AI could grow the economy while more of the gains go to owners of businesses and other assets. That finding in Anthropic's economic scenarios made me ask: what could I build and own?
In the substantial scenario, there are fewer jobs in knowledge work, including programming and analysis. Pay stays roughly flat, although the economy grows at about twice its normal rate.
In the extreme scenario, these jobs fall much further. Pay for knowledge workers drops by more than 10% by 2030 compared with an economy without AI.
Capital owners get a larger share of income in both scenarios. In the extreme case, workers' total income barely changes by 2030, so most of the extra income goes to capital.
These are possible futures, not predictions. They depend on AI's abilities and how widely it is used.
This makes ownership more interesting to me. The same AI that reduces the cost of intellectual work could also help me build a product or business of my own.
Cheaper code makes customers and trust more important
I think a booking tool for repair shops is a useful example. Suppose AI cuts the time needed for the first version from months to weeks.
The developer still needs to understand how shops handle bookings, find an owner willing to try the tool, move their records, and support them when something breaks.
Another developer could use AI to build similar features. They would still need to earn customers' trust.
As software gets cheaper to build, I expect more value to come from:
- Customer access: knowing how to reach shop owners who need the tool.
- Trust: a record of keeping bookings safe and fixing problems.
- Useful data: information that takes time to collect and keep current.
- Market knowledge: understanding details such as repair delays and parts availability.
That is the part of ownership I find interesting: a share of the value that the product and those relationships create over time.
An asset can serve more customers without repeating all the work
By an asset, I mean something I could own that keeps creating economic value after the first version is built. Examples include:
- Subscription software: a booking or reporting tool with a monthly fee.
- A mobile app: a tool with paid features or a subscription.
- An API: a service that converts files or cleans data, charging by usage.
- A browser extension: a tool that saves time on a repeated task.
- A library or plugin: software that teams pay to use under a license.
- A database: information customers pay to access, with regular updates.
- A course, book, or guide: material that can sell many times.
- A newsletter or media business: content funded by readers, sponsors, or product sales.
- A share in a business: ownership of part of a company's future value.
The same library can serve hundreds of teams. The same booking tool can serve many shops. Neither needs to be rebuilt for each customer.
A brand and an audience can help too. Customers who trust one product may try another from the same owner.
These assets need updates, support, and money to run. Their value can grow faster than the owner's working hours, but that is not guaranteed.
AI can help a small team test a business
AI can help write code and tests, draft documentation, and prepare support replies. A developer still needs to check the results, but may need less help to launch a first version.
That creates room to test a business before hiring a full team.
For example, one developer could build the booking tool while a partner finds repair shops willing to test it. They could start with bookings and reminders, then add features based on what paying customers need.
A business run by two people serving one market is a model I find interesting. I would care more about whether customers stay and income covers the costs than about the size of the team.
A side project needs demand to become a business asset
It would be easy to call ten side projects ten assets. But ten unused apps could just mean ten things to maintain.
They can still be useful for learning. For a business, I would look for other signs:
- Users return.
- Customers pay.
- The business can afford the cost of finding new customers.
- Income covers hosting, AI usage, support, and other expenses.
- Each new customer does not require hours of manual work.
A paid extension with returning customers shows demand. A large codebase alone does not.
That would change what I count as progress. A conversation about why someone stopped using the product could matter more than another feature.
I see another path alongside engineering
The usual engineering path is:
Junior → Mid-level → Senior → Staff → Principal
The path I am thinking about looks like this:
Engineer → Builder using AI → Software business owner → Owner of several assets
It brings up questions that writing code alone cannot answer:
- Product: what problem would people pay to solve?
- Distribution: where would the first hundred customers come from?
- Sales: how would I explain the value and ask for payment?
- Business economics: what would it cost to find and serve a customer, and what would be left?
I see this as another direction my engineering skills could take me. Even one product would give me plenty to learn about running a business.
I would want to test the business idea early
If I tried this, I would start with a problem in a market I understand. I would want to know how people solve it today and whether the problem costs them enough to pay for a different solution.
For the booking tool, my first question would be whether shop owners would move their bookings or pay for reminders. A small test could answer more than months spent building features.
This is where AI seems most useful to me: reducing the time and money needed to find out whether an idea works.
If one product found customers, it might also reveal another problem in the same market. That is how I imagine several assets could grow from one business, rather than from a list of unrelated app ideas.
The question I am left with
Anthropic's scenarios made me think about who benefits when intellectual work gets cheaper. More of the gains could go to owners, while AI also makes it easier for a developer to become one.
I do not know which scenario will happen. But it has changed the question I ask about my skills.
Alongside “what can I build?”, there is now another one:
What would I want to own?
Top comments (0)