In our case research on 14 August we wrote this line about an AI business:
⚠️ Listed for sale at $17,000 — concrete evidence of the bias in a seller-populated list
We opened it again today.
What happened in two days
| 14 Aug 2026 (our notes) | 16 Aug 2026 (today) | |
|---|---|---|
| Asking price | $17,000 | $12,000 |
Down $5,000. A 29% cut, in two days.
That is today's screen. The $17,000 from two days ago is our own research record — we did not capture the screen at the time, and we say so rather than implying we did.
The numbers behind it
Displayed by the platform, read from a connected Stripe account.
| MRR | $311 |
| Active subscriptions | 9 |
| Launched | September 2025 → month 11 |
| All-time revenue | $3,085 (about $280 a month) |
| Asking multiple | 2.6× — computed on last 30 days × 12 |
Eleven months, $3,085 in total. An average of $280 a month.
Next to the case we published today
Monthly recurring revenue. A 260× gap.
Both were verified identically — a third party reading the payment processor directly. Verification does not confer success. It only establishes that a number is real. Whether the number is large is a separate question.
When you next read about someone earning six figures with AI, the thing to hold onto is that the same verified table also contains $311 a month — and that one rarely becomes an article.
⚠️ Something else the verified dashboard showed
The page disagrees with itself. Its title reads "$390 last 30 days" while the chart in the body reads $381.
That is the second time we have seen this. The grade-A case we published today showed the same split — $102,660 in the title, $96,516 in the body. Once is chance; twice is a pattern. The title figure appears to come from a different moment than the chart. We could not establish why, so we use neither title figure — only the clearly labelled values.
The "profit margin" field also still reads XX% — an unfilled placeholder. Our
14 August note recorded 40% there. Verified dashboards have blanks too.
Why this is not listed as a case
Our grading rules require a named founder for grade A. This business is operated under a handle only, so identity is unconfirmed. Listed as a case it would be grade B — and grade B cases carry no figures.
Which means that under our own rules we may print $80,601 and may not print $311. There is an asymmetry inside our rules where successes get recorded and failures drop out.
So this is an observation, not a case entry — a record of what a dashboard displayed on a given date. That distinction could look convenient, so we are stating it here. It is the first time our rules have cut against us, and hiding that would be worse than the inconsistency.
⚠️ Limits
✓ Read it this way
- An asking price is what a seller wants, not what anything sold for
- The two-day cut is a fact; the reason for it is not something we know
- The sample is one listing
✕ Reading it this way is wrong
- A price cut means failure — there are many possible reasons
- AI businesses are all like this — one listing generalises to nothing
- $311 is the final outcome — it is still operating
⚠️ We have not named the business or its operator. It is a live sale, and what there is to learn is the shape of the numbers, not the person. Our internal research notes retain the identifying details so the record stays checkable. The platform's marketplace is public, so you can look at comparable listings yourself.
Next
We will look at the same listing in a month and record whether it sold, dropped further, or disappeared. One reading is a photograph, not a trend.