Vibe Coding vs. Agentic Engineering: Where is the line?

I was reading about how 'vibe coding' and 'agentic engineering' are starting to feel like the same thing, and it made me think about how we're using AI in development. It seems like the lines between these two concepts are blurring, which is kind of confusing.

What exactly are these two concepts?

Basically, 'vibe coding' is when someone who isn't a deep programmer asks an AI for something without really focusing on the code quality or the underlying structure. It's more about getting a result quickly, maybe just satisfying a feeling. On the other hand, 'agentic engineering' is what professional software engineers do—they use AI tools but focus heavily on security, maintainability, and making sure the code is solid and reliable. It’s about using tools with a high level of professional understanding.

Why does this distinction matter?

The main takeaway is that while AI makes it easier to generate code, the actual difficulty of producing good software remains high. The real focus shouldn't be on just generating code, but on how professional companies use AI assistance to build better, more secure products. I felt a bit guilty about not always reviewing AI-generated code, which makes me realize that accountability is super important, especially when we're deploying things in the real world.

What am I still unsure about?

I'm still not totally sure where the line should be drawn between casual prompting and professional engineering. I wonder how we can make sure that the tools we use are used responsibly, and I definitely need to keep thinking about how to separate the 'vibe' from the actual engineering process.