not finished within this window doesn't go out that day |
| P0 | Top-priority incident. The grade at which everything else stops to fix it first |
Marketer FAQ
How much does it cost?
Just $50/month (about ₩67,000). No installation fee, no sign-up fee, and you don't need to buy any equipment. Whatever laptop you already have works. The price is the same no matter which country you join from.
Do I have to hand over my account password to the company?
No. Account information never leaves the marketer's laptop. HQ only sends "what to post." This isn't a matter of trust — it's a technical architecture issue. If HQ connected to accounts directly, every account would look like it's logging in from one place, triggering mass suspension (Diagram 5).
Does my laptop need to stay on 24 hours a day? What about electricity costs?
Yes, staying on at all times is a requirement. The actual work mostly happens overnight, so during the day it mostly sits idle. One laptop's power draw isn't much, but we want to be upfront that it isn't ₩0.
How much can I earn?
We won't state a guaranteed amount. There are four revenue sources — 80% commission on small-product sales, 80% split on ad orders, 20% on full-stack, and a referral fee for each new marketer, paid from that marketer's subscription fee (₩5,000 a month, or ₩10,000 once). The non-ad revenue sources depend on your own activity level, while ads depend on how much volume HQ brings in.
At this stage, the honest answer is this — the first ad order has only just come in (Part 5). Until that becomes routine, sales commission is the main source of income. But the fact that 80% goes to marketers once ads are live is already locked in now — not a promise to raise it later, but the rate as of today.
Payouts are made once a month, once the cumulative balance reaches ₩50,000 or more, with 3.3% withholding tax applied. Details are in the separate Compensation Policy Statement.
The 80% stated here is the term for $50/month subscription marketers. Tester marketers who joined before the subscription system are under a separate arrangement, and the terms they were originally given remain in place.
What happens if my account gets suspended?
We've separated voice, topic, visuals, and connection lines to cut the suspension risk, but the risk isn't 0. Support for reopening an account is capped at twice a year. We state the cap because that's better than promising unlimited support and failing to deliver.
What happens to my account if I quit?
The account was always in your own name, so it stays as it is. Only the program stops. The posts and followers you've built up remain yours. Since there's no installation fee, there's no sunk cost when you leave.
What if I want to post something myself?
You can. The system only posts assigned content, and it doesn't stop you from posting your own separately.
Isn't it a problem that this is AI-made content?
What platforms regulate isn't "using AI" itself — it's churning out inaccurate or identical content at scale. That's why fact-checking and review are built into our 12 steps, and each marketer uses a different voice, topic, and length. That said, as noted in the risk chapter, platform policy is an area we can't control.
Can I do this from outside Korea?
It opens in at least 16 languages starting Q1 2027. The price is the same $50 worldwide, and you'll post local-language content to local accounts.
How We Counted the Numbers
Every figure in this document comes from the operating database and direct observation, not from PR-style aggregation. We're publishing the method so it can be checked.
What we measured and how
- Source — the operating records database (62.8MB · last updated 2026-09-24 22:18 · 36 tables)
- Method — direct database queries. We used no pre-built aggregates or estimates
- Measurement date — 2026-09-24
Where each figure comes from
| Figure | Source |
|---|---|
| 98 marketers (87 active · 2 paused · 6 ended) | Full node table · testers excluded · active = laptop that signaled within the last 24 hours |
| 280 accounts | Full channel table · broken down by platform (Instagram 97 · Naver 92 · YouTube 91) |
| 6,663 publish jobs | Publish-type events in the event table · Sep 3–Sep 23 (Korea time). This is a count of jobs that were started, failed, or auto-DM'd together — not a count of confirmed posts |
| 1,064 performance-tracked records · 94 marketers | Full performance table |
| 813 original content pieces | Unique episode values in the performance table |
| 0 scene sheets across 146 of 220 topics | Full topic-inventory audit, 2026-09-07 |
| 3,189 pieces of raw topic material · 15.7GB · 3,951 queued | Topic Sourcing Engine inventory query, 2026-09-24 (topic-videos --remaining) |
| 275 hand-picked tech-category pieces | Category breakdown from the same query |
| Sourcing ceiling of roughly 3,000 pieces/day | A 5.5-hour, unthrottled, 6-second-interval sourcing run (1,409 pieces) scaled to 8 hours |
| 67 seed accounts (14 newly promoted) | Seed promotion run, 2026-09-18 — 6+ pieces sampled, 25%+ pass rate |
| 74% of accounts have 0–2 followers | Account survey, 2026-09-10 |
| Headcount needed by video length | Assignment-history analysis (27–29s needs 7–19 people / 40–43s needs 51–52 people) |
| 785 pieces/day · 2.39x optimism | Throughput recalculation, 2026-09-18 |
| 4.4% payment processing fee | Actual transaction, 2026-08-12 — a ₩159,000 charge netted ₩152,004 |
| Per-channel activity status | Ledger query, 2026-09-24 — Instagram post records · YouTube post URLs · Naver publish records |
| Break-even at 271 / 171 marketers | The ₩29,700/month-era calculation (612 / 140 people) converted to the $50/month basis. The ad-side figure is recalculated with an 80% revenue share (110 people if it were 50%). Fixed costs held constant |
| 21 pieces/day across 16 languages | 1,000 people ÷ up to 3 people per piece of content = 334 pieces → reused across 16 language zones |
What we couldn't measure
Being accurate means also listing what we couldn't measure.
- Google Blogger account counts — these exist in the code, but we couldn't find a path where they're tallied in the records
- Exact time spent per piece of content — the records have no start/end timestamp fields
- Cumulative publish total across all channels — some tables only retain recent entries, so they can't serve as a cumulative figure. The body only uses values with a stated period (6,663 over 21 days)
- Cumulative revenue and net profit — this is why we didn't include financial projections in this document. We'll update once we have grounded results to report
- Actual operating cost of the $50/month system — the break-even headcount above is a converted figure. We'll recalculate from measured data after the global launch
The more numbers a document uses, the more how they were counted matters. Publishing the method means we can be told we're wrong when we are, and confirmed when we're right. A number that can't be checked is the same as no number at all.
The figures in this document are as of Sep 24, 2026, and may change depending on operations.
We do not guarantee profits. Please read Appendix B and Part 9 together for participation terms and risks.