Verified cases

Getting paid for making someone else's AI easier to work in

Open the real page ↗typingmind.com · alive as of 2026-08-31

This product does not replace ChatGPT. It sells a better place to do work with it — and it never pays for the model, because the customer plugs in their own API key. That single decision keeps the operator's AI cost near zero. ⚠️ The founder has published monthly revenue figures. We do not repeat them: no source independent of the founder exists (grade B).

In plain words

ChatGPT already exists. This founder sells a workspace built on top of it for people doing daily work. The customer pays the model provider directly, so the operator carries almost no AI cost.

The whole thing at a glance

  1. To whom

    People and teams using AI in…

  2. What

    AI SaaS

  3. Found how

    X — an audience the founder…

  4. Paid how

    Subscription and licence; te…

Flow: To whom, then What, then Found how, then Paid how

Founder
Tony Dinh
Country
VN
Team
Startup cost
To first revenue
Automation
high
Reported revenue
Revenue not disclosed

What is sold

Subscription and licence; team and enterprise controls sit in the higher tiers

Pricing is published on the official site (liveness confirmed 2026-08-31). ⚠️ We do not copy the figures across. This site publishes only values it has checked itself, and price checks run on a separate monthly routine in the tool directory.

Who buys

People and teams using AI in daily work, plus companies that need an internal deployment

What arrives is not what you keep

  1. What the customer paidThe number on the invoice
  2. AI toolsA share of the monthly subscription
  3. Platform feeDiffers by platform
  4. TaxDiffers by country and business form
  5. What is leftAt or below zero, it is not a price
The rows are not scaled to amounts. Tools, fees, and tax differ per person, so we do not draw a share we did not measure — put your own numbers in the calculator.

This case does not publish its own tool costs, fees or tax. So we show the order and the items only — put your own numbers into the calculator.

AI's part and the person's part

Already had

A developer. This is not a non-technical person's first build. Read the starting line as engineering skill plus an audience that was already there.

If this is large, the AI is not the reason

What the AI did

No model is trained here. It is a client for other people's models. Calls leave on the user's own API key, and what is actually sold is the working experience: organised threads, saved prompts, team sharing, switching models.

the part a tool replaced

What the person did

Product judgement, engineering, and the trust built by working in public. The founder is a developer with an existing audience, and that audience overlaps with the first customers. ⚠️ This is the bottleneck for anyone copying it — the audience existed before the product.

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. X — an audience the founder built by working in public

  2. Word of mouth from individual users into their teams

  3. Team and admin features pulling in whole organisations

  4. Paying customers

    Pricing is published on the official sit…

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

Flow: X — an audience the founder built by working in public, then Word of mouth from individual users into their teams, then Team and admin features pulling in whole organisations, then Paying customers

How we verified this

Evidence strengthB· 2 source(s)

Public dashboard or platform figure. May be self-reported

Confirmed — the service is live (official site HTTP 200, 2026-08-31), the founder is named, and the founder's newsletter post exists (HTTP 200, 2026-08-31). Not confirmed — revenue. The monthly figures and the claim that over half is recurring B2B come from the founder's own newsletter. There is no payment-processor verification and no independent reporting, so this is not grade A and we publish no amount (CLAUDE.md §2.1).

See the grading rules

What made it work

There is a paid position directly above a general-purpose model: the working experience. You do not need to build a model, and every improvement to the underlying model makes your position larger, not smaller. What converted one-off interest into recurring revenue was the admin layer. The moment an individual brings the tool into a team, accounts, sharing and access control become someone's problem — and that is where subscriptions begin.

Risks

  • If the model providers absorb these features, the reason to pay shrinks. The official clients keep adding team functionality.
  • Asking users to supply an API key is a real barrier — for a non-technical buyer the first screen is the hardest one.
  • Revenue rests on the founder's word alone. Do not treat the amounts as established.

Does this work in Korea?

The transferable part is the cost structure. When the model runs on the customer's key, the operator never carries the inference bill — a structural answer to costs that climb as usage grows, and it holds in Korea unchanged. The hard part is the audience. Early demand here came from years of public work. Ship the same product with no audience and you get a good tool nobody has heard of.

Replicability in Korea3/5
Difficulty4/5

Solo-feasible: partially · Automation: high

7-day action plan

  1. Step 1. Write down three things you redo by hand every time you use an AI interface

  2. Step 2. Check in a community whether one of those three is other people's problem too, not just yours

  3. Step 3. Work out your fixed monthly cost if the model ran on the customer's key instead of yours

30-day action plan

  1. Step 1. Build the one fix that annoyed you most and put it in front of ten people

  2. Step 2. Before charging, ask users directly what they would need to bring it into a team

  3. Step 3. Decide which feature separates individual from team use, and price only after that

Run the numbers

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

Break-even calculator →

Get the next verification result

We check the AI income stories people are talking about, grade the evidence, and send you what we found — including what we could not confirm.

Words you may not know

Verification grade
How far we checked this case. A means the revenue was confirmed by someone other than the founder; B means the business exists but the revenue is only the founder's word.
Revenue
Everything that came in. Tools, fees and tax still have to come out before it is what you keep. The two are not the same.
MRR
Money that arrives every month, as opposed to a one-time payment.
Self-reported
A number the person stated about themselves. Nobody else confirmed it, which is why we do not print those numbers.
Replicability
How likely someone else is to get a similar result. The more the founder already knew the field, the harder it is to copy.