Title: Why Software Engineering matters in Web Apps and Why Vibe Coding Worries me
published: true
tags: webdev, software engineering, ai, beginners
Introduction
AI can now build a web app from one prompt. That's exciting, but I think we're skipping an important question: what happens after the app "works"?
1. What is vibe coding?
- You describe what you want, and an AI writes the code.
- You accept it without really reading or understanding it.
- If the page loads and the buttons click, you call it final.
- It's fine for a weekend experiment.
- It's risky once real users enter real data.
2. Why software engineering is relevant in webapps.
Writing code is only one part of building a web app. Engineering also covers.
- Planning what the app should do and who it's for
- Designing how the frontend, backend and database fit together
- Using Git so mistakes can be undone
- Testing features so they keep working
- Reviewing code before it goes live
- Protecting user data and logins
- Deploying safely and noticing when something breaks
A web app isn't finished when it runs on your laptop. It has to survive bad input, attackers, traffic and time.
3. The dangers of vibe coding
-Security holes:
API keys in front-end code, forms that trust any input, weak logins, open databases. If you can't read the code, you can't spot these.
-Code nobody understands:
When it breaks, you don't know where to start. Each AI fix can break something else.
-No tests:
A small change can quietly break the booking form, and you find out when a customer complains.
-Technical debt:
Messy, repeated code makes every new feature harder to add.
-Poor scaling:
An app that works for 5 users can crash at 5,000.
-User data risk:
Weak storage or no backups can mean leaks or lost data.
-Hurts learners:
If AI writes everything, you never build debugging and design skills.
It's not only an AI problem
- Apps built with no planning, Git, tests or documentation were always fragile.
- They work on launch day, then become hard to fix and hard to hand over.
- Vibe coding just gets you to that points faster.**
4. How I think is the best way of how we should use AI
- Read and understand generated code before using it
- Keep secrets in environment variables, never in front-end code
- Test the important features, like forms and payments
- Commit small changes with Git so I can go back
- Plan features and data before I start prompting
- Treat AI as an assistant, not the engineer
5. Conclusion
- Vibe coding is good for quick ideas and prototypes.
- It can't replace software engineering.
- Apps people depend on must be secure, tested and maintainable.
- That comes from knowledge and discipline, not vibes.
- AI makes good engineers faster, but it can also help careless builders ship bigger mistakes faster.
Top comments (0)