Blind Vibe Coding Vs Old Software Engineering

Blind Vibe Coding

In the first approach, you go directly to the AI Coding Agents and tell them what you want to build, and leave everything for them, but here are the issues with them:

• You don't own what you build.

  When you are just giving prompts to Cursor or Claude Code, it is writing the code for you and building the app. You don't know your app workflows, the user stories, or the business logics integrated into it — You don't even know which frameworks and technologies it is using. (some even don't know the programming language it is using :p)

  When you don't know about the technology you are using in your app, you don't have the idea that technology may have been deprecated and replaced by something much better 10 years ago, and the Vibe Coding Assistant may be using it because it is trained on a bulk of past data and code. As you don't know, you can't move your application from that old framework to the newer one, which could have made your app much faster and better in terms of user experience.

• You don't know what frontend and backend architecture it is using.

  Creating an application that runs on localhost is something else; building something so scalable that thousands and millions of people should be able to use it is a totally different world. So developing something that can't scale is nothing but a kid's play. And scalability depends upon our frontend and backend architectures. When we vibe code, we don't even know which programming language it is using, what about the architectures? Who will know whether it has to be a SPA or MPA, a Native app or a Cross-platform one? Whether we have to use a Monolithic architecture or a microservices-based architecture.

• Whether it has even created a separate frontend or backend or not.

  Mostly vibe coding tools just create a frontend and write all the logic onto the frontend, which is very insecure and unscalable.

• You don't know where it stores the data.

  Data is your app's biggest asset. You don't own the data, you don't own anything. Mostly, vibe coding tools just store the data in the local storage on the browser (which is very silly). But what the vibe coding tool doesn't have are any options because the vibe coder doesn't even know what a database is, so they don't provide it to the AI Assistant. Vibe Coders don't even know that there are different types of databases available to store and query data in different ways. They never heard of SQL, NoSQL, Vector, or Graph databases.

Old Software Engineering

If you go with the second approach — the old Software Engineering, it will have only one issue, but a bigger one:

• It takes a lot of time.

  You would be learning everything slowly and gradually, and writing code by hand. People doing vibe coding would be doing things 100x, and until you learn that technology completely, it may get outdated.

The Solution: Vibe Engineering

And the solution is: "Vibe Engineering". Let me define it:

"Write code using AI, but still having the know-how of the major code components, tech stacks, architectures, and data sources."