Pre-launch is a marketing word. Pre-pilot is an honest one. Most teams say the first when they mean the second, and most readers can’t tell the difference. We can. So we say so.
Two weeks ago a journalist asked me, on background, whether Chingu was “already in the field.” She had read the company page, seen backend, dashboard, family app, senior app deployed; 181 tests passing, and assumed that meant a halmoni in Daegu was, at that moment, talking to the product. She wasn’t. Not yet. Chingu has been built. Chingu has been tested. Chingu has not been used in production by a halmoni who didn’t already know us. That is what pre-pilot means.
I corrected her in the call. I asked her why she had assumed otherwise. Her answer was that the page sounded launched. That was useful information.
The two words, plainly.
Pre-launch means the product is finished and the marketing is held back. The plumbing works, the users are paying, the law is satisfied, and the company is sitting on the announcement waiting for a date. The risk has been written down to a press cycle.
Pre-pilot means a real user, in their real environment, has not yet held the product for long enough to tell you whether it actually works. The plumbing is finished, the marketing is honest, and the law is in flight. The risk is the user. The risk is whether the thing you have built behaves the same way at four in the morning, in a Daegu kitchen, in 사투리, on a phone the senior already owns, as it does on a Macbook in Seoul.
Pre-launch is a fundraising posture. Pre-pilot is a research posture. They look the same on a slide; they are not the same thing.
Where the two companies actually are.
Imali. The WhatsApp loop is live at imali.ebstar.co. 1,208 tests pass on every push, ~85% coverage. The free tier is open, the paid tiers (R29 / R99) are wired up. There are merchants typing transactions into WhatsApp; there are not yet merchants paying us R29 for the privilege. We have onboarded through community partners in Gauteng and the Western Cape. The credit-profile generator runs against staging data and produces something a lender could read; it has not yet produced something a lender has, in fact, read. Pre-pilot. Voice playbook says so plainly.
Chingu. Backend, welfare dashboard, family app, senior app: all four deployed. 181 tests passing. PIPA compliance verified through the test suite (consent, encryption, audit log, deletion). The Korean wake-word essay (/journal) was written from a real Daegu session. We have not yet enrolled a single halmoni outside that session. The first paid municipal invoice is targeted for Q1 2027 after a free five-month MOU pilot in Q3 to Q4 2026. Today is May. Pre-pilot, by every working definition.
“Built. Tested. Deployed.” is not the same as “used.” The honest version says all four words.
Why the discipline matters.
The category we sit closest to (African informal-merchant fintech) has a recent body count. Khatabook reached 500,000 merchants on a free bookkeeping app, raised $187M, never monetised, and shut down its Pay and Start lines in October 2023. OkCredit raised $84M, was put up for sale. BukuKas took $140M and was liquidated. Kippa took $14.3M, hit 500,000 users, then made the app inaccessible in January 2024. Users lost their business records overnight. Each of those companies described itself as launched, in the press, the year before it folded.
I don’t cite this list to be morbid. I cite it because it is the precise list a serious LP runs through their head before taking the meeting. If we describe Imali as “launched” today, we walk into that meeting carrying the rhetorical weight of every one of those companies. If we describe it as “pre-pilot, with the WhatsApp loop live and 1,208 tests on every push,” we are doing two things at once: setting expectations honestly, and signalling that we read the same body count they did.
What “pre-pilot” sounds like, in product copy.
Three patterns we hold ourselves to:
- The not-yet block before the proof block. Every company page on this site has a section that lists what we don’t have, by name, before it lists what we do. Counterintuitively, that increases trust. Investors, journalists, regulators: all three tell us so.
- Numbers we measured, never numbers we modelled, on the public surface. “1,208 tests passing” is measurable. “5,000 merchants by month 18” is a milestone. We don’t mix the two on a stat row.
- Composite, when composite, labelled inline. If a journal piece is braided from three pilot interviews, the kicker says Composite scene and the disclosure sits at the top of the article, not the bottom. The receipt piece does this; this one does not need to, because the only person quoted in this one is me.
When does pre-pilot end?
For Imali, pre-pilot ends when the first hundred merchants have crossed sixty days of continuous logging without us nudging them. That is the point at which we can credibly say someone is using the product, not just trying it. We are not there yet. We are some weeks out.
For Chingu, pre-pilot ends when one of the three municipalities currently in conversation signs the free five-month MOU and a halmoni completes her first week. Halmonis don’t enrol on websites. They get enrolled by 보건복지과 caseworkers who trust the city; the city trusts the procurement; the procurement trusts the regulator; the regulator trusts the certification. Each of those trust chains takes its own weeks.
We will say so on the day each chain closes. Until then, the page that describes us has to read the way the work feels, not the way the deck feels.
Honest about stage is the cheapest moat we have. We don’t share it with our competitors because it costs nothing. Only because they don’t want it.
A short test, before any sentence about stage.
Borrowed from the voice playbook, slightly tightened for this question:
- If a journalist asked me to substantiate this sentence right now, could I?
- Does the sentence describe what I have measured, or what I am modelling?
- Have I named the population, the number, and the constraint?
- Is there a “not yet” sentence on the same page, in the same voice?
- Would the sentence still be true if a regulator read it tomorrow morning?
If yes to all five, ship it. If no to any, rewrite it. That includes our own pages.
ETB · written from Seoul, May 2026 · with both companies still pre-pilot, on purpose