Everyone knows Software as a Service—building software and selling it—has been a very promising and attractive dream, especially after Claude Code, ChatGPT, and other AI coding tools made product creation easier.
The dream of building a SaaS, selling it for millions, retiring early, achieving financial freedom—who wouldn't be excited!
However, I don't believe that's still true, unless you're creating something truly groundbreaking, genuinely different in technology or valuable data.
The Problem with Traditional SaaS
Take accounting software as an example.
Previously, only a few large companies could afford to build accounting software. Users paid for licenses to keep books, issue invoices, generate P&L reports, and financial statements.
The cost to build such a system was so high that only those with resources could participate, and once they built it, everyone had to use it.
The problem is, over many years, SaaS typically falls into two scenarios:
-
First, users pay too much compared to actual value received. For example, you buy accounting software, use it for 10 years rotating around just a few basic features, but still have to pay maintenance fees, or lose access.
-
Second, the software becomes too...messy—too many features. Tech companies naturally like to upgrade and add features, but over time the interface becomes increasingly convoluted, cluttered with buttons, making new users spend a long time learning. To the point where at many companies, training interns or mastering the software becomes a separate skill, consuming considerable time.
Building Costs Have Changed
Today everything has changed, because the cost to create an application has decreased dramatically.
Previously, like building a house, you needed a large team and several years. A hundred years ago, the same type of house would take double the time and workforce compared to now, because construction technology wasn't developed.
Software is the same—previously it required an entire company, a team of engineers to build, but now just two people, even one person with AI is enough. AI isn't something mysterious, it's just an acceleration tool, like having an iPhone to take instant photos instead of previously using film cameras and going into a darkroom to develop.
The Future: Custom-Designed Software
Because application building costs are so low, what makes more sense is that each person, each business should have their own software, exactly matching their real needs.
If you only need one set of features, just build that set, no need for excess. The interface is also designed according to your business process, rather than being forced into a generic mold.
For example, some businesses need two sets of books simultaneously; previously popular accounting software didn't allow this, only serving reporting purposes without supporting management, leading to inability to clearly see profit and loss according to management needs.
From now on, I believe it will be the era of custom-designed applications. Building costs are low, so software value no longer lies in "it exists," but in the creator's capability: knowing how to listen to customers, deeply understand problems, and solve in the most suitable way for the owner—the business itself.
Previously, excessive costs forced people to make mass-market products, like H&M or Zara clothing, producing large quantities for everyone. But currently, software can completely be like tailored clothing: compact, easy to use, powerful, and truly fitting the user.