For those who may not know, in simple terms: SaaS is a program that runs on the Internet, not installed on your computer. In fact, this is any online service that you use through a browser or an application.
Now you can hear about vibecoding from every iron. And to be honest, there's a lot of HYPE around him, but overall the idea is correct.
Today, it's really possible to sit down with minimal programming knowledge and start doing a fairly serious project. A social network, CRM, marketplace, SaaS service - all this no longer looks like something unattainable for an ordinary person.
But there is one mistake that I sometimes notice.
People think that because AI has appeared, now there is no need to study anything at all.
It is necessary.
Not to write code by hand for 8 hours a day. And in order to understand what's going on in your project.
Let's say you decide to create your own social network, CRM or any other online service for your tasks.
What, in my opinion, needs to be understood at least at a basic level:
- What is SaaS and how do web applications work?
- Work with code editors (VS Code and others) and integration tools with AI models.
- Which modern and popular technology stack is closer to you (for example, Laravel + PHP, React or Vue, MySQL/PostgreSQL) and what task each component solves.
- What is the difference between a frontend and a backend and how they interact.
- How servers, domains, hosting, and APIs work.
- Basics of working with databases: MySQL, PostgreSQL, SQLite, tables, relationships, queries.
- Basics of working with Git: commits, branches, pushes, rollbacks of changes.
- Working with the command line (CLI).
- How to run a project locally and deploy it to a server.
- How to read logs and look for error causes.
- Basic understanding of the project architecture.
And one of the fundamental points in this whole story is understanding the subject area and the problem that your product solves.
Who needs it, what kind of pain it covers, why people should choose it, and what they will be willing to pay for.
Because this is exactly what the architecture of the project and its logic are based on. And it will be almost impossible to roll it back or redo it later.
In fact, it's not as much as it seems. Especially when you consider that previously you had to understand everything at the application level - write code, tests and check it all yourself.
My thought is simple: vibecoding does not cancel the foundation. It removes a huge amount of routine.
Previously, you had to lay every brick yourself. AI is laying the bricks now.
But if you don't understand where the foundation is, where the load-bearing wall is, and why the roof suddenly fell, you won't get far.
