This month I watched a four-minute Douyin video by a creator called 澳洲 AI 教父. He answers a question I hear from students all the time: "I learned so much AI. So where is the money?" His answer is that the money is not in which tool you use. It is in what you sell. He splits the AI business into seven layers, from selling your own time to building the AI model itself. The map is his. Below I walk through it, then add the part where I see it differently.
这个月我看了澳洲 AI 教父的一支抖音，四分钟。他回答了一个我的学生常常问的问题：「学了一堆 AI，钱在哪里？」他的答案是：钱不在你用什么工具，在你卖什么。他把 AI 生意分成七层，从卖你自己的时间，一直到做 AI 模型本身。这个框架是他的。下面我先把它讲清楚，再讲我看法不一样的那一半。A new AI tool makes you faster at the layer you are already on. It does not move you up. To move up, you change what the customer pays for: your time, your content, a result, a product, a system. My own lens: a product that only wraps AI is the half that gets cheaper fastest. What holds its value is the fit, built for one industry.
新的 AI 工具，只会让你在现在这一层做得更快，不会让你往上走。要往上走，就要换客户付钱买的东西：你的时间、你的内容、一个结果、一个产品、一套系统。我的看法：只是把 AI 包一层的产品，是最快变便宜的那一半。真正保值的，是为一个行业做的「合身」。
Whose idea this is
The seven layers are 澳洲 AI 教父's framework, not mine. I only add my own lens on top.
I am writing about it because it is one of the clearest maps I have seen of a simple truth: people don't pay you for knowing AI. They pay for what you hand them.
The seven layers
Read it from the bottom up. Each layer changes the thing the customer pays for.
-
7
Build the base model
Sells: capital and research Runs on money and resources, not tool skills. Only a handful of teams in the world can do it.
-
6
Sell AI capability
Sells: an API For example an image API. Many buyers are designers and marketers, not programmers, so pure tech drifts away from the market unless it is wrapped into a tool they can use.
-
5
Sell an AI system
Sells: a department rebuilt around AI He calls this FDE. It needs deep understanding of the business and of AI together. Far harder than making content.
-
4
Sell an AI product
Sells: software AI wrapped into an easy tool, on subscription or licence, like a one-click image maker. The tech bar is not high. Product and market sense are.
-
3
Sell AI results
Sells: a result More customers or lower costs for a traditional business. The owner only looks at the outcome. His example: work that needed 10 people now needs 2.
-
2
Sell AI content
Sells: content Videos, articles, templates made in bulk, paid by ads or courses. Easy to start, so it gets crowded. The content itself is cheap. Traffic and trust are what's worth something.
-
1
Use AI to do the work
Sells: your time A designer, salesperson or owner uses AI to work faster. Income is still tied to hours. You can take more jobs, but nothing works while you rest.
The line that matters most
The whole video turns on one idea.
A tool skill only makes your current layer more efficient. To change layers, you have to change what you sell.
澳洲 AI 教父 · paraphrased from the video
His example: going from doing design to selling a design tool. Same skill underneath. Completely different thing on the invoice.
This is why learning one more tool so often doesn't change your income. If you are on layer 1, a better tool makes you a faster layer 1. You still sell hours. The tool was never the step. The step is the thing the customer pays for.
Where I see it differently
I agree with his map. I would cut two places differently.
First, the word FDE. He uses FDE for layer 5, rebuilding a whole department around AI. On this blog I use FDE for Forward Deployed Engineer: the person who goes inside a client's business and makes AI actually run there. Different words, same work. Layer 5 is where an FDE lives.
Second, layer 4 in 2026. A product looks like the dream layer for one person, because it keeps selling while you sleep. But earlier this month I wrote that every piece of software has two halves: the feature, and the fit. The feature is how it works. The fit is that it runs every day inside one business, with that business's data and someone looking after it.
Sells the feature
- One clever thing AI can do
- Same tool for everyone
- Easier to copy every month
The half getting cheaper.
Sells the fit
- Built on that industry's real data
- Shaped to how their day actually runs
- Someone keeps it running
The half that holds its price.
A product that only wraps AI is mostly the feature. That is exactly the half AI is making cheapest to copy. So if you build at layer 4, build where the fit is: one industry, its real data, its real daily work.
Don't ask which AI tool to learn next. Ask what your customer will pay for next.
Weiss Ang
Find your layer this week
You don't need a new tool for this. You need ten minutes and an honest list.
- Write down the last three times someone paid you. What exactly did they pay for?
- Label each one: time, content, result, product or system. Be honest. Most of us start at time.
- Ask one question: if I stop working tomorrow, does this income stop? If yes, you are selling time, whatever the tool.
- Pick one thing you could sell one layer up. Not a new tool. A new thing on the invoice. Instead of charging per post, charge for the outcome the posts are meant to bring, and agree with the client how you will measure it.
- Test it with one real customer before you build anything. One person saying yes tells you more than a month of planning.
If you want the bigger picture of working this way, start with what a One-Person Company actually is. The whole system I use is free in my community.
I'm Weiss Ang. One person, plus an AI department, is a company.
I ran mine first, from Sabah, Malaysia. Now I help people in their own industry build a line that keeps running by itself. I have trained more than 1,300 people to use AI, and I run the One-Person Company community, where the whole system is free.
我是洪家徽。一个人，加 AI 部门，就是一家公司。我先在沙巴把自己的跑通了，现在帮你那一行，搭一条会自己跑的线。我教过一千三百多人用 AI，也在经营 One-Person Company 社群，整套系统免费。
FAQ
What are the seven layers of the AI business?
Will learning more AI tools make me more money?
What is the best layer for a solopreneur?
What does FDE mean here?
Who created the seven-layer framework?
Sources
- 澳洲 AI 教父, 学了一堆 AI，钱在哪里？你对 AI 的生意了解多少？ Douyin, 9 October 2026, about 4 minutes. v.douyin.com/ieG0Nw1
- Weiss Ang, Software Can Be Read Now. The Fit Cannot. 13 October 2026. solopreneur-os.pages.dev/blog/reverse-engineer-anything
Build your One-Person Company with me.
The whole system and the course are free inside my community. Put your name on the waiting list.
Join the waiting list