Cases

81 transactions prove demand — but this one is not a solo side business

A listing that sells AI model design and implementation as project work. It has the strongest demand evidence in this list (81 transactions, 100% satisfaction) and the lowest replicability. It is a two-person company with a floor price near ₩800,000. Read it as proof that money moves in this market, not as a side-hustle template.

Founder
StellaDev (Stella Dev Co., Ltd.)
Country
KR
Team
2
Startup cost
To first revenue
Automation
low
Reported revenue
Revenue not disclosed

What is sold

Project work — a floor price plus surcharges by data type

From ₩800,000 · 14-day turnaround · 2 revisions. Data types beyond text and images carry a surcharge. Checked 2026-08-15.

Who buys

Companies and teams that need an AI capability built; client names appear in the portfolio

AI's part and the person's part

Already had

MSc from a university AI lab, former researcher at KAIST, former R&D at a large Korean tech group's AI division. The profile states two programmers, both with master's degrees.

If this is large, the AI is not the reason

What the AI did

Selects a model for the client's problem, implements it from the literature, builds training data where needed, and runs training and validation. The portfolio includes a deep-learning demo mobile app and a motion-recognition model.

the part a tool replaced

What the person did

Requirement analysis and model selection are effectively the whole job. What is sold here is not tool operation but knowing which approach to use.

stories that omit this do not replicate

The three columns are not proportional to contribution — we did not measure that. This is a schematic for separating them.

How customers were found

  1. Kmong — marketplace search placement

  2. Portfolio with client names

  3. Platform-assigned senior seller level

  4. Paying customers

    From ₩800,000 · 14-day turnaround · 2 revisions. Data types beyond text and images carry a surcharge. Checked 2026-08-15.

Channels we could confirm from public sources. Conversion rates were not disclosed, so they are not shown.

Flow: Kmong — marketplace search placement, then Portfolio with client names, then Platform-assigned senior seller level, then Paying customers

How we verified this

Evidence strengthB· 1 source(s)

Public dashboard or platform figure. May be self-reported

Checked the Kmong listing directly on 2026-08-15. Platform-displayed values — 81 completed transactions, 100% satisfaction, 31 reviews averaging 4.9, corporate account. The profile states two programmers, so this is not a one-person business. No revenue figure is published, so we print none. ⚠️ The 81 transactions are the seller's total, not this listing's.

See the grading rules

What made it work

The use of this case is the price floor, not the playbook. Eighty-one completed transactions at a floor of ₩800,000 with a 14-day turnaround says demand exists at that band. Put next to the two cheaper cases in this list, the shape of the Korean AI-services price range starts to show.

Risks

⚠️ Lowest replicability in this list. The research and industry track record is the product. ⚠️ Two people, so it is not evidence that one person can do this. ⚠️ A 14-day engagement is hard to run alongside a full-time job — unlike the 3-to-7-day chatbot case.

Does this work in Korea?

A domestic company selling on a domestic platform, so it transfers directly. Corporate buyers effectively require tax invoices, which in turn requires business registration.

Replicability in Korea1/5
Difficulty5/5

Solo-feasible: no · Automation: low

7-day action plan

  1. Step 1. Do not copy this one. Instead list only the services in the same category with a turnaround of three days or less

  2. Step 2. Write down that list's price distribution and find the band you could enter

  3. Step 3. In one sentence, write what separates the ₩800,000 listing from the ₩150,000 one

30-day action plan

  1. Step 1. Buy or at least request a quote from three of the cheapest listings in your band

    see what is actually delivered

  2. Step 2. Pick the one point where you could do better

  3. Step 3. List a service covering only that point. Do not widen the scope

Run the numbers

We do not print a single projected figure here. Put your own numbers in instead.

Break-even calculator →